Upload
anthony-thompson
View
98
Download
9
Tags:
Embed Size (px)
Citation preview
Mental Causation and Ontology
This page intentionally left blank
Mental Causationand Ontology
edited byS. C. Gibb, E. J. Lowe,and R. D. Ingthorsson
1
3Great Clarendon Street, Oxford, ox2 6dp,United Kingdom
Oxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide. Oxford is a registered trade mark ofOxford University Press in the UK and in certain other countries
# the several contributors 2013
The moral rights of the authors have been asserted
First Edition published in 2013
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored ina retrieval system, or transmitted, in any form or by any means, without theprior permission in writing of Oxford University Press, or as expressly permittedby law, by licence or under terms agreed with the appropriate reprographicsrights organization. Enquiries concerning reproduction outside the scope of theabove should be sent to the Rights Department, Oxford University Press, at theaddress above
You must not circulate this work in any other formand you must impose this same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
ISBN 978–0–19–960377–0
Printed by the MPG Printgroup, UK
Links to third party websites are provided by Oxford in good faith andfor information only. Oxford disclaims any responsibility for the materialscontained in any third party website referenced in this work.
Contents
List of Contributors vii
Acknowledgements viii
Introduction 1
S. C. Gibb
Part 1: Levels of Being, Properties, and Mental Causation
1. Mental Causation 18
John Heil
2. Physical Realization without Preemption 35
Sydney Shoemaker
3. Mental Causation in the Physical World 58
Peter Menzies
4. Mental Causation: Ontology and Patterns of Variation 88
Paul Noordhof
5. Causation is Macroscopic but Not Irreducible 126
David Papineau
Part 2: Causal Relata, Substances, and Powers
6. Substance Causation, Powers, and Human Agency 153
E. J. Lowe
7. Agent Causation in a Neo-Aristotelian Metaphysics 173
Jonathan D. Jacobs and Timothy O’Connor
8. Mental Causation and Double Prevention 193
S. C. Gibb
9. The Identity Theory as a Solution to the Exclusion Problem 215
David Robb
10. Continuant Causation, Fundamentality, and Freedom 233
Peter Simons
11. There is No Exclusion Problem 248
Steinvor Tholl Arnadottir and Tim Crane
Index 267
vi CONTENTS
List of Contributors
Steinvor Tholl Arnadottir, University of Stirling
Tim Crane, University of Cambridge
S. C. Gibb University of Durham
John Heil, Washington University, St Louis
Jonathan D. Jacobs, St Louis University
E. J. Lowe, University of Durham
Peter Menzies, Macquarie University
Paul Noordhof, University of York
Timothy O’Connor, Indiana University
David Papineau, Kings College London
David Robb, Davidson College
Sydney Shoemaker, Cornell University
Peter Simons, Trinity College Dublin
Acknowledgements
We are very grateful indeed to the Arts and Humanities Research Council
(AHRC) for funding the research project on the new ontology of the
mental causation debate, of which this book is a consequence. (Arts and
Humanities Research Council’s Research Grant AH/F009615/1 ‘The New
Ontology of the Mental Causation Debate’.) We would also like to thank
the AHRC, the Mind Association and the Analysis Trust for helping to
fund the conference held at Durham University on the ontology of mental
causation, at which many of the contributors to this volume presented
their work.
Introduction
S. C. GIBB
The hypothesis of this book is that many of the central issues in the mental
causation debate are metaphysical, where metaphysics is to be understood as
the branch of philosophy that is concernedwith the fundamental structure of
reality as a whole. More specifically, we propose that it is ontology—the
domain of metaphysics which concerns what categories of being there are
and how they are related to one another—that is of particular concern to
the mental causation debate. It seems clear that, insofar as all branches of
philosophy deal with aspects of reality, they all incorporate—whether expli-
citly or implicitly—ontological claims. This is no less true of the philosophy
of mind and, in particular, the debate about mental causation. If this is the
case, then solutions to the problem of mental causation that attempt to
divorce themselves from ontology or which are based on ad hoc ontological
assumptions will inevitably prove to be inadequate. Let us begin by con-
sidering what the problem of mental causation is a problem about.
The thought that we are capable of performing intentional actions which
result in the movement of our limbs is central to our pretheoretical concep-
tion of human agency. It is my belief that it is going to rain together with my
desire not to get wet that causes me to run indoors. It is my desire to quench
my thirst in combinationwith my belief that there is a Coke in the fridge that
causes me to open the fridge door. However, despite the initial plausibility of
the claim that the mental is of causal relevance in the physical domain,
specifying a relation between the mental and the physical that is both
consistent with psychophysical causation and independently plausible has
proven to be one of the most enduring problems in the philosophy of mind.
Causal closure arguments appeal to the existence of psychophysical
causation as a premise, along with a further premise that the physical domain
is in some sense causally closed. All such arguments share the conclusion that
mental causes (that have physical effects) are identical with physical causes.
The structure of a causal closure argument will differ according to the
strength of causal closure principle that it appeals to. The most popular
form of causal closure argument, owing largely to the relative weakness of
the causal closure principle that it requires, is the argument from causal
overdetermination. It can be set out as follows:
(1) Some mental events are causally relevant in the physical domain.
(Call this ‘the principle of psychophysical causation’.)
(2) At every time at which a physical event has a cause it has a sufficient
physical cause. (Call this ‘the principle of causal closure’.)1
(3) There is no systematic causal overdetermination. (Call this ‘the
principle of causal non-overdetermination’.)
Mental events (that are causally relevant in the physical domain) are
identical with physical events.2
To explain the argument: In accordance with the principle of psycho-
physical causation, say thatM1 is a mental event existing at time t and thatM1
is a cause of physical event P2. The principle of causal closure states that at
every time at which a physical event has a cause it has a sufficient physical
cause, where for event e1 to be causally sufficient for event e2 is for e1 to
causally necessitate e2. Given the principle of causal closure, it follows that as
P2 has a cause at t1, it must have a sufficient physical cause at t1. Call this
sufficient physical cause P1. Now the principle of causal closure does not
entail that M1must be a physical event, despite the fact thatM1 has a physical
1 Note that if (2) were to be substituted for the weaker claim that ‘Every physical event that has a cause
has a sufficient physical cause’, then, provided that causation is transitive, the combination of the
premises of the argument from causal overdetermination would not be enough to establish its conclu-
sion. This is because, given the transitivity of causation, a physical event would have a sufficient physical
cause if it had a sufficient mental cause which in turn had a sufficient physical cause. If a principle of
causal closure allows mental events this causal role in the physical domain, then the three premises of the
argument from causal overdetermination are quite clearly compatible with various forms of dualism.
(2) avoids this problem because, according to it, at every time at which a physical event has a cause it has a
sufficient physical cause. However, it is arguable that an appeal to the possibility of simultaneous
causation demonstrates that even (2) is to too weak to establish the argument from causal overdetermi-
nation’s conclusion. For a detailed discussion of all of these points, see Lowe (2000a).2 For various formulations of this three-premise argument, see Hopkins (1978), Crane (1995), Lowe
(2000b), and Papineau (2001), among others. The formulation that I have presented here is Lowe’s (see
Lowe 2000b, 27).
2 S . C. GIBB
effect. Even if it is the case that at every time at which a physical event has a
cause it has a sufficient physical cause, some physical events might have non-
physical causes as well. It is only the principle of causal closure in combin-
ation with the principle of causal non-overdetermination that entails that
non-physical events are causally irrelevant in the physical domain. Given the
latter principle then arguably, as a general rule, if an event e has a sufficient
cause c at t, no event at t distinct from c can be a cause of e. It follows that if P1
is a sufficient cause of P2 at t1 andM1 is also a cause of P2 at t1, thenM1must
be identical with P1 or part of a cause that is identical with P1.
Both the principle of causal closure and the principle of causal non-
overdetermination enjoy general support in the mental causation debate.
The principle of causal closure is widely accepted, largely as a result of the
empirical support that it is thought to enjoy—that is, the great number of
cases in current physics that confirm this principle and the lack of any
disconfirming cases. Current physics, it is argued, has found sufficient
physical causes for many different kinds of physical events. Furthermore,
physics has never needed to appeal to a non-physical cause to provide a
sufficient cause for a physical event. This, it is thought, provides inductive
evidence for the claim that for every physical event, at every time at which it
has a cause, it has a sufficient physical cause.
Regarding the principle of causal non-overdetermination, depending on
one’s theory of causation, one may of course want to allow that there are
special cases in which events are causally overdetermined. For example, if
two shots are independently fired and each bullet reaches the victim’s heart
at exactly the same time, provided that each bullet striking was causally
sufficient for the victim’s death, the victim’s death was arguably causally
overdetermined by the shootings. However, even if there are isolated cases
of causal overdetermination, the claim that as a general rule some events are
causally overdetermined seems implausible. It is this claim that the principle
of causal non-overdetermination rules out—according to it, events are not
systematically causally overdetermined. And it is precisely this kind of sys-
tematic causal overdetermination that the combination of the principle of
psychophysical causation and the principle of causal closure seems to give
rise to, unless, of course, mental events and physical events are identified.
Despite the apparent plausibility of the argument from causal overdeter-
mination, the acceptance of its conclusion is problematical. Largely as a
result of the argument from multiple realizability—the argument that
INTRODUCTION 3
mental properties are multiply realized by, and hence cannot be identical
with, physical properties—most philosophers in the mental causation debate
would want to accept the following principle:
The principle of property dualism: Mental properties are not identical with physical
properties.
The acceptance of the principle of property dualism is, given certain accounts
of an event, directly inconsistent with the conclusion of the argument from
causal overdetermination. If, for example, events are property-instantiations
and the identity of property-instantiations requires the identity of their
properties, then the principle of property dualism entails that mental events
cannot, contrary to the argument from causal overdetermination, be physical
events. Thus, for example, take Kim’s account of an event.3 According to it,
the event (s1, p1, t1) exists just in case the substance s1 instantiates the property
p1 at time t1. A mental event is the instantiation of a mental property by a
substance at a time. A physical event is the instantiation of a physical property
by a substance at a time. Events (s1, p1, t1) and (s2, p2, t2) are identical if and
only if s1= s2, p1= p2, and t1= t2. Hence, the identity ofmental and physical
events requires the identity of mental and physical properties. Thus the
acceptance of the principle of property dualism entails the rejection of the
conclusion of the argument from causal overdetermination.
One might instead attempt to advance an account of events which allows
one to combine an event monism with a property dualism—events involve
properties, but the identity of events does not require the identity of these
properties. Given such an account, the principle of property dualism does
not entail the rejection of the conclusion of the argument from causal
overdetermination. Mental events may be physical, even though mental
properties are not. But even if it is combined with an event monism,
maintaining a property dualism still presents a problem given the argument
from causal overdetermination, as questions about the causal redundancy of
the mental simply arise at the level of properties, as opposed to the level of
causes. Events are presumably causes in virtue of the properties that they
involve—it is, in other words, the properties of an event that make the
causal difference. Furthermore, events are rarely causes in virtue of all of
their properties. Consequently, the question of whether a mental event is
3 See, for example, Kim (1993a).
4 S . C. GIBB
ever causally relevant within the physical domain in virtue of its mental (and
hence, non-physical) properties arises. If one denies mental properties such
causal efficacy, then one abandons any serious commitment to the principle
of psychophysical causation. Alternatively, if one allows them such causal
efficacy, then one seems forced to abandon either the principle of causal
closure or the principle of causal non-overdetermination. For these reasons,
it seems that one must either abandon the principle of property dualism or
provide a reason to reject the argument from causal overdetermination.
With the exception of the psychophysical reductionist, most of those in the
mental causation debate take the second option, and reject or disambiguate
one of the premises of the argument from causal overdetermination. Hence,
for example, eliminativists reject the existence of mental entities, and hence,
mental causes. They therefore reject the principle of psychophysical causation.
Interactive substance dualism and anti-physicalist forms of property dualism
typically deny the principle of causal closure. Most forms of non-reductive
physicalism deny or disambiguate the principle of causal non-overdetermin-
ation. Thus some argue that the combination of the principle of psychophys-
ical causation, the principle of causal closure, and the principle of property
dualism does not, given non-reductive physicalism, result in a systematic causal
overdetermination of the worrying type. For it to do so, it would have to be
the case that the effect would have occurred if only one of the causes was
present. On the non-reductive physicalist’s account, mental and physical
causes are not metaphysically independent—mental properties ontologically
depend on physical properties—and hence this condition cannot be satisfied.
But all of this is only to paint a partial picture of the possible responses to
the problem. An alternative kind of response is to argue that the acceptance
of the three premises of the argument from causal overdetermination does
not entail that mental causes are physical causes, and hence does not conflict
with the principle of property dualism. It is only given the implicit accept-
ance of further claims—claims that are themselves open to question—that
this would be so. Hence, for example, dualist attempts to respond to the
problem of mental causation by denying the homogeneity of the causal
relata are of this type, as are attempts to respond to the problem by denying
the homogeneity of the causal relation.4
4 See Lowe (1993), (2000a), and (2008) for the first kind of position. See Crane (1995), }9 for a
discussion of the second kind of approach.
INTRODUCTION 5
Despite the great amount of effort that has been devoted to responding to
the problem of mental causation, a number of central figures in the mental
causation debate have questioned whether we are any closer to a satisfactory
solution or, indeed, given the current form of the debate, whether we will
ever be. Hence, Kim concludes that those in the mental causation debate are
‘up against a dead end’ (Kim 1993b, 367), while McGinn suggests that a
solution to the problem of mental causation is forever beyond our reach
(McGinn 1989).
It is our hypothesis that the apparent insolubility of the problem of mental
causation stems, in part, from the fact that the debate has not always been
framed with sufficient ontological precision. Basing it in a plausible onto-
logical framework will constitute, we contend, a significant step towards its
resolution. Not only will this add clarity to the existing debate and serve to
identify those positions that are ontologically unsound, but it will also
potentially reveal new ways of responding to the problem of mental caus-
ation that have not been previously explored.
The metaphysical issues that are of relevance to the mental causation
debate fall into three interrelated groups. The first concerns the nature of
the causal relata. The second concerns the nature of a property. The third
concerns the nature of the causal relation. Let us consider each of these
issues in turn.
Assuming that causation is a relation, what is it a relation between? What
one takes the causal relata to be depends on the ontological system in which
one is basing one’s account of causation, and hence the ontological categories
to which one appeals. Making the plausible assumption that singular caus-
ation is a relation between particulars, there are several possible candidates for
the role. These include events, states of affairs, individual substances, and
tropes. In formulating the argument from causal overdetermination, we have
assumed that causation is ‘event causation’—that is, that causation is a
relation between events. Furthermore, in explicating what it is to maintain
an event dualism we have assumed that events involve properties. These
assumptions are open to question, and whether or not they are correct
impacts on the mental causation debate. Davidson’s theory of anomalous
monism and the discussion that surrounds it serves to provide an excellent
demonstration of this point (Davidson 1980b, 1980c, 1980d, 1993).
According to Davidson’s account of mental causation, mental events reduce
to physical events, but mental concepts do not reduce to physical concepts.
6 S . C. GIBB
Like the argument from causal overdetermination, Davidson’s argument for
the conclusion that at least some mental events are identical with physical
events has three premises and appeals to a premise that is similar to the first
premise of the argument from causal overdetermination:
(1) The Principle of Causal Interaction (CI): Some mental events and
physical events are related as cause and effect.
(2) The Principle of the Nomological Character of Causality (NCC):
Events related as cause and effect fall under strict laws.
(3) The Principle of the Anomalism of the Mental (AM): There are no
strict psychophysical laws.
To explain this argument, in accordance with CI, say that mental event, M,
and physical event, P, are related as cause and effect. Given NCC, it follows
that M and P must be characterizable in terms that allow them to fall under a
strict law. Given AM, if M is only characterizable in mental terms, M and
P cannot fall under a strict law. Consequently, M must also be characteriz-
able in physical terms. For this to be the case, M must be a physical event.
Furthermore, as there are no strict psychophysical laws that would support
the reduction of mental concepts to physical concepts, anomalous monism
leads to the rejection of any conceptual reduction.
Now anomalous monism is often accused of resulting in a property
epiphenomenalism.5 Davidson appears to be combining an event monism
with a property dualism. The worry, therefore, is how, according to
anomalous monism, the mental properties of an event can be causally
efficacious in the physical domain. More specifically, how can anomalous
properties of an event be causal properties given NCC?
If causal relations obtain between events because events involve certain
properties, this problem is arguably insoluble. However, unlike his critics,
Davidson does not consider events to involve properties. For Davidson,
properties are not objective aspects of things in the world. What makes an
event mental (or physical) is whether or not it has a mental (or physical)
description. Although a mental (or physical) description of an event is either
true or false of it, there is no ontological fact about or feature of an event that
makes its description as mental (or physical) true or false.6 Given this
5 See, for example, Honderich 1982, Kim 1993c, McLaughlin 1993, and Sosa 1993.6 See, for example, Davidson 1980b, 215.
INTRODUCTION 7
account of an event, it makes no sense to suggest that events are causes in
virtue of their properties. Therefore, anomalous monism cannot be accused
of property epiphenomenalism. It is arguably true that if one were to
combine, for example, a Kimean theory of events with anomalous monism,
then the resulting position would result in a property epiphenomenalism.
But Davidson’s account of mental causation should not be detached from
his account of an event. Of course, whether the ontological system in which
Davidson bases anomalous monism is plausible is a further issue. My point
here is simply that, given this ontological system, anomalous monism cannot
be accused of property epiphenomenalism.7
Even if one accepts, contra Davidson, that a cause is a cause in virtue of
the properties that it involves, there are still many further questions to be
raised regarding the nature of the causal relata. The first set of questions
concerns the nature of each of the categories which one’s account of the
causal relata invokes. Take, for example, the popular claim that events are
the causal relata, where events are property-instantiations—that is the
instantiation of a property by a substance at a time. Several questions arise
regarding this account: Are the properties that substances instantiate uni-
versals or particulars? What is the ontological status of substance—are
substances reducible to ‘bundles’ of properties or is this ontological category
fundamental? If the category of substance is fundamental, then what is
its nature? Is a substance to be understood as a ‘substratum’ or ‘bare
particular’—that is, as an entity which in itself has no properties, but
which plays the role of supporting and uniting an object’s properties? This
would suggest an ontology in which states of affairs are the basic building
blocks of reality. This is because, as Armstrong argues, given this notion of
substance, something is needed to weld substance and property together,
and it is states of affairs that most plausibly play this role. Where a substance a
instantiates property F, a’s being F is a state of affairs, and a and F are
constituents of the state of affairs which it holds together in a non-mereo-
logical form of composition. If so, then property-instantiations are states of
affairs.8 Alternatively, substances might be Aristotelian in nature.9
According to this account, a substance is not something that is wholly
separate from its properties in the way that a bare particular is (and for this
7 See further Crane (1995) and Gibb (2006). 8 See Armstrong (1997).9 See, for example, Lowe (2006) for such an account.
8 S . C. GIBB
reason, states of affairs are arguably not required to weld substance and
property together). But nor should it therefore be concluded that a sub-
stance just is a state of affairs, for unlike states of affairs, substances are not
complex entities. Rather, according to this account, substances comprise a
basic, irreducible category, which do not have properties as constituents,
but upon which properties ontologically depend. Property-instantiations
are therefore not to be identified with states of affairs, but with states of
substances.
A further set of questions concern the homogeneity of the causal relata. Is
all causation a relation between entities of the same kind? According to
several philosophers, it is not. Some advance an account of events and of
facts that allows them to distinguish between event causation and fact
causation. That is, causation is neither exclusively a relation between events
nor exclusively a relation between facts, but instead event causation and fact
causation are both basic and irreducible—either an event or a fact can cause
either an event or a fact.10 Alternatively, rather than drawing a distinction
between event causation and fact causation, others propose a distinction
between event (or fact) causation and agent causation. In the case of agent
causation, an agent causes an event (or fact).11 An ‘agent’ is to be understood
as an individual substance—whether the term is to be restricted to those
individual substances that are capable of performing intentional actions is a
matter for discussion.
These issues regarding the nature of the causal relata all have a bearing on
the mental causation debate. Indeed, the importance of some of them has
already been made clear by those in the debate. Robb’s ‘trope solution’
provides one such example (Robb 1997). Robb argues that if a cause is the
instantiation of a trope by a substance, as opposed to the instantiation of a
universal by a substance, then this resolves the problem of mental causation.
According to Robb, if properties are tropes then one can accept the
conclusion of the argument from causal overdetermination whilst avoiding
the problem of multiple realization—the denial of the principle of property
dualism is only problematical if properties are universals.
Lowe’s dualist response to the problem of mental causation provides
another example of the importance of the issue of the causal relata to the
10 See, for example, Bennett (1988).11 See, for example, O’Connor (2000) and Taylor (1966).
INTRODUCTION 9
mental causation debate.12 Lowe accepts the premises of the argument from
causal overdetermination, whilst rejecting its physicalist conclusion. He
does this by defending a psychophysical causal interactionism that rests on
a distinction between event causation and fact causation. According to
Lowe, mental events do not cause physical events. Rather mental events
cause physical facts. This, Lowe argues, is wholly consistent with each of the
premises of the argument from causal overdetermination.
Let us now turn to the second issue—that of property analysis. For those
in the mental causation debate who admit the ontological category of
property, the issue of how to analyse properties is a central one. What is it
for a property to exist? What is it for one property to be identical with
another? These questions matter to the mental causation debate at the most
general of levels. To determine whether mental properties exist, and, hence
whether mental causes exist, one must adopt some criterion of property
existence. To determine whether a mental property is identical with a
physical property, one must adopt some criterion of property identity.
A further closely related issue concerns the different kinds of relationship
that properties can stand in—both to each other and to items of any other
ontological category that one would want to admit. Of particular import-
ance to the mental causation debate, and more specifically to non-reductive
physicalism, is the question of whether properties can be said to depend
upon each other and, if so, in what sense.
The fundamental importance of property analysis to the mental causation
debate is made clear by Heil’s arguments regarding the mental causation
debate and levels of being.13 As Heil observes, a familiar point—but one that
is not always sufficiently respected in the mental causation debate—is the
importance of distinguishing property-types from predicates (see, for
example, Heil 2003, Ch. 3). A plausible analysis of the existence and identity
conditions for properties reveals that not every meaningful predicate picks
out a real property-type. This is the case regardless of whether properties are
universals or particulars and, hence, regardless of whether a property-type is
a universal or, for example, a set of resembling tropes. Whilst few in the
mental causation debate would admit to conflating predicates with proper-
ties, Heil considers that some of the central arguments and assumptions in
this debate are guilty of just such a conflation. In particular, Heil considers
12 See, for example, Lowe (1993). 13 See, for example, Heil (2003, 2004).
10 S . C. GIBB
non-reductive physicalism to be guilty of confusing the idea that mental and
physical predicates operate at different levels of description, with the idea
that mental and physical properties exist at different ontological levels.
Now, whether this criticism of non-reductive physicalism is correct is, of
course, a matter for debate. What seems to be clear, however, is that one
cannot take part in this debate without engaging in the analysis of a
property’s existence and identity conditions.
Finally, we would suggest that the mental causation debate is affected by
what the causal relation is. Some theories of causation say that c causes e
when e is counterfactually dependent on c (that is, if c had not existed, then
e would not have existed). Others maintain that for c and e to be causally
related they must exemplify types that are lawfully connected, but disagree
about what this lawful connection consists in. Still yet others maintain that
causation is the transference of some quantity from cause to effect, for
example, the transference of energy or momentum. Alternatively, others
consider that causation is to be accounted for in terms of irreducible causal
powers. Properties confer causal powers on the substances that they char-
acterize. Causation is the exercise of these causal powers. These are but a
few of the many different theories of causation that have been presented in
the literature on causation.
All theories of causation fall into one of two groups—those that consider
causal laws to be more basic than causal relations and those that consider
causal relations to be more basic than causal laws. In line with this, one may
distinguish between generalist and singularist accounts of causation.
A singularist denies that singular causation is grounded by type-level rela-
tions, while generalists maintain that for particular events to be causally
related at the token level, they must instantiate types of events that bear
suitable objective relations to one another. Most plausibly, laws furnish
causation’s generalist component. For example, the regularity theory of
causation is a generalist theory of causation. According to it causation is
grounded by laws, where laws are regularities. By contrast, the energy
transference theory of causation is a singularist account of causation—
according to it causation is not grounded by type-level relations.
There is also a narrower understanding of what a generalist theory of
causation is. According to this understanding, whether or not two particular
events are causally connected depends upon things that happen elsewhere or
elsewhen—causation is not a purely local matter. Contrary to this, others
INTRODUCTION 11
consider causation to be a purely local matter. The energy transference
theory of causation, for example, adopts the latter position—whether or not
two events are causally connected simply depends upon whether there is
transference of a quantity between them. Whether there is such a transfer
does not depend upon what is happening elsewhere or elsewhen. By
contrast, the regularity theory of causation adopts the former position.
According to it, a particular sequence is causal by virtue of being an instance
of a general pattern and the same particular sequence coupled with no
general pattern would not be causal. It is, however, quite clearly the case
that not all nomological theories of causation are generalist in this narrow
sense.
Here we have mentioned but a few of the more popular theories of the
causal relation. However, what should be obvious from this brief and partial
summary is just how much theories of causation can differ. It seems
reasonable to raise the question of whether any of these differences matter
to the mental causation debate.
Some of the main issues that might be raised regarding the mental
causation debate and theories of causation can be recognized by focusing
on the principle of causal closure. Very few theories of the causal relation
entail that causation must be a physical relation—one exception is an energy
transference theory of causation that identifies ‘energy’ with ‘physical
energy’. But, from the fact that few theories of the causal relation entail
that causation is physical causation, it would be far too hasty to conclude
that one’s theory of causation therefore does not affect the plausibility of the
principle of causal closure.
We have formulated the principle of causal closure as the claim that at
every time at which a physical event has a cause it has a sufficient physical
cause. What it is to be a ‘cause’ and what it is to be an ‘effect’, and thus what
it is to be a sufficient cause of an effect, depends on one’s theory of the causal
relation. Could it therefore be the case that viewed from the perspective of
one theory of causation, the empirical evidence suggests that at every time at
which a physical event has a cause it has a sufficient physical cause, whilst
viewed from the perspective of another theory of causation which entails a
broader understanding of what it is to be a cause, the very same empirical
evidence suggests no such thing?
A further issue concerns whether it is correct to assume, as this formula-
tion of the principle of causal closure does, that every physical event that has
12 S . C. GIBB
a cause has a sufficient cause. More specifically, is the principle’s assumption
that complete causes are always sufficient for their effects (where a cause is
sufficient for an effect if it causally necessitates it) correct according to all
theories of causation? Or, do some theories of causation allow cases in
which effects do not have sufficient causes? If so, does this leave a potential
gap in the argument from causal overdetermination?
Finally, even if one accepts the principle of causal closure, do certain
accounts of the causal relation still provide reasons to question other aspects
of the argument from causal overdetermination? Is the principle of causal
non-overdetermination equally plausible under various accounts of the
causal relation? Do certain accounts of the causal relation raise issues
regarding how this principle should be formulated? Is the argument from
causal overdetermination’s assumption that the causal relation is homo-
geneous correct? Or, do certain accounts of the causal relation allow one
room to question whether mental and physical causation differ, not merely
in what they relate, but also in the kind of causation that they involve?
Here, we have presented some of the issues that might be raised regarding
the ontology of the mental causation debate. We invited philosophers both
from within the mental causation debate and from within metaphysics to
think about the ontology of the mental causation debate. This book is a
product of what they think. The papers fall into two groups—those that
focus on issues relating to non-reductive physicalism and those that are
concerned with providing an alternative account of mental causation based
either on the acceptance of a powers ontology and/or a specific account of
the causal relata. All of the papers provide crucial insights into the relevance
of the causal relata or property analysis or the causal relation to the mental
causation debate, and, indeed, some of the papers touch upon all three of
these ontological areas.
John Heil’s paper ‘Mental Causation’ serves as an ideal point of entry into
the issues concerning the ontology of mental causation, highlighting some
of the questionable metaphysical assumptions that have arguably played a
role in the mental causation debate. One of Heil’s central aims in this paper
is to show that, given a proper understanding of properties, it is plausible
that the mental–physical distinction is not ontologically deep—that the
distinction is a distinction of conception only. This draws into question
appeals to levels of being—and, more specifically, non-reductive physicalist
attempts—to solve the problem of mental causation.
INTRODUCTION 13
Unlike Heil, the next three authors are united in considering that non-
reductive physicalism provides an attractive response to the problem of
mental causation. However, their accounts of non-reductive physicalism
differ greatly. According to Sydney Shoemaker’s account of non-reductive
physicalism, mental states are distinct from, but realized by, physical states,
where ‘realization’ is to be understood according to the subset account of
realization. In ‘Physical Realization without Preemption’, Shoemaker both
expands and corrects the account of the realization relation between mental
and physical states that he presented in his book Physical Realization.
In ‘Mental Causation in the Physical World’, Peter Menzies presents a
new version of Kim’s exclusion argument that targets all forms of physical-
ism, both reductive and non-reductive.14 Focusing on the principle of causal
exclusion to which this argument appeals, Menzies explores the extent to
which this principle is supported by different theories of the causal relation.
He argues that within the framework of a difference-making account of
causation, the principle is false. However, under special conditions, a
more plausible version of the principle of causal exclusion is supported
by this account of the causal relation. Menzies goes on to consider the
consequences—both surprising and encouraging—that this has for non-
reductive physicalism.
Paul Noordhof agrees with neither of these non-reductive physicalist
approaches. In ‘Mental Causation: Ontology and Patterns of Variation’ he
instead takes elements of those accounts which seek to understand the
efficacy of the mental non-ontologically in terms of patterns of variation
and elements of those accounts which attempt to resolve the problem of
mental causation by identifying mental property instances with physical
property instances, to motivate his own distinctive kind of non-reductive
physicalism.
David Papineau in ‘Causation is Macroscopic but Not Irreducible’ argues
that causation is not physically fundamental, but instead an essentially
macroscopic phenomenon. Causal processes are, according to Papineau,
analogous to thermodynamic processes, depending on probabilistic facts
about the ways in which macro-states are realized at the micro-level. This
is compatible with the popular non-reductive physicalist claim that causes
should be ‘proportional’ to their effects, and that the requirements of
14 For Kim’s version of this argument, see, for example, Kim (2005).
14 S . C. GIBB
proportionality can favour mental states as causes over their physical real-
izers. However, Papineau argues that these claims do not vindicate non-
reductive physicalism. He defends the thesis that mental properties are
instead reducible to some common physical feature of their realizers.
The first five papers in the second part of this book, unlike many of the
papers in the previous part, reject non-reductive physicalism. The different
accounts of mental causation that they advance hinge either on their
acceptance of a powers ontology and/or their understanding of the causal
relata. The first three papers place their discussion of the mental causation
debate in the context of recent developments in the ontology of powers.
In ‘Substance Causation, Powers, and Human Agency’, E. J. Lowe bases
his discussion within the general framework of a power-based theory of
substance causation. Lowe argues that the human will is a unique kind
of spontaneous power that is distinguished by the facts that it is a ‘two-way’
power and that it can be exercised rationally. Having such a power provides
human agents with all of the control that they could need over their
voluntary actions. Moreover, to deny that we have such a power is to
deny our own rationality.
Jonathan Jacobs and Timothy O’Connor’s paper ‘Agent Causation in a
Neo-Aristotelian Metaphysics’ explores how embracing a neo-Aristotelian
metaphysics of causal powers affects accounts of the nature of metaphysical
freedom. Jacobs and O’Connor argue that different versions of the neo-
Aristotelian metaphysics result in different metaphysical accounts of free and
responsible action. Consequently, action theory cannot properly be pursued
in isolation from general metaphysics.
In ‘Mental Causation and Double Prevention’, S. C. Gibb offers a new
solution to the problem of mental causation which has emerged from her
acceptance of a powers theory of causation. Gibb proposes that the causal
role of mental events in the physical domain is to serve as ‘double prevent-
ers’. She argues that if mental events are double preventers, and a powers
theory of causation is accepted, then the premises of the argument from
causal overdetermination can be reconciled with dualism.
David Robb’s response to the problem of mental causation depends on
his account of the causal relata, and, more specifically, his understanding of
the nature of a property. According to Robb, if properties are tropes then
this allows one to accept the conclusion of the argument from causal
overdetermination. In ‘The Identity Theory as a Solution to the Exclusion
INTRODUCTION 15
Problem’ he elaborates on this solution and responds to various objections
that have been raised against it.
Like Robb’s account, Peter Simons’ response to the problem of
mental causation arises from careful consideration of the causal relata. His
paper ‘Continuant Causation, Fundamentality, and Freedom’ distinguishes
between ‘continuant causation’ and ‘occurrent causation’. Simons maintains
that in the case of mental causation, the cause is not an event or other
occurrent, but instead a continuant. Hence, mental causation is continuant
causation. Simons goes on to argue that continuant causation is not funda-
mental, but is instead ontologically secondary to occurrent causation.
In the final paper, ‘There is no Exclusion Problem’, Steinvor Tholl
Arnadottir and Tim Crane argue that, contrary to many of the papers in
this book, to resolve the problem of mental causation one does not need to
take a stand on any controversial issues regarding either the nature of
causation or the ontology of the mental. A non-reductive physicalist or an
emergentist who accepts that mental and physical causes are not independ-
ent of one another does not face the problem of mental causation. They
would only face the problem if it were the case that an event cannot have
distinct but dependently sufficient causes. This is implausible. One does not
need to make any heavy-duty ontological claims to establish this.
References
Armstrong, D. (1997). A World of States of Affairs. Cambridge: Cambridge Univer-
sity Press.
Bennett, J. (1988). Events and their Names. Oxford: Clarendon Press.
Crane, T. (1995). ‘The Mental Causation Debate’. Proceedings of the Aristotelian
Society, Supp. Vol. 69: 211–36.
Davidson, D. (1980a). Essays on Actions and Events. Oxford: Clarendon Press.
—— (1980b). ‘Mental Events’. In Davidson (1980a): 207–28.
—— (1980c). ‘Psychology as Philosophy’. In Davidson (1980a): 229–44.
—— (1980d). ‘The Material Mind’. In Davidson (1980a): 245–60.
—— (1993). ‘Thinking Causes’. In Heil and Mele (1993): 3–18.
Gibb, S. C. (2006). ‘Why Davidson is not a Property Epiphenomenalist’. Inter-
national Journal of Philosophical Studies, 14(3): 407–22.
Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.
16 S . C. GIBB
—— (2004). Philosophy of Mind. Cambridge: Cambridge University Press.
—— A. R. Mele (eds.) (1993). Mental Causation. Oxford: Clarendon Press.
Honderich, T. (1982). ‘The Argument for Anomalous Monism’. Analysis, 42: 59–64.
Hopkins, J. (1978). ‘Mental States, Natural Kinds and Psychophysical Laws’.
Proceedings of the Aristotelian Society, Supp. Vol. 52: 221–36.
Kim, J. (1993a). ‘Events as Property Exemplifications’. In Supervenience and Mind.
Cambridge: Cambridge University Press: 33–52.
—— (1993b). ‘Postscripts on Mental Causation’. In Supervenience and Mind. Cam-
bridge: Cambridge University Press: 358–68.
—— (1993c). ‘Can Supervenience and “Non-Strict Laws” Save Anomalous
Monism?’. In Heil and Mele (1993): 19–26.
—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-
sity Press.
Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.
—— (2000a). ‘Causal Closure Principles and Emergentism’. Philosophy, 75: 571–86.
—— (2000b). An Introduction to the Philosophy of Mind. Cambridge: Cambridge
University Press.
—— (2006). The Four-Category Ontology: A Metaphysical Foundation for Natural
Science. Oxford: Oxford University Press.
—— (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford
University Press.
McGinn, C. (1989). ‘Can We Solve the Mind–Body Problem?’Mind, 98: 349–66.
McLaughlin, B. (1993). ‘On Davidson’s Response to the Charge of Epipheno-
menalism’. In Heil and Mele (1993): 27–40.
O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:
Oxford University Press.
Papineau, D. (2001). ‘The Rise of Physicalism’. In C. Gillett and B. M. Loewer
(eds.), Physicalism and its Discontents. Cambridge: Cambridge University Press:
3–36.
Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:
178–94.
Sosa, E. (1993). ‘Davidson’s Thinking Causes’. In Heil and Mele (1993): 41–50.
Taylor, R. (1966). Action and Purpose. Englewood Cliffs, NJ: Prentice-Hall.
INTRODUCTION 17
1
Mental Causation*
JOHN HEIL
Howcouldmental events, states, or properties causally affect, or be affected by,
physical events, states, or properties? The question persists even though few
philosophers nowadays find themselves attracted to the kind of uncomprom-
ising dualism associated with Descartes, the most prominent source of the
modern mind–body problem. “Physicalism” rules. Mental items are, at the
very least, taken to be ‘dependent on’ physical goings-on.Cartesian sentiments
linger, however, in the widespread acceptance of a sharp-edged distinction
between mental properties and physical properties. “Non-reductive physical-
ists,” for instance, hold that, although mental properties are in some way
wholly dependent on physical properties, mental properties are nevertheless
distinct from physical properties. Physical properties, or their instances, ‘real-
ize’ mental properties, or their instances. A denial of distinctness is taken to
require either reduction (the mental is reducible to the physical; mental
properties are physical properties) or outright elimination (nothing thinks,
perceives, or feels: mentality is an illusion).
By my lights, almost everything expressed in the previous paragraph stems
from philosophical confusion that finds its source in the linguisticization of
metaphysics. I do not claim that the problem ofmental causation can be swept
away by a few deft philosophical flourishes. I do claim, however, that we have
managed to distract ourselves—and our students—from the real problems.
My procedure will be to offer up without much argument a handful of
observations that, taken together, are meant to afford a measure of hope.
* The paper has benefited from comments from a number of people, especially S. C. Gibb and
E. J. Lowe.
Causation
The problem of mental causation has many faces. You might wonder
whether mental states or events cause, or are apt to cause physical states or
events. You might wonder whether mental states or events cause whatever
they cause in virtue of beingmental. The question I shall address, however, is
not whether mental goings-on cause, or are “causally relevant” to, physical
goings-on, but how this could be possible, how it might work. In answering
this question, it is not to the point to call up examples of accepted causal
explanations in psychology or neuroscience as philosophers of science have
sometimes done (see, e.g., Woodward 2008). Princess Elizabeth does not
doubt that minds and bodies interact. Her challenge to Descartes is to
account for mind–body interaction given Descartes’ commitment to dual-
ism. Nowadays the difficulty is, more often than not, taken to be that of
accounting for mental causation, or the “causal relevance” of the mental,
given the truth of non-reductive physicalism.
I believe that the prospects of reconciling mental causation and non-
reductive physicalism are vanishingly small. This I find not at all troubling
because I have little enthusiasm for non-reductive physicalism. Indeed
I regard non-reductive physicalism as a kind of blight on the late twentieth-
century philosophical imagination. Abandoning non-reductive physicalism is
not something many philosophers seem prepared to do, however. Those
philosophers owe us all an account of how mental causation could work if the
non-reductivist picture is correct.
I have noted that one unsatisfying kind of answer would feature appeals to
empirical researchwidely taken to supportmental–physical causal claims. The
issue before us is notwhether the mental bears causally on the physical, but how
this could be so given the non-reductive physicalist’s metaphysical picture.
Another equally unsatisfying answer would incorporate a weakening of
the notion of causation so as to get the right answer. On the one hand, you
have mental–physical causation. On the other hand, you have mental
properties or states being distinct from, but dependent on physical proper-
ties and states. You need only massage the concept of causation in a way that
reconciles these two facts about the world.
Philosophers who solve the problem of mental causation by taking caus-
ation to be bare counterfactual dependence (or counterfactual dependence
MENTAL CAUSATION 19
suitably constrained), provide an illustration of what I have in mind. What,
you might wonder, grounds the pertinent counterfactuals? What is it about
the world in virtue of which these counterfactuals hold true? If you allow the
counterfactuals to be “barely true”, and if you embrace the idea that causality
is solely a matter of counterfactual dependence, the mind–body problem, the
problem of how the mental and physical could causally interact, evaporates.
Such an account would rescue a Cartesian, but that very fact ought to arouse
suspicion. To the extent that you regard the mind–body problem as a genuine
problem for Descartes, you are thinking of causation as something more than
bare counterfactual dependence.
A third ontologically barren category of response involves appeal to
psychological or psychophysical ceteris paribus laws figuring in psychological
explanation. Grant that accepting such laws amounts to accepting psycho-
physical causation. We are now back with the question how such causation
could be possible if non-reductive physicalism is true. In every case we are
led back to the non-reductivist’s ontology.
Non-Reductive Physicalism
Mainstream non-reductive physicalists embrace three theses.
(1) Distinctness: mental properties are distinct from physical properties.
(2) Dependence: mental properties depend on physical properties.
(3) Autonomy: the physical realm is causally self-contained.
Taken together, these three theses make it hard to see how mental properties
could “make a causal difference” to physical goings-on.Autonomy implies that
physical effects have purely physical causes. Distinctness ensures that mental
properties are removed from the physical causal mix. Dependence seems to
render mental properties inefficacious even as regards mental effects: a mental
effect depends on a physical effect; a mental property could be “causally
relevant” to the production of a mental effect only by playing a role in the
production of the physical effect on which the mental effect depends.
Figure 1.1, associated with the work of Jaegwon Kim, will be familiar to
most readers (Kim 1978, 1979, 1982, 1984, 1993a, 2005). HereM1 andM2 are
mental property instances, states, or events, P1 and P2, physical “realizers” of
these. Vertical double-arrows * represent ‘vertical’ dependence relations,
20 JOHN HEIL
and the horizontal arrow ! indicates a causal relation. P1 figures unprob-
lematically in the production of P2. It looks as though, owing to autonomy,
M1 could play no role in the production of P2, and thus, given dependence,
no role in the production of M2.
An industry dedicated to resolving this problemwhile preserving (1)–(3) has
yielded an assortment of ingenious philosophical theses. Kim himself thinks
that the only solution is to identify mental properties with their physical
realizers. To the extent that this appears unpromising (as in the case of qualia),
mental phenomena must be epiphenomenal. Sydney Shoemaker (2001, 2003)
and Dirk Pereboom and Hilary Kornblith (1991; see also Pereboom 2002)
defend an account of the “realizing” relation that builds instances of, or causal
powers “bestowed” by, mental properties into their realizers. John Gibbons
(2006), taking off from Fodor’s (1997) conception of properties as figuring
essentially in causal laws, argues that mental causes could have mental effects
without compromising autonomy. Karen Bennett (2003) suggests that mental
causes could “over-determine” physical effects in an unobjectionable sense
consistent with autonomy (see also Yablo 1992; Mills 1996; Thomasson 1998).
And Cynthia and Graham Macdonald (2006) echo others, including LePore
and Loewer (1987), andWoodward (2008) in construing causation as a species
of counterfactual dependence: becauseM2 (and very likelyP2, aswell) depends
counterfactually onM1, we are entitled to acceptM1 as a causal factor.
I don’t think it would be much of an exaggeration to say that none of
these views has attained anything approaching universal support. Their costs
exceed their benefits.
“Mental” and “Physical”
The tenets of non-reductive physicalism revolve around the distinctness of
the mental and the physical. The key breakthrough, it has been thought, lay
in recognizing that distinctness can be consistent with dependence: mental
M1 M2
P1 P2→
⇑ ⇑
Fig 1.1
MENTAL CAUSATION 21
properties, although distinct from physical properties, in some way depend
on physical properties. This idea was expressed in the 1970s and 1980s in
terms of supervenience: the mental supervenes on the physical, a doctrine
traceable to Davidson’s “Mental Events”: “there cannot be two events alike
in all physical respects but differing in some mental respect, or an object
cannot alter in some mental respect without altering in some physical
respect” (Davidson 1970, 214).
For a while supervenience seemed to provide an elegant, cost-free solu-
tion to the problem of mental causation. If mental properties supervened on
physical properties, and physical properties figured unproblematically in
causal transactions, thenmental properties, too, could be creditedwith causal
efficacy. When this happens, we have “supervenient causation” (Sosa 1984).
Initial euphoria over supervenient causation quickly faded, however.
Supervenience yields dependence, but dependence is trumped by distinct-
ness. Frederick Stoutland (1976) was (to my knowledge) the first of many to
identify the problem of “causal relevance.” Even if you concede the “token
identity” of mental and physical events, you can still ask whether a given
event had a particular effect in virtue of that event’s being mental or in
virtue of its being physical. Suppose your experiencing a painful sensation is
“token identical” with your C-fibers’ firing, and suppose your C-fibers’
firing causes your arm to move in a particular way. You can still ask whether
the event that caused your arm to move did so qua physical or qua mental,
whether the event caused your arm to move in virtue of its physical
properties or in virtue of its mental properties. We have token identity,
but type diversity—distinctness. And type diversity, coupled with depend-
ence and autonomy, threatens what Brian McLaughlin (1989) dubbed “type
epiphenomenalism” (nicely captured by Figure 1.1).
Most parties to these discussions regard causal relations as holding among
distinct token events. Thus, in a token identity thesis, tokens are meant to
be particular events; types are properties. Davidson, who is often credited as
the inspiration behind non-reductive physicalism, is taken to defend token-,
but not type-identity: mental events are physical events, but mental prop-
erties are distinct from physical properties. Indeed, the three principles with
which I led off are directly traceable to Davidson’s defense of “anomalous
monism.” But what, for Davidson, distinguishes the mental from the
physical? What distinguishes mental and physical types?
22 JOHN HEIL
Before answering this question, let me call your attention to a remarkable
feature of much of the discussion of mental causation over the past forty
years. Although philosophers insist on distinctness—mental properties are
distinct from physical properties—scant attention has been paid to what the
distinction is (Mellor 1973 and Crane and Mellor 1990 represent notable
exceptions). Philosophers involved in the debate seem to think that the
distinction is so obvious as not to be worth further discussion. This stems
in part from views on the autonomy of psychology (and other higher-level,
special sciences). If psychology is not reducible to neurobiology, psycho-
logical propertiesmust be distinct from neurological or biological properties.1
Philosophers sometimes characterize properties by reference to predicates
discoverable in formulations of laws of nature. We can see that predicates
deployed in psychology differ in their application conditions from predi-
cates occurring in various physical sciences: physics, chemistry, biology. We
have no reason to think that psychological predicates could be analyzed in
terms of, translated into, or otherwise defined in a non-psychological
vocabulary. Psychological descriptions and explanations appear autono-
mous with respect to the physical sciences. Physical properties, then, are
properties named by predicates figuring in laws of nature investigated by the
various physical sciences; mental properties are those designated by respect-
able psychological predicates.
On a view of this kind, there will be hosts of properties other than
those designated by predicates of the fundamental physical sciences
(biological properties, meteorological properties, geological properties,
sociological properties), mental properties being only one class of these.
Non-reductivists are fond of pointing out that philosophers who doubt the
causal efficacy of mental properties are obliged to doubt as well the causal
efficacy of biological, meteorological, geological properties, and socio-
logical properties. Scientific practice renders such a view comical, an
expression of the worst sort of philosophical hubris.
Returning to the mental–physical distinction, the move, as I have char-
acterized it, is a move from differences in conception to distinctions in
reality. Such a move is resistible. You can accept fundamental differences in
1 And of course neurological or biological properties will be distinct from properties figuring in laws
of still lower level sciences, thus spawning a hierarchical conception of reality, an updated version of the
Great Chain of Being.
MENTAL CAUSATION 23
conception, while doubting that these reflect real distinctions, fundamental
differences in reality.2 This is how a philosopher such as Spinoza might
think of the mental–physical dichotomy. It is also how Davidson (who cites
Spinoza explicitly: 1970, 212) is thinking of it. Events, for Davidson, are
mental if they answer to mental descriptions, physical if they are physically
describable. Every mental event, every event satisfying a mental predicate,
Davidson thinks, is a physical event, an event describable in a physical
vocabulary. On such a view, the mental–physical distinction is not onto-
logically deep, not a real distinction.
Many readers will be surprised to hear such a view attributed to David-
son. But only a philosopher who hears “predicate” as “property,” only a
philosopher who regards predicates, or predicates ineliminably deployed to
express truths about the world, as invariably naming real properties, could
possibly read Davidson any other way.3 My aim is not to engage in
exegetical debates about Davidson, however, but merely to place on the
table a position that seems rarely to have been considered by metaphysically
inclined philosophers of mind. If the mental–physical distinction is a
distinction of conception only, then there is no pressing need to nail
down an interesting, non-circular way of distinguishing mental and phys-
ical properties. There are properties, all right, properties of objects that can
serve as truthmakers for claims about the world framed in a mental or a
physical vocabulary. If Davidson is right, whenever you have a true mental
description of an event, a description framed in a psychological vocabulary,
a description of the same event couched in the language of physics could, at
least in principle, be formulated—although not necessarily by the wielder
of the mental description. This is what supervenience boils down to for
Davidson.
A view of this kind ought not be counted “physicalist.” In no sense does
it privilege the physical. Mental events can be picked out using a physical
2 Descartes was acutely aware that many of his critics were skeptical of his defense of the “real
distinction.” Modern interpreters have understood the critics as arguing that one and the same substance
could possess both mental and physical properties. A view of this kind, supplemented by Dependence,
amounts to non-reductive physicalism. I think it likely that at least some of Descartes’ critics regarded the
mental–physical distinction as one of conception only, not a real distinction—between kinds of
substance or kinds of property. See Strawson 2008.3 In the literature on mental causation, it has been customary to slide from talk of predicates to
property talk. Davidson, in contrast, regards talk of properties as an oblique way of talking about
predicates; see Davidson (1993).
24 JOHN HEIL
vocabulary, but physical events can be referred to via mental descriptions.
Application conditions for mental predicates are orthogonal to those for
physical predicates, so there is no possibility of reduction—at least not if
reduction requires translation of statements in one vocabulary to statements
in another, or the expression of application conditions of predicates in one
vocabulary in the vocabulary of the other.
I have put all this in terms of descriptions and predicates, but really all you
need is the thought that worldly goings-on can be regarded in very different
ways for different purposes. This could be so without its being the case that
there is just one correct way to regard the world. We turn to physics to
provide something like an exhaustive description of the fundamental things.
But biologists, meteorologists, and sportscasters also manage to give us true
descriptions that fasten onto broad similarities and differences that serve us
well in making sense of and predicting features of the world around us.
These similarities and differences are perfectly objective, mind-independent
in the pertinent sense.
You might be appalled at the thought that the mental–physical distinc-
tion is not ontologically deep. But bear with me, and consider its implica-
tions for “non-reductive physicalism.” If the physical—that is, the world
described via a physical vocabulary—lacks priority, reduction is out of the
picture. With the threat of reduction off the table, the motivation for
regarding mental items—items picked out by means of mental terms—as
“higher-level” entities dependent on, but distinct from, lower-level phys-
ical items evaporates. Mental states and events can have physical conse-
quences. And, as Davidson notes, the question whether these consequences
had mental or physical causes cannot arise.
Token Identity, Type Diversity
Non-reductive physicalism is regarded by its proponents as embracing
token identity and rejecting type identity. If you thought of causation as a
relation among events (and who doesn’t?), the tokens would be particular
events.4 What of types? Non-reductivism takes types to be properties,
4 Do not read this as an endorsement of “event causation.” I have severe reservations about the idea
that causal relations are asymmetrical relations among events.
MENTAL CAUSATION 25
properties to be universals.5 An event is mental if it includes a mental
property, physical if it includes a physical property. If you think that the
mental is in some fashion “determined” by the physical, then every mental
event will be a physical event: every event including a mental property
includes, as well, a physical property.
This is the picture that gives rise to the problem of causal relevance.
Given that mental events are physical events, mental events could be causes
if physical events could be causes. Difficulties arise, however, when you ask
whether a given event had the effect it had in virtue of being a mental event
or in virtue of being physical. Your forming the intention to move your
finger (thereby flipping the switch) is identical with some event in your
nervous system. When your forming the intention to move your finger
causes your finger to move, does it do so quamental, qua being an intention,
or qua physical, qua being a particular kind of neurological event? The
question needs only to be asked to raise doubts that the event’s being mental
could have played a role in the production of a physical sequence culminat-
ing in your finger’s moving.
Many philosophers have weighed in on this topic, offering ingenious
ways of insinuating mentality into the causal act, but if we are honest, we
shall have to admit that the results have not been encouraging. My own
belief is that a sensible answer to the qua problem requires a rejection of the
muddled ontology that gives rise to it in the first place. I have hinted at
how this might work in discussing Davidson. Before saying more, how-
ever, it is worth pausing to look more carefully at the underlying onto-
logical picture.
First, what exactly is an event? Philosophers sometimes divide the terri-
tory between “Davidsonian events” and “Kim-style events.” Davidsonian
events are thought of as “coarse-grained”; Kim’s are “fine-grained.”
According to Kim, an event is a substance’s having a property at a time:
a’s being F at t. (I shall omit the time reference in what follows.) Now
suppose thatM is a mental property and P is a physical property. In that case,
a’s being M and a’s being P are distinct events. There is token identity, a is
5 A variant substitutes collections of exactly resembling tropes for universals. Although ontologically
momentous, the distinction has no discernible effect on arguments I address here, so I shall ignore it in
what follows.
26 JOHN HEIL
identical with a, but the events are distinct—in the way a ball’s being red and
its being round are distinct. The events are not token identical.6
What of Davidson? I have argued that, for Davidson, the mental–physical
distinction is one of conception only. So for Davidson, a given token event
can answer to both a mental and a physical description. A single token event
answers to each description. Notice that this has nothing to do with the
event’s being “coarse-grained.” Were he ontologically less timid, Davidson
could accept Kim’s accounting of the ontology of events without modifying
his account of mental causation.
So? Well, it is hard to see how the qua problem could arise for either
Davidson or Kim. For Davidson, or my Davidson, it is crazy to ask whether
a given event caused what it did because it was described as mental or
because it was described as physical. And for Kim, the mental and physical
events are distinct if their constituent mental and physical properties are
distinct. You might wonder which event had a given effect, but not
whether a single event had the effect it had because it was mental or because
it was physical.
Properties of Events
In a paper published in 2006, Cynthia andGrahamMacdonald accept a Kim-
style conception of events, but proceed to argue that particular events can
cause what they do in virtue of their mental properties. When I introduced
this topic, I spoke of events “including” mental and physical properties.
I adopted this awkward form of speech because I wanted to leave open
whether the properties in question were constituents of the events. For the
Macdonalds, an event, a’s being F, can itself have a property. This property
can be, indeed must be, distinct from F, yet be “causally relevant” to the
event’s producing a given effect, provided various conditions are satisfied.
A view of this kind, might seem to allow that a mental event (an event
possessing a mental property) could be token identical with a physical event
(an event possessing a physical property) thereby leaving room for a qua
6 Does it matter that M supervenes on P? Not if supervenience is understood in the usual way: the
bearer of M is not P, M is not a second-order property, not a property of a property, but a property
possessed by a in virtue of a’s possession of P.
MENTAL CAUSATION 27
problem. The question is, why would you want to do this? Why go to a lot
of trouble to re-introduce a problem, then to a lot more trouble to devise a
convoluted solution to cope with it? It is hard not to think that the
Macdonalds are impressed by the usual way of describing Davidson’s view
as implying token, but not type, identity, and providing an explicit onto-
logical picture of events as, not only being partly constituted by properties but
as, in addition, possessing properties.
I find the Macdonalds’ position deeply uncongenial, but I am not con-
cerned here with its details. Rather I want to focus just on the idea of events
bearing properties. You can predicate—truly—many things of events, but it
needn’t follow from this that the truthmakers of such predications are
properties of those events. This is not a deep or technical point, merely an
expression of bewilderment over what it could be for an event—and here
we are thinking of “Kim-style” events, a’s being F—to bear a property.
Substances bear properties. I doubt that properties have properties, but even
if they did, it is very hard to see the Macdonalds’ events as being ontologic-
ally suited to be bearers of properties. Think of it. You have a’s being F
possessing, say G. The bearer of G here is not a; a’s being the bearer would
be fine, but then G would belong to a, not to the event, not to a’s being F.
Further,G’s belonging to a would in fact be a constituent of a new event: a’s
being G. Similarly, the bearer of G couldn’t be F. No, G must be possessed
by a’s being F, or, more particularly, by a’s being F at t.
You will not share my bewilderment at this thesis if you are content to
suppose that every true predication designates a property. Such a view turns
properties into what David Armstrong calls shadows of predicates. Under-
standing what it could mean for an event—a’s being F—to possess a
property, requires more than an appeal to the fact that we can and do
describe events using an assortment of predicates. Events can be brief or
drawn out, for instance, but what makes it true that an event is temporally
extended in a particular way need not be the possession by the event of a
temporal property, whatever that might mean.
The worry here is less an expression of a quirky ontological preference,
than a simple inability to comprehend what it would be for an event, as
characterized by the Macdonalds, to bear a property. There is certainly a
place for a relaxed conception of properties, according to which having a
property is simply a matter of answering to a predicate. When the question
is whether properties of a certain kind could figure in causal relations,
28 JOHN HEIL
however, you need an ontologically serious conception, you need to move
beyond the thought that, if you can predicate something truly of an event,
this must be because the event possesses a property corresponding to your
predicate (and shared by anything to which the predicate truly applies).
I do not consider these remarks on properties of events to amount to
anything approaching a knock-down argument. I will be satisfied if I have at
least pricked the conscience of those for whom talk of properties of events
trips off the tongue.
Monism
Moving on, then, let us suppose, at least for the sake of argument, that
causation is a relation among events and that events are substances’ possess-
ing properties at times. Now consider a particular causal sequence in which
your forming an intention to move your finger causes your finger to move.
If Davidson is right, there is a description of this sequence expressible in the
language of fundamental physics that is an instance of a fundamental,
“exceptionless” law. For Davidson, laws are linguistic items, statements.
Davidson is silent as to the truthmakers for law-statements, but we need not
be.7 If you are a Humean, the truthmakers will be uniformities, patterns of
events of particular types specifiable in the language of fundamental physics.
If you are like me, you will think the truthmakers are powers possessed by
the objects involved in virtue of those objects’ properties and relations
among these.
Whatever your view, the idea would be that what answers to a statement of
this fundamental law also, on this occasion, answers to a singular causal claim
couched in a psychological vocabulary. Because application conditions for
psychological predicates differ markedly from application conditions for
predicates deployed in fundamental physics, there is no prospect of framing
the former in terms of the latter. This, this conceptual or taxonomic mis-
match, is the anomalousness of the mental. The truthmaker for the original
psychological assertion is a state of the world that could be described in a fine-
grained physical vocabulary. The description would be, from the perspective
7 You might think that Davidson rejects the notion of truthmaking but I don’t think he has the
concept.
MENTAL CAUSATION 29
of physics, sprawling and ungainly. There is little reason to think—and if
arguments for “multiple realizability” are to be taken seriously, every reason
not to think—that there could be interesting connections between mental and
physical descriptions.
The emerging picture is of one world describable—truly describable—in
many different ways. The world, or rather ways the world is, serve as truth-
makers for these descriptions. If fundamental physics is in the business of
giving an exhaustive, maximally fine-grained description of the world, this in
no way excludes biological, anatomical, geological, meteorological, psycho-
logical, anthropological, or journalistic descriptions of worldly goings-on.
The mistake is to imagine that biology and the rest describe worlds distinct
from but dependent on the world described by physics. If physicalism is the
doctrine that every truth about the world could be expressed in the vocabu-
lary of fundamental physics, then this is not physicalism.
Powerful Qualities
All well and good, but doesn’t monism face the problem of finding a place
for the qualia in a universe of quarks and electrons? If you can move beyond
the thought that mental predicates designate a distinctive realm of proper-
ties, you can begin to think seriously about what truthmakers for psycho-
logical assertions, including assertions concerning qualities of conscious
experiences might be.
This is not the place to attempt a full-scale discussion of qualia, but I can at
least say why I think the non-reductive physicalist mind-set makes a difficult
problem much more difficult.
One facet of what might be thought of as the standard conception of the
physical world is the thought that qualities are exclusively mental. The
ancestor of this idea lies in the use to which the distinction between primary
and secondary qualities has been put by philosophers who have tended to
regard the physical sciences as the measure of all things, the source of all
truths. The primary qualities (shape, size, mass, charge, and the like) are
taken to be properties of physical objects. Secondary qualities (colors,
sounds, tastes, smells, etc.) are relegated to the minds of observers.
A conception of this kind bifurcates the world into minds and everything
else. Certain properties belong to minds that do not, indeed could not, belong
30 JOHN HEIL
to physical objects. If that is right, then there is no prospect of finding a place
for minds in the physical order.
You can see this conception at work in the thesis that physical properties
are powers: to have mass is to have the power to affect and be affected by
other massy things in particular ways. Most often, the thought that physical
properties are powers includes the thought that this exhausts their nature.
The result is a conception of the physical world as a world of “pure
powers,” a conception that apparently fits well with scientific practice.
Coupled with the idea that powers are individuated relationally, we are
led to the “structural realist” construal of the world as constituted by
relations (Ladyman 2007). Finding a place for qualities of conscious experi-
ences in such a world looks hopeless. At best such qualities are add-ons,
ontological fuzz, foam on the sea of being.
Suppose, however, you distinguished features of scientific accounts of the
world from features of the world in virtue of which those accounts are true.
A relational vocabulary, for instance, might suffice to represent the truths of
fundamental physics (Dipert 1997). From this, however, it does not follow
that truthmakers for assertions couched in a relational vocabulary are irredu-
cibly relational states of affairs (Heil 2009; see also Parsons 2009). Nor
from the fact that scientific explanation omits reference to qualities does it
follow that the world thus explained is qualitatively barren. To think
otherwise, to think that the assertion that the fundamental properties are
powers implies that they are not qualities, is to move from “the a’s are F” to
“the a’s are not G.” Properties could be powers, and powers could be
qualities; properties could be powerful qualities.
I believe there are excellent reasons to think this is so, to think that
properties are powerful qualities, for reasons having to do with the individu-
ation of powers.8You need not be fully convinced of such a view, however,
to grant that it is at least an option. The idea that properties are powerful
qualities narrows the gap between the qualia and everything else. If everything
has qualities, it is unsurprising that conscious states are qualitatively imbued.
States of mind differ qualitatively from states of an ordinary computing
machine. But qualitative differences are potentially explicable.
8 Berkeley argued the point but assumed that qualities must be immaterial. More recently, it has been
argued by Campbell (1976, 93), Martin (1997, 2008), Armstrong (1999), and Unger (2006). See Heil
(2008, 2010) for discussion.
MENTAL CAUSATION 31
Ontological Progress
I am not so foolish as to imagine that these brief remarks could persuade
skeptical readers that considering properties as powerful qualities affords an
immediate solution to the mind–body problem. Thinking of properties this
way does, however, nudge us in the right direction. Earlier, I described
Davidson as advancing the thesis that the mental–physical distinction is a
distinction of conception only, not a real distinction. This is all to the good,
but Davidson is concerned with just one class of psychological states, the
propositional attitudes. A natural reaction to Davidson’s argument would be
to concede that the argument might possibly provide a way of understand-
ing cases of mental causation involving beliefs, desires, intentions, and the
like, but to note, as Davidson himself notes, that the argument fails to
extend to conscious experiences.
This is where you need to move beyond Davidson’s hands-off approach
to ontology. Suppose I am right, suppose properties are powerful qualities.
This would mean that the distinction between qualities and powers is a
distinction of conception only, not a real distinction. Were that so, not only
would it be unsurprising that conscious experiences are qualitatively satur-
ated, but also that their qualities “make a causal difference.”
I believe we have excellent reasons to embrace an ontological picture that
regards properties as powerful qualities, reasons that have nothing in particular to
do with the mind–body problem. The mind–body problem as we have it
today—the problem of mental causation, the problem of causal relevance—
arises from acceptance of a very different ontological picture, one that has little to
recommend it. The influence of that picture stems, not from its being founded
on compelling arguments but from its permeating philosophers’ thoughts about
the mind, from its functioning as a lens through which we see problems and
evaluate competing solutions. Whether I am on the right track or not, the way
out is not to be found by making incremental ad hoc ocular adjustments—
adding epicycles to the prevailing,RubeGoldberg ontology—but by coming to
see that ontology as merely one among many possible ontologies.
References
Armstrong, D. M. (1999). ‘The Causal Theory of Properties: Properties According
to Ellis, Shoemaker, and Others’. Philosophical Topics, 26: 25–37.
32 JOHN HEIL
Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just
Maybe, to Tract It’. Nous, 37: 471–97.
Campbell, K. (1976).Metaphysics: An Introduction. Encino: Dickenson Publishing Co.
Crane, T. and D. H. Mellor (1990). ‘There is no Question of Physicalism’. Mind,
99: 185–206.
Davidson, D. (1970). ‘Mental Events’. In L. Foster and J. Swanson (eds.),Experience and
Theory. Amherst: University of Massachusetts Press: 79–101. Repr. in Davidson,
Essays on Actions and Events. Oxford: Clarendon Press. (1980, 2001): 207–25.
—— (1993). ‘Thinking Causes’. In Heil and Mele (1993): 3–18.
Dipert, R. R. (1997). ‘The Mathematical Structure of the World: The World as
Graph’. Journal of Philosophy, 94: 329–58.
Fodor, J. (1997). ‘Special Sciences: Still Autonomous after All These Years’.
Philosophical Perspectives, 11: 149–63. Repr. in Fodor (1998). In Critical Condition:
Polemical Essays on Cognitive Science and the Philosophy of Mind. Cambridge, MA:
MIT Press: 9–24.
Gibbons, J. (2006). ‘Mental Causation Without Downward Causation’. Philosoph-
ical Review, 115: 79–103.
Heil, J. (2008). ‘Critical Study of Peter Unger’s All the Power in the World’.Nous, 42:
336–48.
—— (2009). ‘Relations’. In R. Le Poidevin and R. Cameron (eds.), Routledge
Companion to Metaphysics. London: Routledge: 310–21.
—— (2010). ‘Powerful Qualities’. In A. Marmodoro (ed.), The Metaphysics of
Powers: Their Grounding and their Manifestations. London: Routledge: 58–72.
—— and A. R. Mele (eds.) (1993). Mental Causation. Oxford: Clarendon Press.
Hohwy, J. and J. Kallestrup (eds.) (2008). Being Reduced: New Essays on Reduction,
Explanation, and Causation. Oxford: Oxford University Press.
Kim, J. (1978). ‘Supervenience and Nomological Incommensurables’. American
Philosophical Quarterly, 15: 149–56.
—— (1979). ‘Causality, Identity, and Supervenience in the Mind–Body Problem’.
Midwest Studies in Philosophy, 4: 31–49.
—— (1982). ‘Psychophysical Supervenience’. Philosophical Studies, 41: 51–70.
—— (1984). ‘Epiphenomenal and Supervenient Causation’. Midwest Studies in
Philosophy, 9: 257–70.
—— (1993a). ‘The Non-Reductivist’s Troubles with Mental Causation’. In Heil
and Mele (1993): 189–210. Repr. in Kim (1993b): 336–57.
—— (1993b). Supervenience and Mind: Selected Philosophical Essays. Cambridge:
Cambridge University Press.
—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-
sity Press.
MENTAL CAUSATION 33
Ladyman, J. (2007). ‘Structural Realism’. In E. N. Zalta (ed.), The Stanford Encyclo-
pedia of Philosophy, available online at: http://plato.stanford.edu/archives/
win2007/entries/structural-realism/.
Lepore, E. and B. Loewer (1987). ‘Mind Matters’. Journal of Philosophy, 84: 630–42.
Macdonald, C. and G. Macdonald (2006). ‘The Metaphysics of Mental Causation’.
Journal of Philosophy, 103: 539–76.
McLaughlin, B. P. (1989). ‘Type Epiphenomenalism, Type Dualism, and the
Causal Priority of the Physical’. Philosophical Perspectives, 3: 109–35.
Martin, C. B. (1997). ‘On the Need for Properties: The Road to Pythagoreanism
and Back’. Synthese, 112: 193–231.
—— (2008). The Mind in Nature. Oxford: Oxford University Press.
Mellor, D. H. (1973). ‘Materialism and Phenomenal Qualities’. Aristotelian Society
Proceedings, Supp. Vol. 47: 107–19.
Mills, E. (1996). ‘Interactionism and Overdetermination’. American Philosophical
Quarterly, 33: 105–17.
Parsons, J. (2009). ‘Are there Irreducibly Relational Facts?’ In E. J. Lowe and
A. Rami (eds.), Truth and Truth-Making. Stocksfield: Acumen Publishing: 217–26.
Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,
99: 499–531.
Pereboom, D. and H. Kornblith (1991). ‘The Metaphysics of Irreducibility’.
Philosophical Studies, 63: 125–45.
Shoemaker, S. (2001). ‘Realization and Mental Causation’. In C. Gillett and
B. Loewer (eds.), Physicalism and Its Discontents. Cambridge: Cambridge Univer-
sity Press: 74–98.
—— (2003). ‘Realization, Microrealization, and Coincidence’. Philosophy and
Phenomenological Research, 67: 1–23.
Sosa, E. (1984). ‘Mind–Body Interaction and Supervenient Causation’. Midwest
Studies in Philosophy, 9: 271–81.
Stoutland, F. (1976). ‘The Causation of Behavior’. In J. Hintikka (ed.), Essays on
Wittgenstein in Honor of G. H. von Wright (Acta Philosophica Fennica, 28).
Amsterdam: North Holland Publishing: 286–325.
Strawson, G. (2008). Real Materialism and Other Essays. Oxford: Clarendon Press.
Thomasson, A. (1998). ‘A Non-Reductivist Solution to Mental Causation’. Philo-
sophical Studies, 89: 181–91.
Unger, P. (2006). All the Power in the World. New York: Oxford University Press.
Woodward, J. (2008). ‘Mental Causation and Neural Mechanisms’. In Hohwy and
Kallestrup (2008): 218–62.
Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.
34 JOHN HEIL
2
Physical Realization withoutPreemption
SYDNEY SHOEMAKER
I
This paper expands, but also corrects, the account I gave in my book
Physical Realization of how we can hold that mental states are physically
realized without being forced to allow that the causal efficacy of mental
states is preempted by their physical realizers. In part the correction of the
account in my book involves returning to an account of realization I put
forward in an earlier paper, and retracting a concession I made nearly thirty
years ago to an objection to the original formulation of my causal theory of
properties. The expansion of the account involves clarifying and developing
the claim that all properties of macroscopic objects are realized in what I call
MSE properties, properties things have in virtue of their careers embedding
microphysical states of affairs of a certain sort.
II
I begin with the well-known multiple realization objection to type physic-
alism. The objection can be presented as follows. Ask yourself what would
be the best possible evidence that a particular mental state, say pain, is
identical with a particular physical state, say C-fiber firing. One might at
first think that the best evidence would be the discovery of a universal
correlation between the mental state and the physical state—the discovery
that a subject is in pain if and only if there is C-fiber firing occurring in her
brain. But this first thought is certainly wrong, for the existence of such a
correlation is perfectly compatible with a dualist view which holds that the
mental and physical states are distinct, although universally correlated,
perhaps because the mental state is always caused by the physical one.
A much better answer is that the best possible evidence would be the
discovery that the physical state has the same causes, and the same effects,
as the mental state—e.g., that C-fiber firing is caused by cuts and burns, and
causes grimaces, moans, the taking of aspirins, and calls to the doctor. This is
sometimes put by saying that the physical state plays the same causal role as
the mental state. Playing the causal role of a state can be equated with
sharing its causal profile, where the causal profile of a state is constituted by
the ‘forward-looking’ causal features of the state, its being apt to contribute
in certain ways to the causing of certain effects, and its “backward-looking”
causal features, its being such that its instantiation can be caused in certain
ways. It is hard to see what could be a better reason for identifying pain with
C-fiber firing than their having the same causal profile.
But now we face the objection that having discovered that a certain
physical state seems to play the causal role of pain in us, we might discover
another species in which some different physical state plays this causal role.
Thus David Lewis imagined that while C-fiber firing plays the causal role of
pain in Earthlings, in Martians this causal role is played by the inflation of
tiny cavities in the feet (Lewis 1980). And Hilary Putnam suggested that
what plays the causal role of pain in octipi is a physical state quite different
from the one that plays that role in us (Putnam 1967). We might even
discover that this causal role is played by different physical states in different
members of our own species, or even that it is played by different physical
states at different times in the history of the same person. Obviously the
same mental state cannot be identical to a number of different physical
states. But if states are identical, they have to be necessarily identical, which
means that there is no possible circumstance in which they are distinct. So if
it is so much as possible that different physical states can play the causal role
of a given mental state, playing that causal role cannot make a physical state
identical with the mental state.
The different physical states that play the causal role of a mental state can
also be said to play its functional role. The property of having some state or
other that plays this role is a functional property, which is realized in, or has
as possible realizers, those physical states. And now it seems that it is this
36 SYDNEY SHOEMAKER
functional property, not any of its physical realizers, that a physicalist should
identify with pain.
But now we face a problem. One problem which the psychophysical
identity theory did not face is that of explaining how, on a physicalist view,
mental states can have causal efficacy. For if a mental state is identical with a
physical state, it will obviously cause whatever that physical state causes. If
pain is identical with C-fiber firing, then pain will cause whatever C-fiber
firing causes—and as noted earlier, what would motivate the identification
of pain with C-fiber firing is precisely the fact (or supposed fact) that C-fiber
firing causes the things we take pain to cause, and is caused by the things we
take to cause pain. But if pain is instead identical with a functional property
of which C-fiber firing is one of the realizers, we do have a problem. What
would make C-fiber firing a realizer of pain would be the same thing that
initially made it a promising candidate for being identical with pain, namely
that it plays the causal role of pain. But if it plays the causal role of pain, but is
not identical with pain, how can pain itself play the causal role of pain?
More generally, if it is one or another of the realizers of a mental state that
plays its causal role when the state is instantiated, how can the mental state
itself be said to play the causal role? To suppose that a wince or groan
is caused both by the instantiation of some physical realizer of pain, say
C-fiber firing, and by the instantiation of pain, is to suppose that it is
overdetermined, And it seems implausible, at best, to suppose that every
case of mental causation involves overdetermination. It thus seems that if
the mental state is a multiply realizable functional state, its causal efficacy is
“preempted” by the realizers of that functional state. There is thus the threat
that the mental realm turns out to be causally inefficacious—that it is
epiphenomenal. But how can this be, given that we initially picked out
the physical realizers as states that play the causal role of the mental state they
realize? How can a mental state have a causal role if it doesn’t do any
causing? It may seem that we are led to the conclusion that our common-
sense psychological theory is really an error theory, and that instead of saying
that the physical realizers of a mental state are physical states that play the
causal role of that mental state we should say that they are physical states that
play the causal role that our commonsense theory mistakenly says is played
by that mental state.
One version of the preemption doctrine focuses on the relation between
instances of a functional property and instances of the physical properties
PHYSICAL REALIZATION WITHOUT PREEMPTION 37
that realize it. It seems prima facie plausible that if the property of being in
pain is realized on a given occasion by C-fiber firing, then the instance of
pain that occurs at that time is identical with an instance of C-fiber firing.
This goes with the idea that the causal powers of the instance of pain will be
identical with the causal powers of the instance of C-fiber firing, which is
what Jaegwon Kim’s Causal Inheritance Principle tells us. It will be in virtue
of being an instance of C-fiber firing that it will have these causal powers.
If the property of being in pain is not identical with the property of having
C-fiber firing occurring in one, there seems nothing for it to do—no way it
can contribute to bestowing causal powers on its instances. More generally,
the property instance that is an instance both of a realized property and one
of its physical realizers will have its causal powers in virtue of being an
instance of the physical realizer, and the fact that it is an instance of the
realized property will play no role in explaining why it has these causal
powers.
There have been various responses to this problem. One is eliminativism.
This is the view that there are no mental properties—there are only physical
properties that play the causal roles that we mistakenly take mental proper-
ties to play. This may seem to be forced on us if we hold the plausible view
that any genuine property must have a causal profile. Another response is
reductionism. This is the view that the mental is causally efficacious because
it somehow reduces to the physical. The identity theory we started with is
of course one version of reductionism, which we might call global reduc-
tionism. That theory we have seen reasons for rejecting. But some philoso-
phers, e.g., Jaegwon Kim, have suggested that we can have reductions that
are not global—that are species relative, or structure relative. We cannot
reduce pain, simpliciter, to something physical, but perhaps we can reduce
human pain, or pain of creatures with a certain physical structure, to
something physical. For each species, or for each structural type within a
species, we identify the pain of creatures in that species, or of creatures
having that structure, with a certain physical state. Viewed one way, this is a
version of eliminativism.We abandon pain as a state or event type, deny that
there is any property of being in pain, but we retain human pain, dog pain,
Martian pain, etc., about each of which we put forward an identity theory.
It will be allowed, on this view, that there is a single concept of pain that
applies to all of these states, but it is denied that there is a single property, or
state type, corresponding to this concept.
38 SYDNEY SHOEMAKER
But there is also a response which holds on to the view that mental states
are causally efficacious, without fragmenting them into different species-
relative or structure-relative states, and holds that they are not identical
with, although they are realized in, physical states. This is sometimes called
non-reductive physicalism. I will be presenting a version of this.
III
Such a view must hold that mental properties have causal profiles, and that
their instantiation involves their being realized by physical properties with
associated causal profiles, without there occurring causal overdetermination
of an objectionable sort. To see how this is possible we need to investigate
the notion of realization, and in particular the notion of physical realization.
I take the notion of a mental property being physically realized to be the
notion of there being a constitutive relation between the instantiation
of the property and something physical; the instantiation of the mental
property consists in the existence of its physical realizer, and the existence
of the realizer is metaphysically sufficient for the occurrence of the mental
property instantiation. As the term “realize” suggests, what realizes a
property instance is what makes it real. Of course, it is not only mental
properties and their instantiations that can be said to be realized in this
sense. Functional properties like being a braking system, or being a clock,
can be realized in mechanical systems of various sorts, and colors can
presumably be realized by such things as spectral reflectances. In fact, as
I will claim later on, every property of concrete things that we can refer to
is realized in other properties.
I distinguish two sorts of realization, which turn out to be closely related.
One sort, property realization, is what we have already met with—it is what
we have in the claim that a property like having C-fiber firing going on in
one can realize the property of being in pain but cannot be identical with it.
We can speak of this as a relation between properties, but at bottom it is a
relation between instantiations of properties, or what I will sometimes refer
to as property instances. One property is realized by another property if
instantiations of the one property can be realized by instantiations of the
other property—i.e., if the having of the one can be realized by the having
of the other. In the central case, the instantiation of a property and the
PHYSICAL REALIZATION WITHOUT PREEMPTION 39
instantiation of its realizer will occur in one and the same thing—although
we can define a derivative sense in which the instantiation of a property in
one thing is realized in the instantiation of a different property in a different
thing that is coincident with that thing (as it might be, a person’s body’s
having a certain property realizes the person’s having a certain other prop-
erty, although the person and her body are numerically different things). So,
as we imagined earlier, the instantiation of the property of being in pain
might be realized by the instantiation of the property of having C-fiber
firing occurring in one.
In the other sort of realization,microphysical realization, the instantiation of a
property will be realized in a microphysical state of affairs, one consisting of
micro-entities (atoms, electrons, quarks, or whatever) having certain proper-
ties and being related in certain ways—for short, it consists in micro-entities
being propertied and related in certain ways. Here the realizer of a property
instance will not be another property instance, as it will be in cases of property
realization, but will instead be such a state of affairs. On the physicalist view
that the microphysical facts about the world fix all of the facts about it, every
property instantiation will be microphysically realized. There will of course
be one massive microphysical state of affairs, involving all of the micro-
entities there are, that realizes every property instantiation. But there will be
less global states of affairs that realize some property instantiations and not
others, and it seems reasonable to assume that every property instantiationwill
have a microphysical realizer that is minimal in the sense that it does not
contain any proper part that is a realizer of that property instantiation.
These two sorts of realization are related, because for every type of
microphysical state of affairs whose members can realize a property instanti-
ation there will be a property something has just in case its career embeds a
state of affairs of this sort. One can speak of these as MSE properties, for
microphysical-state-of-affairs-embedding properties. But here I will reserve
the term “MSE property” for cases in which the embedded state of affairs is
maximally determinate. All microphysical states of affairs are realized by
maximally determinate ones, and every property instance will have a max-
imally determinate microphysical realizer. So every case of microphysical
realization, except for cases in which the realized property is itself an MSE
property, is a case of property realization in which the realizer is an MSE
property. Since all properties are microphysically realized, all properties that
are not themselves MSE properties are multiply realized by MSE properties.
40 SYDNEY SHOEMAKER
The usual characterization of property realization is what figured in my
introductory remarks. This says that the realized property is the second-order
property of having some property or other that plays a certain causal role, and
that its realizers are the properties that play that role. This is the formulation
that suggests that the realizers preempt whatever causal role we might think
belongs to the realized property. I favor an account that avoids that sugges-
tion. This is what is sometimes called the subset view, which I share with
Lenny Clapp andMichaelWatkins (Clapp 2001; Watkins 2002). My version
of it says, with certain qualifications that I will not go into, that property P has
property Q as a realizer if the forward-looking causal features of P—its
aptness to contribute when instantiated to the production of certain
effects—are a subset of the forward-looking causal features of Q. So, sticking
to the example of pain, the things we take pain to cause are included among
the things C-fiber firing causes. Initially I saw this view as a competitor to the
higher-order property account of property realization, but I have come to
see that it can be seen instead as a version of that account: let it be that P has
Q as a realizer if P is the property something has just in case it has some
property whose forward-looking causal features include those of P as a
subset. Notice that this view does not say that the realizers play exactly the
causal role that we take the realized property to play. There is a close relation
between the causal profile of the realized property and the causal profiles of
its realizers, which is given by the subset relation, but this relation is not
identity. This will be important later on.
Here I have reverted to the account of property realization I put forward
in my first published discussion of it (Shoemaker 2001), one in which it
consists in a subset relation between the forward-looking causal features of
the realized property and its realizers, and have abandoned the account
I gave in a later paper (2003) and in my book Physical Realization (2007),
which adds as a further requirement that the backward-looking causal
features of the realizer be a subset of the backward-looking causal features
of the realizer properties. This is not because I think that the holding of that
subset relation is not required, but because I think that it necessarily holds
whenever the subset relation between forward-looking causal features
holds. This would not be so if it were possible for two different properties
to share the same forward-looking causal features but differ in their back-
ward-looking causal features. Many years ago Richard Boyd offered a
putative case of this as an objection to my causal theory of properties. In
PHYSICAL REALIZATION WITHOUT PREEMPTION 41
his example substance X is what you get when you combine substances
A and B and substance Y is what you get when you combine the different
substances C and D, and the properties of being composed of X and being
composed of Y are exactly alike in their forward-looking causal features.
These properties are supposed to be different because of their different
compositional histories. I was not convinced by the example, but I saw
that the causal theory of properties could easily be modified so as to allow
for its possibility by having it individuate properties in terms of their
backward-looking causal features as well as their forward-looking causal
features (Shoemaker 1980, postscript). And so I thought that it would do no
harm to allow for the possibility of such cases in the account of realization by
adding the requirement about backward-looking causal features.
But I was wrong to think that doing this would do no harm. This was
shown by Brian McLaughlin (McLaughlin 2007). In a critique of my
account of realization he pointed out that, on the assumption that different
properties can have the same forward-looking causal features, there being an
instantiation of a property which my account says is a realizer of a given
property does not guarantee that there is an instantiation of that property. It
guarantees that there is an instantiation of a property having that property’s
forward-looking causal features, but it does not guarantee that there is an
instantiation of a property having that property’s backward-looking causal
features. The putative realizer’s backward-looking causal features will be a
subset of the backward-looking causal features of the supposedly realized
property, but its instantiation does not guarantee that there is instantiated a
property having precisely that set of backward-looking causal features.
The objection disappears, of course, if we reject the assumption that
properties can differ despite having the same forward-looking causal fea-
tures. And I now do reject this. Returning to Boyd’s example, the most it
could show is that compositional properties can be alike in forward-looking
causal features while differing in backward-looking causal features. But
I don’t think it shows even that. The fact that we get instances of the
property of being made of X by combining substances A and B and get
instances of the property of being made of Y by combining substances C and
D doesn’t show that being made of X is not the same property as being
made of Y—instances of the same property can be caused in different ways.
Of course, if being an X thing consists in being composed of A and B while
being a Y thing consists in being composed of C and D, then being an
42 SYDNEY SHOEMAKER
X thing and being a Y thing will be different properties. But given that
being an X thing and being a Y thing share all their forward-looking causal
features, there is no possibility of decomposing such things into their
supposedly different constituents. And that is a reason for denying that
they have different sets of constituents. I think that if we ran across a case
like this, the best thing for us to say would be that we have a single substance
that can be produced by combining either of two pairs of substances. If
someone insists that it is logically possible, although unverifiable, that in
such a case the substances would differ in composition and would for that
reason be different substances, and so the properties of being composed of
those substances would be different, then I will restrict my account of
property realization to properties that are not of this sort.
IV
I turn now to microphysical realization. Just as properties have causal
profiles, so too do types of microphysical states of affairs. A given type of
microphysical states of affairs will be such that members of that type are apt to
cause, or contribute to causing, microphysical states of affairs of certain other
types, andwill be caused bymicrophysical states of affairs of certain types.My
account of microphysical realization says that a microphysical state of affairs
realizes an instance of a particular property just in case the microphysical state
of affairs belongs to a type whose causal profile matches in a certain way
either the causal profile of the property whose instance it realizes or the causal
profile of a property realizer of that property, and the microphysical state of
affairs is embedded in the career of the subject of the property instance and is
simultaneous with it. Suppose again that pain is realized by C-fiber firing.
This formulation allows an instance of pain to have as a realizer a state of
affairs that is a microphysical realizer of an instance of C-fiber firing, and so is
of a type whose causal profile matches that of the property of C-fiber firing
rather than that of the property of being in pain. But it will also have a
microphysical realizer that is of a state of affairs type, a more abstract one,
whose causal profilematches that of the property of being in pain. Suchmore
abstract states of affairs will be realized by more concrete ones.
What is the relation of “matching” between causal profiles that I speak of
here? I would like to be able to say that it is identity, and I think that there is
PHYSICAL REALIZATION WITHOUT PREEMPTION 43
a good sense in which it is. But the causal features of the property instances
have to do with the causing of other property instances, while those of the
microphysical states of affairs have to do with the causing of other micro-
physical states of affairs. It is true that in contributing to the causing of other
microphysical states of affairs a microphysical state of affairs contributes to
the causing of the property instances realized by those states of affairs; and
that is a reason for saying that the states of affairs share the causal features
of the property instances they realize. But that invokes the notion of
microphysical realization, which is what I am trying to explain. So to say
this in explaining the notion of microphysical realization would involve a
kind of circularity. In Physical Realization I said that the causal profile of the
state of affairs type is isomorphic with the corresponding causal profile of the
instantiated property. But if that means only that there is a structural
similarity between the causal profiles involving a one–one correspondence
which pairs each causal feature in the one with a causal feature in the other,
such matching does not by itself make the state of affairs a realizer of the
property instance. It must further be the case that the property is instantiated
if there occurs at the same time and in the same career a state of affairs of the
corresponding type, and that this is true as a matter of necessity. And it must
be the case, again as a matter of necessity, that when an instance of the
property causes or contributes to causing an instance of another property,
the corresponding state of affairs causes or contributes to causing a state of
affairs of a type that is paired with that other property in the one-one
correspondence—likewise, when an instance of the property is caused by
an instance of another property, the corresponding state of affairs is caused
by a state of affairs of a type that is paired with that other property in the
one–one correspondence. A further requirement is that if we trace the
causal histories of the property instance and the microphysical state of affair,
both into their pasts and into their futures, they will converge—the very
same things will be involved in causing causal ancestors of both the property
instance and its microphysical realizer, and the very same things will be
caused by causal descendants of both. What all of this is intended to
guarantee is that the occurrence of the microphysical state of affairs consti-
tutes the instantiation of the realized property.
The microphysical states of affairs will be partly concrete, a matter of
particular micro-entities being propertied and related in certain ways, but
will consist in part in the truth of positive and negative existential propositions
44 SYDNEY SHOEMAKER
saying that there are, or are not, micro-entities related in certain ways to the
constituents of the concrete part of the state of affairs. The account involves
these complications because of the fact that what realizes a property instance
must guarantee the existence of the thing having the property instance, and
must guarantee the instantiation of whatever other properties a thing of that
sort must have in order to exist. For I take realizers to be sufficient for what
they realize—otherwise they would not be what make the realized things
real—and what is sufficient for the instantiation of a property must be
sufficient for the existence of something having it, and for that thing having
whatever properties it must have in order to exist. I deal with this by
suggesting a way of factoring the state of affairs that is a microphysical realizer
of a property instance. This will consist partly of what I call its core, which is
made up of states of affairs that contribute directly to implementing the causal
profile of the realized property. But it also consists in part of existential states
of affairs that exist in virtue of the concrete states of affairs involved in the
realizations of other properties of the thing that has the property. So, for
example, a microphysical realizer of an instance of the property of having a
certain height will have a core consisting of states of affairs that contribute
directly to instantiating the causal profile associated with that height, but will
also contain existential states of affairs that guarantee that it has some width or
other, some mass or other, and so on. And a microphysical realizer of an
instance of the mental property of having a certain belief will have a core
consisting of states of affairs that contribute directly to implementing the
causal profile of that belief, but will also contain existential states of affairs
that guarantee that the subject has whatever other mental properties a
subject of that belief must have. This is only a rough sketch of the account,
the details of which are too complex for me to go into here. (For a fuller
account see my 2007.)
V
As is apparent from this brief summary, the accounts of both sorts of
realization take it that realized properties, e.g., functional properties and
determinables of all sorts, have causal profiles, and so can be causally
efficacious. Certainly it should be the default view that mental properties,
and other physically realized properties, are causally efficacious. But how do
PHYSICAL REALIZATION WITHOUT PREEMPTION 45
I deal with the threat, mentioned in my opening remarks, that the causal
efficacy of realized properties is preempted by their realizers—and in par-
ticular, that the mental efficacy of mental properties is preempted by their
physical realizers, making mental properties causally inert?
I do this in part by focusing on JaegwonKim’s formulation of what he calls
the Causal Inheritance Principle.When Imentioned this earlier I put it as the
principle that the causal powers of an instance of a higher-order property are
identical with those of the instance of the lower-order property that realizes
it. This goes with the view that the instance of the higher-order property and
that of its lower-order realizer are one and the same—so, e.g., the instance of
pain might be identical with the instance of C-fiber firing. That is what Kim
thinks is true in the central case, and that is what seems to lead to the view
that higher-order properties have no causal role to play—for presumably the
instance of the physical property that is the lower-order realizer has its causal
powers in virtue of that physical property’s having the causal profile it has,
and that leaves the higher-order property, e.g., the mental property, with no
contribution to make to the causal powers of what is supposed to be an
instance of both it and its realizer. But the formulation of the Causal Inherit-
ance Principle that Kim gives in several places says that the causal powers of a
higher-order property are “identical with (or a subset of ) the causal powers of
[its] realizer” (Kim 1998, 116, my emphasis). His reason for including (in
parentheses) the phrase “or a subset of ” is to allow for cases in which the
realizer is a conjunctive property having as one of its conjuncts a property
that is itself a realizer of the higher-order property. Thus the conjunctive
property has C-fibers firing and is six feet tall will count as a realizer of pain if
having C-fibers firing is a realizer of it, and in this case the causal powers of an
instance of this conjunctive property will have as a proper subset those of an
instance of having C-fibers firing, where the latter are the causal powers of
the instance of pain. But the subset account of property realization suggests
that in general, and not just in the special case of such conjunctive realizers,
the causal powers of the realized property are a proper subset of those of the
realizer instance. This implies that the instance of the realized property is not
identical to the instance of the realizer, for the instances could be identical
only if their causal powers were the same. And this undermines the case for
the view that the higher-order property is causally idle.
But it is compatible with the claim that the instance of the higher-order
property and that of its realizer are not identical that the former is part of the
46 SYDNEY SHOEMAKER
latter. And that seems the right conclusion to draw from the fact that the
causal powers of the former are a proper subset of those of the latter. And
then it is open to us to say that while it is true that the instance of the realizer
property causes the various effects we attribute to the realized property, it
does so because it includes as a part the instance of the realized property. For
example, the instance of C-fiber firing causes, or contributes to causing, the
moaning, groaning, and calls to the doctor, but it does so because it includes
the instance of pain. It includes the instance of pain because its instantiation
guarantees, constitutively, the instantiation of a property having the causal
profile of pain—this because of the subset relations between the causal
profiles of the two properties. The part of the causal profile of C-fiber
firing that is exercised here is precisely the part it shares with the causal
profile of pain in virtue of having the forward-looking causal features of
pain as a subset. So while it is true that the instance of C-fiber firing “does
the causal work,” it does not do so in a way that leaves the instance of pain
with no work to do; on the contrary, it does the causal work because it
includes as a part the instance of pain.
Now let me turn to microphysical realization. As I mentioned earlier, a
property instance can have more than one microphysical realizer. If the
property instance is property-realized by another property instance, then it
will be realized by the microphysical state of affairs that realizes that property
instance, but will also be realized by a microphysical state of affairs that is
specific to it—one whose causal profile matches that of the property of
which it is a realizer. So, to stick with the rather tired example, the instance
of pain will be realized both by the microphysical state of affairs that realizes
the instance of C-fiber firing that property-realizes that pain instance, and
by a more abstract microphysical state of affairs that is peculiar to it.
A property that realizes another property may itself be realized by still
another property. So whenever there is property realization there will be
a hierarchy of property instances, having two or more members, each
member of which other than the one at the bottom is property realized
by those below it in the hierarchy. And there will be a corresponding
hierarchy of microphysical states of affairs, each member of which realizes
the corresponding property instance and all of the property instances above
it in that hierarchy. And each of these states of affairs can be said to realize
the states of affairs above it in the hierarchy of microphysical states of
affairs—there will be the same subset relations between the causal profiles
PHYSICAL REALIZATION WITHOUT PREEMPTION 47
of states of affairs at different levels in this hierarchy as there are between the
causal profiles of the properties instanced in the hierarchy of property
instances. As we go up the hierarchy of states of affair the states of affairs
will become more abstract. At the bottom of such a hierarchy will be a state
of affairs that is maximally determinate. The MSE properties referred to
earlier will be properties things have in virtue of their careers embedding
maximally determinate states of affairs of this sort.
VI
A few words of clarification about MSE properties. While these can be
regarded as a kind of microstructural, or micro-based, properties, not all of
them fit one natural characterization of microstructural properties. That
characterization says that a microstructural property is one that something
has just in case it contains micro-entities of certain sorts that are related to
one another in certain ways. So conceived, microstructural properties are
what I call thin properties. They are properties that can be shared by things
of different kinds. If, as I believe, there can be coincident entities—e.g., a
person and that person’s body—which are numerically different despite
being composed of the very same matter, then such coincident entities
share all of their thin properties and so share all of the same microstructural
properties of this sort. By contrast, thick properties are not shared by coinci-
dent entities, or by entities that are of different kinds and have different
persistence conditions. I take it that mental properties are thick properties.
Although I and my body are coincident entities, I have mental properties
and my body doesn’t. There aren’t two things here thinking my thoughts,
and feeling my sensations. Whatever properties are property realizers of a
thick property are themselves thick properties. And MSE properties that are
realizers of thick properties are themselves thick properties—these are the
ones that don’t fit the natural characterization of microstructural properties.
I said that an MSE property is one something has just in case its career
embeds a maximally determinate microphysical state of affairs that is a micro-
physical realizer of a property instance. But there are different things that can
be meant by ‘embed’ here. In one sense, for the career of a thing to embed a
state of affairs at a time it is sufficient that the state of affairs occurs at that time
in the career of the thing—e.g., in the case of microphysical states of affairs, it is
48 SYDNEY SHOEMAKER
sufficient that it is a state of affairs consisting in some of the micro-constituents
of the thing being propertied and related in certain ways. Call this weak
embedding. It is in this sense that microphysical realizers of thin property
instances are embedded in the careers of their possessors. But in the case of
micro-realization of thick properties we need a stronger sort of embedding,
which I will call strong embedding. We can think of the career of a thing of a
certain sort as consisting in a series of sets of property instances that are so
related, causally, that the persistence conditions for things of that sort, together
with the causal profiles of the properties instantiated, make that series the
career of a single thing of that sort. Assuming physicalism, this will be realized
in a series of microphysical states of affairs that realize the property instances.
Let’s say that a microphysical state of affairs that is a member of such a series at a
time is a momentary stage of that thing at that time. This stage will bemade up
of smaller states of affairs that are realizers of property instances occurring in the
thing at that time. What makes the stage made up of these states of affairs the
stage of a thing of a particular sort is the fact that the causal profiles of the states
of affairs that make it up are such that their occurrence contributes to the
implementation of the persistence conditions for things of the sort in question.
This will involve some of the states of affairs being realizers of thick properties
that can be instantiated only in careers of things of that sort. A microphysical
state of affairs is strongly embedded in the career of a thing of a given sort at a
time if it is such a part of themicrophysical state of affairs that is themomentary
stage of the thing at that time, i.e., if it has a causal profile that makes the
requisite contribution to the implementation of the thing’s persistence condi-
tions, qua thing of that sort. In the case of pairs of coincident entities, like me
and my body, there will be two different careers involving the same micro-
entities, and some of the microphysical states of affairs involving these entities
will be strongly embedded in one of these careers and some will be strongly
embedded in the other. Which is to say that one of these coincident entities
will have one set of thick MSE properties, and the other will have a different
set of such properties. A thick MSE property will be a property a thing has in
virtue of its career strongly embedding a microphysical state of affairs. Of
course, the things will share a number of thin properties, and the thin proper-
ties will also be realized by thinMSE properties. And here theMSE properties
will be realized by states of affairs that are weakly embedded in the careers of
their possessors. Some of these states of affairs will also be strongly embedded in
the careers of one or another of the coincident entities.
PHYSICAL REALIZATION WITHOUT PREEMPTION 49
Consider a microphysical state of affairs that realizes an instance in me of a
mental property. This state of affairs occurs in my career, and it also occurs
in the career of my body. But while it is strongly embedded in my career, in
the sense just explained, it is not strongly embedded in my body’s career,
although it is weakly embedded in it. And so my body, unlike me, does not
have the MSE property constituted by the strong embedding of that state of
affairs. The occurrence of that state of affairs in my body’s career does
guarantee that something has the mental property in question. But that
something is me, not my body. What its occurrence in my body entails is
that there is something coincident with my body that has the MSE property
of strongly embedding it and so has the mental property it realizes.
Every property instance that is not itself an instance of an MSE property
will be property realized by an instance of an MSE property. This will be
true of instances of properties that would not ordinarily be thought of as
higher-order properties—properties like having a certain shape, or a certain
mass. Such properties will have an infinite number of MSE properties as
possible realizers. The MSE properties will be epistemically inaccessible
to us, and will not be properties that figure in ordinary thought and
discourse—or, for that matter, in scientific thought and discourse. We can
easily know that two things differ in their MSE properties, but it will be
beyond our ability to know that two things share an MSE property, or that
something has remained unchanged with respect to MSE properties. And
sharing of MSE properties, or retention of them over time, will be exceed-
ingly rare, for the slightest difference in the location of a single electron
or quark will give us a difference in MSE properties. Such properties could
not figure in the taxonomy of any science, and could not figure in any laws
that we can formulate. Only an omniscient deity could have any knowledge
of them.
VII
Sometimes it is suggested that only first-order properties can be causally
efficacious and figure in causal laws. This goes with the view that the causal
efficacy ascribed to functional properties and other higher-order properties
is preempted by their lower order realizers. Those who hold this usually
assume that the first-order properties that “do the causal work” are ones that
50 SYDNEY SHOEMAKER
we can have knowledge of, and that the laws governing them are ones we
learn about from fundamental science. But if only MSE properties count as
first-order properties, this cannot be right. If the only causal laws are laws
about these, causal laws are unknowable by the likes of us. This is highly
implausible.
Consider the property of having a mass of one gram. One might have
thought that this is a paradigm first-order property. But it will have as
property realizers a vast—I think infinite—number of MSE properties.
These will differ from one another in the sorts of micro-entities that make
up the things having this mass, and in the ways these micro-entities are
configured. What they have in common is that the instantiation of each of
them bestows the causal powers bestowed by the property of having a mass
of one gram. When something has the mass of one gram in virtue of having
one of these MSE properties, it would be absurd to suggest that the effects
we attribute to its having this mass are really due instead to its having that
MSE property. For one thing, this would fly in the face of a plausible
proportionality constraint on the relation between causes and effects, for
the vast majority of the causal features of the MSE property will be irrele-
vant to the effects associated with having a mass of one gram (see Yablo
1992). It is by abstracting away from these causal features that we get to the
causal profile of the mass property. To be sure, it is in virtue of having that
MSE property that the thing has the mass of one gram, and so has the causal
powers that go with the possession of that mass. But it is the having of that
mass that gives it the relevant powers. In line with what I suggested earlier,
we can say that the instance of the MSE property contains as a part the
instance of the mass property, and that it is because of this that it bestows the
relevant powers.
What I have said about the property of having a mass of one gram, versus
the various MSE realizers of it, is similar to what Hilary Putnam said in his
famous discussion of why a cubical peg won’t go through one hole and will
go through another (Putnam 1975). The appropriate explanation of this, the
one having the generality we want from such an explanation, will be in
terms of the rigidity of the peg and board and the dimensions of the peg and
holes as described in macroscopic terms. The explanation that would be
perverse even if we were in a position to give it—Putnam says it is a “terrible
explanation” if it counts as an explanation at all—is one in terms of the
distribution of the micro-entities making up the objects. And that would
PHYSICAL REALIZATION WITHOUT PREEMPTION 51
amount to an explanation in terms of the MSE properties of the peg and of
the board that realize the shapes of the peg and the board and their rigidity.
We don’t, of course, think of properties like having a certain mass as
being property-realized by other properties that are not MSE properties—
these are usually thought of as ground level, first-order properties. On the
other hand, we do tend to think that the property of being in pain
has property realizers, perhaps C-fiber firing and the like, that are not
MSE properties. Let me observe in passing that it is not a consequence
of physicalism that this is so. Someone could be both a physicalist and a
functionalist while thinking that the property of being in pain has as good a
right as the property of having a mass of one gram to be regarded as a first-
order property, and that its only realizers are MSE properties. Whether it
has other realizers is an empirical question. But if it does, those realizers must
be realized, directly or indirectly, in MSE properties, and so also must the
property of being in pain. And what I just said about the property of having
a mass of one gram applies as well to the property of being in pain—and to
any other psychological property you care to mention. I said earlier that it is
only because the C-fiber firing instance contains the pain instance as a part
that it has the relevant effects. We can think of this as a consequence of the
relation between the microphysical realizers of these instances—it is because
the microphysical realizer of the C-fiber firing instance contains as a part the
microphysical realizer of the pain instance that it has the relevant effects.
And it is only because the maximally determinate microphysical state of
affairs that realizes an MSE property instance contains a state of affairs that is
a pain instance realizer (perhaps by way of containing a C-fiber firing
instance realizer) that it has the relevant effects.
How are we to understand this talk of states of affairs containing other states
of affairs as parts? As a first pass, state of affairs P contains state of affairs Q ifQ’s
existence is entailed by P’s existence. But the entailment can be seen as
holding in virtue of a relationship between the causal profiles of the states of
affairs that parallels the relationship between properties and their property
realizers. Where state of affairs P contains state of affairs Q, the forward-
looking causal features of P will contain as a subset the forward-looking causal
features of Q, and the backward-looking causal features of Q will contain as a
subset the backward-looking causal features of P. And this relationship, in
turn, will hold in part because the ways micro-entities are propertied and
related in Q are realized by the ways micro-entities are propertied and related
52 SYDNEY SHOEMAKER
in P. The contained states of affairs will be abstract relative to the containing
states of affairs, and can be said to be realized by them.
VIII
I said earlier that every case of microphysical realization, except for cases in
which the realized property is itself an MSE property, is a case of property
realization in which the realizer is an MSE property. This seems to be true
no matter what the world is like, and whether or not physicalism is true.
Will it also be true that every case of property realization will also be a case
of microphysical realization—i.e., that whenever a property instance is
realized by an instance of a different property, it is realized by a micro-
physical state of affairs? Assuming a version of physicalism on which the
microphysical facts fix all of the facts, this will be true. But there are
conceivable worlds, dualist ones, in which it is not true—in these many
property instances, in particular mental ones, do not have microphysical
realizers, but it may still be true that many property instances are realized by
other property instances. For example, the property of being a clock will be
realized by various mechanical and electronic properties. And there are
perhaps worlds in which the instantiation of physical properties in macro-
scopic entities is not realized by microphysical states of affairs. If so, in such
worlds there can be physical property realization unaccompanied by micro-
physical realization. My physicalist assumption is that the actual world is one
in which the microphysical facts fix all of the facts, and so one in which all
realization involves microphysical realization.
I began my discussion of realization by mentioning the multiple realiza-
tion argument against type psychophysical identity theory—the view that
mental properties are identical with physical properties. The objection was
that where a physical property might seem to be identical with a mental
property, because it plays its causal role, it is always possible for there to be
other properties that play the causal role of the mental property equally well
and so are equally good candidates for being identical with it, and that such
properties are at best realizers of the mental property rather than being
identical with it. But associated with any mental property there will be one
physical property that seems an especially good candidate for being identical
with it. If I had not restricted the application of the term “MSE property” to
PHYSICAL REALIZATION WITHOUT PREEMPTION 53
cases in which the embedded states of affairs are maximally determinate,
I could call these MSE properties. So let me call them MSE* properties.
Every property instance has a microphysical realizer of a type whose causal
profile exactly matches that of the property of which it is an instance, and
the thing having that property instance thereby has the property, an MSE*
property, something has just in case it embeds a microphysical state of affairs
of that type: (MSE properties are maximally determinate realizers of MSE*
properties). These MSE* properties should count as physical properties,
given that their instantiation just consists in microphysical entities being
propertied and related in certain ways. And they look to be excellent
candidates for being identical with the properties whose instances are realized
by the embedded microphysical states of affairs. So, for example, every
instance of the property of being in pain is directly realized by a microphysical
state of affairs of a certain type, and someone will have the property of being
in pain just in case he has the MSE* property something has in virtue of its
career embedding a microphysical state of affairs of that type.
But what I just said will be true only in worlds—of which I assume the
actual worlds is one—in which all of the facts are fixed by the microphysical
facts. In such worlds every property, including every mental property, will
be coextensive with an MSE* property. But if there are possible worlds in
which dualism is true, or worlds in which the microphysical facts do not fix
all of the facts, and if the properties instantiated in these worlds include some
of the properties, including the mental properties, instantiated in the actual
world, then these properties will not be necessarily coextensive with MSE*
properties and so cannot be identical with them.
Whether there are such possible worlds is a difficult question I cannot go
into here. I do not think that the conceivability or imaginability of such
worlds, or the fact that they are in some sense epistemologically possible,
shows that they are possible in the relevant sense—that they are metaphysic-
ally possible. And it is worth asking what the implications would be for the
philosophy of mind if it should turn out that they are not possible, and that
all properties of persisting things, including mental properties, are MSE*
properties, and so are physical properties.
Would this undermine the case for non-reductive physicalism? It would do
so, of course, if it is made definitive of non-reductive physicalism that it denies
any sort of type identity between mental properties and physical ones. But
I don’t think this should be made definitive of non-reductive physicalism.
54 SYDNEY SHOEMAKER
Notice that MSE* properties, though physical, are not for the most part
properties that figure under their physical descriptions in the laws of physics,
or are likely ever to so figure. And the ones that would be identical with mental
properties, on the supposition we are making, are not properties anyone has, or
ever will have, the ability to describe or define in canonical physical terms.
Accepting such an identification would not in the least support the claim that
psychology is reducible to physics, and would not put into question the
autonomy of psychological explanation, whether it be of the scientific sort or
the commonsense sort. Non-reductive physicalists should be happy to accept
psychophysical property identities involving MSE* properties.
Suppose, however, that dualist worlds aremetaphysically possible, and that
the coextensiveness of mental properties to MSE* properties is limited to a
subclass of possible worlds, those in which themicrophysical facts fix all of the
facts. What then is the relation between mental properties and coextensive
MSE* properties, given that it cannot be identity? It seems that it should be
realization or constitution of some sort. It could be property realization of the
sort defined earlier—the forward-looking causal features of the mental prop-
erty are a subset of those of the corresponding MSE* property and the
backward-looking causal features of the MSE* property are a subset of
those of the mental property. There is also a different subset relation between
them. The microphysical states of affairs that are realizers of instances of an
MSE* property would be a proper subset of the states of affairs that realize the
associated mental property. Just what the realizing states of affairs for mental
properties would be in a dualist world is hard to say—maybe, following
Putnam’s tongue-in-cheek suggestion, they would be ectoplasmic states of
affairs (Putnam 1967). Notice that we have this subset relation also in other
cases of property realization; if pain has C-fiber firing as a property realizer,
then given all of the states of affairs that obtain in all possible worlds, the states
of affairs that aremicrophysical realizers of instances of C-fiber stimulation are
a proper subset of the states of affairs that are realizers of pain. This gives us an
alternative account of property realization, one defined in terms of state of
affairs realization. Property P is a realizer of property Q if and only if the
possible states of affairs that are realizers of instances of P are a proper subset of
the possible states of affairs that are realizers of instances of Q—where a
possible state of affairs is a state of affairs obtaining in some possible world.
It seems to me that whether or not there are dualist worlds, mental
properties have as good a claim to be physical properties as automotive
PHYSICAL REALIZATION WITHOUT PREEMPTION 55
properties, architectural properties, computer properties, and botanical
properties. It is true of all these properties that all of their actual world
property realizers are physical, and that all of their actual world instances are
realized by microphysical states of affairs. If there are dualist worlds, mental
properties will in these have non-physical realizers—but so too, I think,
might automotive properties and botanical properties.
IX
To return from this digression, and conclude my paper, let me restate my
central claim. It is that a proper understanding of what it is for property
instances to be physically realized removes any threat that the causal efficacy
we ascribe to mental properties is preempted by their physical realizers. The
subset view of property realization suggests that the proper understanding of
“causal inheritance” is that the causal powers of an instance of a realized
property are always a proper subset of the causal powers of the instance of its
property realizer, and this suggests the view that the realized property
instance is included in the realizer property instance as a part. And the
same conclusion is suggested by a consideration of microphysical realization;
if microphysical realizers of higher-order properties are abstract states of
affairs that are realized by more concrete states of affairs that realize lower-
level properties that are property realizers of them, it seems appropriate to say
that the former states of affairs are included in the latter, and likewise that the
property instances realized by the former are included in those realized by
the latter. And then we can say that the lower-level property instances do the
relevant causal work because they have the higher-level property instances as
parts. In the case that primarily concerns us, physical property realizers of
mental property instances cause the effects we attribute to those mental
properties because they contain those mental property instances as parts.
References
Clapp, L. (2001). ‘Disjunctive Properties; Multiple Realizations’. The Journal of
Philosophy. 98(3): 111–36.
Kim, J. (1998). Mind in a Physical World. Cambridge, MA: MIT Press.
56 SYDNEY SHOEMAKER
Lewis, D. (1980). ‘Mad Pain and Martian Pain’, in N. Block (ed.), Readings in
Philosophy of Psychology, vol. 1. Cambridge, MA: Harvard University Press: 216–32.
McLaughlin, B. P. (2007). ‘Mental Causation and Shoemaker-Realization’.
Erkenntnis, 67: 141–72.
Putnam, H. (1967). ‘The Nature of Mental States’. First published as ‘Psychological
Predicates’ in W. H. Capitan and D. D. Merrill (eds.), Art, Mind and Religion.
Pittsburgh, PA: University of Pittsburgh Press. Repr. in Putnam,Mind, Language
and Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University
Press: 429–40.
—— (1975). ‘Philosophy and Our Mental Life’. In Putnam, Mind, Language and
Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University Press:
291–303.
Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and
Cause. Dordrecht: D. Reidel: 109–35.
—— (2001). ‘Realization and Mental Causation’. In C. Gillett and B. Loewer
(eds.), Physicalism and Its Discontents. Cambridge: Cambridge University Press:
74–98.
—— (2003). ‘Realization, Micro-realization and Coincidence’, Philosophy and
Phenomenological Research, 67(1): 1–23.
—— (2007). Physical Realization. Oxford: Oxford University Press.
Watkins, M. (2002). Rediscovering Color. Dordrecht: Kluwer.
Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.
PHYSICAL REALIZATION WITHOUT PREEMPTION 57
3
Mental Causation in thePhysical World
PETER MENZIES
1. Introduction
Mental causation is the phenomenon in which a mental state causes another
mental state or causes some behaviour. As Jaegwon Kim (1998, 2005)
reminds us, not much of commonsense psychology would make any sense
if mental causation were not real. Our conception of ourselves as conscious,
intentional agents capable of perception, memory, and reasoning is tied up
with the assumption of the reality of causal processes involving cognitive
phenomena.
However, philosophical questions about mental causation revolve
around, not so much whether it is important, but rather how it is possible
in the first place in the light of certain metaphysical assumptions and
principles. The classic instance of the philosophical problem of mental
causation is Descartes’ discussion of how mind–body interaction is possible
in his dualist metaphysics, according to which mind and body are two
radically different kinds of substances. Philosophers from Pierre Gassendi
onwards have pointed out such causal interaction is impossible within
Descartes’ metaphysics which accords primacy to causation by contact
forces and in which minds do not have any spatial location or extension.
The general consensus among philosophers is that Descartes was not suc-
cessful in solving this problem.
In contrast to Descartes’ problematic dualism, the monist metaphysics of
physicalism is thought to be more congenial to explaining how mental
causation is possible. For example, on the identity theory of the mind,
mental states are just neural states of the brain, so that mental causation is a
simple instance of neurophysiological causation. Unfortunately, this simple
solution to the problem doesn’t work if the identity theory isn’t tenable.
Many philosophers of mind now believe this to be the case in view of the
many serious objections faced by the theory, the most serious of which turns
on the multiple realizability of mental states (Fodor 1974; Putnam 1975).
However, physicalism as a more general framework has not lost its appeal
despite the waning popularity of the identity theory. Physicalists continue to
believe that the world and its contents are nothing over and above the
structures described by fundamental physics. The idea that physicalists try to
capture is that all the objects in the worlds are constituted out of physical
particles, and that all the properties and relations that these objects enjoy
depend, in some constitutive sense, on the properties and relations mapped
out in fundamental physics.
While there are still unresolved problems about the precise formulation
of this metaphysical view, most physicalists accept a formulation of physic-
alism in terms of a supervenience thesis. The following formulation by
Frank Jackson (1998) has become reasonably standard:
Physicalism about the mental: Any world that is a minimal physical duplicate of the
actual world is also a mental duplicate of it.
A minimal physical duplicate of the actual world is a world that contains the
same physical objects, physical properties and relations, and physical laws as
the actual world; and nothing else. It is important to restrict the set of worlds
used in the supervenience thesis to the set of minimal physical duplicate
worlds. For the physicalist should not accept that anyworld that is a physical
duplicate of the actual world is a mental duplicate. For there are worlds that
duplicate the physical entities of the actual world but include, in addition, a
number of non-physical entities such as Cartesian minds. Such worlds are
not relevant to the formulation of physicalism, which is supposed to be at
best a contingent truth about the actual world, not a necessary truth about
all worlds. Physicalists need not deny that there are such worlds with
Cartesian minds in the remote regions of logical space; they need insist
only that the actual world is not such a world.
The supervenience thesis above expresses a minimal commitment of
physicalism. It is a thesis endorsed by both reductive physicalists who
accept the identity of mental with physical properties and non-reductive
MENTAL CAUSATION IN THE PHYSICAL WORLD 59
physicalists who do not. In recent years, non-reductive physicalists have
tended to outnumber reductive physicalists, mostly because considerations
about multiple realizability have been regarded as biting into the plausibility
of the identity theory. Non-reductive physicalists express the hope that they
can explain how mental causation is possible within the austere metaphys-
ical framework of physicalism while avoiding the reductionism of the
identity theory. Indeed, they hope that it is possible to vindicate not only
the reality of mental causation, but also its independence and autonomy
from physical causation.
This paper will divide into two parts. In the first part I shall argue that
physicalism, whether of the reductive or the non-reductive variety, faces a
challenge just as serious as that faced by Cartesian dualism. I shall outline an
argument that proceeds from physicalist premises to the conclusion that
mental states are causally inert or epiphenomenal. The argument is related
to the well-known exclusion argument advanced by Jaegwon Kim (1998,
2005) that purports to show that non-reductive physicalism is an unstable
position that should be replaced by reductive physicalism. Like Kim’s
argument, the argument I advance appeals to an exclusion principle about
causation to the effect that a state that is causally sufficient for some effect
excludes any mental state that supervenes on it from being causally effica-
cious with respect to the effect. However, the exclusion principle I appeal
to is weaker than Kim’s principle. Also my argument is directed at all
versions of physicalism, reductive as well as non-reductive. Ultimately,
I shall conclude that the argument I describe is not sound: physicalism can
escape the conclusion about the causal inertness of the mental but only by
abandoning the exclusion principle about causation. Many physicalists will
find this conclusion hard to swallow, as the exclusion principle appears to be
very intuitive to them, with Kim (2005), for example, claiming that it is an
analytic truth. We shall see that the principle, when appropriately formu-
lated, is not a general truth of any kind, as there are straightforward counter-
examples to it.
The second half of the paper takes up the issue whether there is a better
formulation of the exclusion principle. Philosophical discussions of exclu-
sion principles seldom proceed in terms of a well-ground theory of
causation. I plan to remedy this defect by motivating a conception of
causation as difference-making and then using it to formulate an alterna-
tive, more satisfactory version of the exclusion principle not vulnerable to
60 PETER MENZIES
the counterexamples to the earlier version. Much of this discussion reports
on work done in collaboration with Christian List (List and Menzies 2009;
Menzies and List 2010). We have argued that the new principle is at best a
contingent truth about causal systems and have identified the conditions
that a causal system must satisfy in order for the principle to be true. It turns
out that the principle can apply in two non-trivial ways to a causal system.
The first—the case of upwards exclusion—is familiar from the argument
against physicalism: here a lower-level cause excludes a higher-level cause.
But the second—the case of downwards exclusion—is often overlooked:
here a higher-level cause excludes a lower-level one. These cases of
downwards exclusion are particularly interesting, as they support the causal
autonomy of higher-level properties. This is a surprising turn of events: far
from supporting reductionist thinking, the exclusion principle actually
turns out to be the linchpin of an argument that vindicates the causal
autonomy of mental properties. In the last section of the paper I turn to
consider the implications of this result for recently popular compatibilist
forms of non-reductive physicalism. Compatibilists attempt to answer the
exclusion argument against mental causation by claiming that mental and
physical states work in tandem to cause to behaviour in a form of non-
standard overdetermination. I concentrate on Sydney Shoemaker’s (2007)
version of compatibilism, arguing that the downwards exclusion result
demonstrates the untenability of his view that mental causation involves a
kind of non-standard overdetermination in which one kind of cause rides
piggyback on another.
2. A New Exclusion Argument
It may be best to illustrate the new exclusion argument against non-
reductive physicalism by way of an example, first introduced into philo-
sophical discussions by James Woodward (2008). The real-life example
concerns the research of Richard Andersen and colleagues at Caltech on
the neural encoding of intentions to act (Mussallam et al. 2004).1 Andersen
1 The ultimate goal of Andersen’s work is to develop neural prosthetics for paralysed subjects that
decode their intentions to reach for specific targets from neural signals and use these to control external
devices.
MENTAL CAUSATION IN THE PHYSICAL WORLD 61
and his colleagues made recordings from individual neurons in the parietal
reach region (PRR) of the motor cortex of monkeys. This region is known
to encode intentions or higher-order plans to reach for specific targets, say a
piece of fruit in a particular location. Andersen developed a programme that
correlated the monkey’s intentions to reach for specific goals, as revealed in
their movements, with certain patterns in the recorded firings of neurons in
their PRR. Using neural recordings, the programme was able to predict
with 67.5 per cent accuracy the reaching behaviour of the monkeys towards
eight targets.
The neural signals that encode themonkeys’ intentions to reach for certain
targets were recorded as averages of the firing rates (spikes per second) of
individual neurons. But clearly the same aggregate firing rate in a group of
neurons is consistent with a lot of variation in the behaviour of individual
neurons. For example, very different temporal sequences of neural firings
can give rise to the same firing rate. So an intention to reach for a certain
target can be realized in many different ways at the level of individual
neurons. Nonetheless, each intention is associated with a distinctive aggre-
gate pattern of firing rates. It is useful to introduce some simple notation.
Suppose themonkeys can have intentions to reach for certain targets, I1, I2, I3etc., and can perform the corresponding actionsA1,A2,A3, etc. Suppose that
intention Ii can be realized at the level of individual neurons in different
token patterns of neural firings, Ni1, Ni2, Ni3, etc. Suppose that on some
specific occasion a monkey forms the intention Ii to reach for a particular
object and performs the corresponding action Ai. Suppose further thatNi1 is
the particular token pattern of neural firing that realizes or encodes the
intention Ii on this occasion. What was the cause of the monkey’s action
Ai? Was it the intention Ii, or its particular neural realization Ni1? Let’s
assume that both the intention and its neural realization are causally sufficient
for the action. It is very tempting for physicalists to answer that it was the
highly specific, neural state Ni1 that caused the monkey to perform the
action. But if the neural state did all the causal work, it would appear that
the intention, which we are assuming is numerically distinct from its highly
specific neural realizer, has no causal role and so is epiphenomenal.
Let’s look at the argument in more detail. The argument relies on a
number of assumptions or principles, some of which were implicit in this
informal presentation of the argument. Let’s make these assumptions and
arguments explicit.
62 PETER MENZIES
(1) Supervenience and realization: mental properties supervene on distinct physical
properties; and so any given instance of a mental property will have an
instance of a distinct physical property as its supervenience base (alternatively,
any mental state will be realized by a distinct physical state).
This supervenience of mental properties on physical properties is a simple
consequence of the contingent supervenience thesis that we are taking to be
the minimal commitment of physicalism. Since we are discussing non-
reductive physicalism, the supervenience thesis is spelled out in terms of
mental properties supervening on distinct physical properties.
(2) Causation entails causal sufficiency: if the state S1 causes another state S2, then S1causally sufficient for S2.
What is meant by causal sufficiency here? I shall understand this as follows: a
state S1 is causally sufficient for S2 (in the actual world) if and only if all the
worlds among the set of minimal physical duplicates of the actual world in
which S1 holds are worlds in which S2 holds. Given that the minimal physical
duplicates of the actual world hold fixed the fundamental physical laws, this
means that these laws entail that an S1 state will lawfully evolve into an S2 state.
Of course, this is a questionable assumption that commits us to a deterministic
conception of causation.While I concede that causationmay involve probabil-
istic rather than deterministic processes, I make this assumptionmostly because
it simplifies our discussionwithout any significant loss of generality. It would be
misguided, I think, to imagine that the assumption of determinism is the source
of the difficulties affecting mental causation, which can be solved by repudiat-
ing this assumption. If there is a solution to the mental causation problem, it is
one that surely holds good even if we assume that causation is deterministic.
(3) The transmission of causal sufficiency across realization: if a mental state M is
causally sufficient for a behavioural state B and M is realized by a distinct
physical state P, then the physical state is causally sufficient for the behav-
ioural state B.
This principle should be no more controversial than the definition of causal
sufficiency given above, since it follows as an analytic consequence of this
definition. To see that the principlemust be true given the definition of causal
sufficiency, suppose, for reductio, that it is false; that is, suppose that the
mental state M is causally sufficient for B and that M is realized by a distinct
physical state P but P is not causally sufficient for B. Then it follows that
MENTAL CAUSATION IN THE PHYSICAL WORLD 63
among the minimal physical duplicate worlds there are someworlds in which
P holds but B does not. But by definition of supervenience, the P-worlds are
allM-worlds, so the worlds in question must be ones in whichM holds but B
does not hold. But this contradicts the assumption thatM is causally sufficient
for B, so demonstrating the falsity of our initial supposition.
(4) The new exclusion principle: if a mental state M is realized by a distinct physical
state P that is causally sufficient for B, then M does not cause B.
This principle is related to an exclusion principle that Jaegwon Kim (1998,
2005) formulates as follows:
(5) Kim’s exclusion principle: if a state S1 is causally sufficient for a state S2, then no
distinct state obtaining at the same time as S1 can cause S2.2
It’s easy to see that the new exclusion principle above follows from Kim’s
principle. Suppose that a stateM is realized by a physical state P that is causally
sufficient for B. ThenM and P obtain at the same time, and so it follows from
Kim’s principle thatM can’t cause B. On the other hand, it can be seen that
the new exclusion principle doesn’t imply Kim’s principle. Suppose that
mental stateM and physical state P obtain at the same time but are not related
by supervenience. Since the new exclusion principle only applies to pairs of
events related by supervenience, nothing follows from the principle con-
cerning whether M excludes P from causal efficacy. These considerations
show that the new principle is weaker than Kim’s. Kim (2005) says that his
exclusion principle is an analytic, a priori truth; and if this is correct it would
follow that the new exclusion principle is also such a truth. We shall see later
that neither principle is a truth of this kind. For now I simply rest content that
the new principle is no more implausible than Kim’s principle.
We are now in a position to formulate the new exclusion argument.
I present the argument schematically, but with a little effort it can be easily
2 Kim usually formulates his exclusion principle with the qualifying clause ‘unless it is a case of
genuine overdetermination’. However, I omit this qualification in order to simplify my discussion. By
the term ‘a case of genuine overdetermination’, Kim means the kind of situation in which the multiple
causes operate independently of each other in much the manner of two assassins who, completely
independently of each other, shoot the same victim at the same time. Since there is general agreement
that mental causation doesn’t involve this kind of overdetermination, it’s reasonable to assume that the
condition specified in the clause is not satisfied in the case of mental causation and so can be safely
ignored. Nonetheless, the whole question of whether mental causation involves overdetermination is
discussed further in section 6 below.
64 PETER MENZIES
translated into a concrete argument using the example about the monkeys’
neurally encoded intentions.
Suppose, for reductio, that a particular state M causes some behaviour B.
By (1) the state M is realized by a distinct physical state P.
By (2) the state M is causally sufficient for B
By (3) the state P is causally sufficient for B.
By (4) the state P excludes M as a cause of B.
Hence, a contradiction.
This argument is extremely simple. But a physicalist who accepts the
assumptions and principles listed above must accept the conclusion that
the causal efficacy of any mental state is excluded by that of its underlying
physical realizer.
This argument is related to Jaegwon Kim’s famous exclusion argument
(1998, 2005), which has a slightly broader target. Kim’s exclusion argument
attempts to show that any kind of property dualism that implies the mental
properties are distinct from physical properties is committed to epipheno-
menalism about the mental. Property dualists include both non-reductive
physicalists who accept the supervenience of mental properties on physical
and non-physicalists who deny this. Kim’s argument proceeds from slightly
different premises to the same conclusion about epiphenomenalism about the
mental. The differences between the arguments result from the fact that they
have different targets. Kim’s argument establishes a stronger conclusion that all
forms of property dualism are committed to epiphenomenalism about the
mental, but it proceeds from stronger premises, in particular relying on a
contingent principle to the effect that the physical world is causally closed.
3. Descartes’ Revenge
I have argued that the new exclusion argument is a problem for non-
reductive physicalists. However, it might be doubted whether the argument
is effective against reductive physicalists who assert the identity of mental
properties with physical properties. There is some reason for this scepticism,
as the premises of the argument all concern the relationship between mental
properties and distinct physical properties that form their supervenience
bases. For example, the supervenience premise states that a mental property
MENTAL CAUSATION IN THE PHYSICAL WORLD 65
supervenes on a distinct physical property, or alternatively, that the mental
stateM is realized by a distinct physical state P. In any case, even without the
distinctness qualification, the argument would be powerless against the
identity theory. If M is identical to P, then the argument generates no
contradiction at all: the fact that P is causally sufficient for B doesn’t
undermine the causal efficacy of M with respect to B given that P and M
are the same state.
In this connection, it is worth noting there was something unrealistic in
the way I presented the example about Andersen’s work on the neural
encoding of monkeys’ intentions. I ran an informal version of the argument
by supposing that the monkey’s intention Ii is realized by a distinct neural
state Ni1; and then appealing to exclusion reasoning to show that the causal
powers of the intention are excluded by those of the underlying neural state.
But it would be more reasonable to think in this example that the monkey’s
specific intention is actually identical with a neural state: namely, the
aggregate neural state of neurons in the cluster having a collective average
firing rate. This abstract neural state is plausibly represented as a disjunction
of all the more specific realizers Ni =Ni1 vNi2 vNi3 v . . . vNin, where each
disjunct represents a highly specific neural state consisting of a temporally
ordered sequence of the individual neural firings such that together these
states have a specified average firing rate. It is indeed reasonable to think that
Andersen and his colleagues thought of the monkeys’ intentions as being
identical with such aggregate neural states, as what they were directly trying
to manipulate were the aggregate patterns of firings. So these experimenters
are best seen as reductive physicalists who could plausibly claim that the new
exclusion argument is not effective against their view.
However, let’s subject this thought to more scrutiny. Does the reductive
physicalist really emerge unscathed by the new exclusion argument? Let’s
suppose that every mental state is identical to a physical neural state, perhaps
a fairly abstract one. Let’s suppose that each of the monkey’s intentions can
be identified with an abstract aggregate pattern of neural firing, as suggested
above. It would still seem that the argument should go through to show that
this abstract physical state will have its causally efficacy pre-empted by the
more specific neural state which realizes it. The only assumption required
to kick-start the argument is the assumption that the target state is realized
by a distinct underlying physical state. And this is true for the neural state
Ni, which on a given occasion will be realized by one state from the set
66 PETER MENZIES
{Ni1,Ni2,Ni3, . . . ,Nin}. These are more highly specific versions ofNi, each
being a complex state consisting in a temporally ordered sequence of
individual neural firings that satisfies the aggregate average associated with
Ni. These highly specific states are distinct fromNi and so the new exclusion
argument goes through to show that the causal efficacy of Ii =Ni is excluded
or pre-empted by the causal sufficiency of one of the highly specific realizer
states. It is easy to see that the argument can be reproduced for any physical
property that one cares to identify with a mental property: by taking its
realizers to be more fine-grained specifications of this physical property, one
can show that the physical property has its causal powers pre-empted by its
more fine-grained realizers.
It doesn’t require much effort to see that the physicalist world picture
itself will lead to a natural generalization of the argument. As defined above,
physicalism is committed to a multilayered model of reality, stratified into
different levels and bottoming out in fundamental physical level.3 Entities
belonging to a given level have an exhaustive decomposition without
remainder into entities belonging to entities belonging to the next level
down. So living organisms can be decomposed into cells, which can be
decomposed into molecules, then atoms and so on to the basic fundamental
physical particles—perhaps the quarks, leptons, and bosons of the standard
model. Physicalism, as I have defined it, accepts that the distribution of
fundamental particles with their properties and relations, together with the
way this distribution evolves in conformity with the fundamental physical
laws, will fix everything else in reality.
If this picture is correct, then any non-fundamental state, whether it be a
mental state, a neurophysiological state, a biochemical state, molecular state,
or atomic state, will have fine-grained realizers at the next level down that
are distinct from it. These realizers will consist in lower-level states involv-
ing constituents of the objects of the higher-level states configured into
3 It might be argued that reductive physicalism is better understood as committed to a flat, one-layer
view of reality rather than a multilayered view like non-reductive physicalism. However, contrary to this
view, I maintain that while reductive physicalists believe that the fundamental physical level is the basis
for all real entities (objects and properties), they must nevertheless accept the derivative reality of entities
at higher levels. Reductive physicalists can, for example, accept the reality of biological organisms and
biological properties because they are constituted out of fundamental physical objects and properties.
Accordingly, even the austere metaphysics of reductive physicalism makes room for a multilayered
conception of reality: the different levels of reality are ordered in terms of the complexity of the
constructions of the higher-order objects and in terms of the supervenience relations between properties.
MENTAL CAUSATION IN THE PHYSICAL WORLD 67
complex arrangements. So for example, if we take one of the fine-grained
neural states Nij above, which consists in a specific temporally ordered
sequence of individual neuron firings, then this state can be decomposed
into a highly complex state involving molecular arrangements and processes
that actually realize the sequence of neural firings. This complex molecular
state will be distinct from the given neural state because there are many
other arrangements of molecules and molecular processes that could realize
the same neural state. This process of decomposition can reiterate until one
finally reaches a decomposition of the neural state in terms of a state at the
level of fundamental physics. An application of the new exclusion principle
at each stage of the decomposition process will show that the causal efficacy
of each state resulting from the decomposition is rendered void by the causal
powers of its realizer, until we finally reach the last stage where only the
fundamental physical realizer state has any causal powers.
In this way, I would argue, the new exclusion argument can be general-
ized so that it applies even in the situation in which the mental state is
identified with a neural state. The generalization of the argument depends
only on the assumption guaranteed by physicalism that any non-fundamental
state, whether it be a mental, neural, chemical, or atomic, will be realized by a
distinct finer-grained physical state, and indeed ultimately by the finest-
grained state of all, the realizing state specified by fundamental physics. The
upshot of the generalization of the argument is that the only causal powers are
those possessed by fundamental physical properties and states. Ned Block
(2003) aptly calls this the causal drainage problem because the causal powers of
properties and states at all but the lowest level drain away to the lowest level.
This is an indeed serious problem for any physicalist, whether reductive or
non-reductive, who accepts the five assumptions and principles required to
generate the exclusion argument.
Let’s take stock of where we have arrived in our reasoning. At the outset,
we saw that Descartes’ substance dualism was jettisoned because of its
inability to explain the causal interactions between mind and body. It’s
impossible, so it is said, to vindicate mental causation within the metaphys-
ical model of substance dualism. At first sight, physicalists seem to be in a
better position to carry out the project of vindication. After all, what could
be clearer than the fact that mental states at the very least supervene on
physical neural states that can evidently produce or give rise to behaviour?
But now it would seem that the new simple exclusion argument threatens to
68 PETER MENZIES
muddy this clarity. By generalizing the argument in an apparently unprob-
lematic way, one can show that not only mental causation is unreal, but
biological, chemical, atomic causation are all unreal too. Moreover, the
premises from which the new exclusion argument proceeds are principles
that are definitive of physicalism (such as the supervenience thesis), or are
analytic or close to being so (like the principle of transmission of causal
sufficiency across realization), or are assumptions that many physicalists take
to be unexceptionable (like the new exclusion principle). If we should
repudiate substance dualism because of its failure to vindicate mental caus-
ation, what should we say about physicalism, both of the reductive and non-
reductive variety, that evidently fails to vindicate any kind of upper-level
causation? I entitled this section “Descartes’ Revenge”, because it would
seem that, from the point of view of commonsense plausibility, physicalism
is in a much worse position than substance dualism.
I have argued that the new exclusion argument applies more generally to
show that all non-fundamental properties and states are epiphenomenal.
This surely gives us reason to suspect that the argument is unsound. I will
eventually point the finger at the new exclusion principle as the false
premise that physicalists should reject. However, before moving on to
this, I wish to consider whether some philosophical manoeuvre will stop
the drainage of causal powers to the fundamental levels of physics. In this
connection, a number of philosophers have claimed that Kim’s superve-
nience argument, which is closely related to his exclusion argument, also
generates a causal drainage problem and must be defective for this reason.
However, Kim has responded that his supervenience argument doesn’t
generalize in this untoward way. So let’s examine Kim’s defence to see
whether it can be used to invalidate the generalization of the new exclusion
argument mooted above.
Kim’s defence (1998, 2005) involves his distinctive conception of the
hierarchy of levels. According to Kim, the hierarchy of levels applies in the
first instance to objects: objects are ordered into the hierarchy of levels on
the basis of the part–whole relation so that entities at one level are composed
out of smaller constituent entities at the next level down. The crucial point
that Kim makes is that his supervenience argument posits a supervenience
relation between mental and physical properties, which must, by virtue of
the definition of property supervenience, belong to the same object. In
order for the supervenience argument to give rise to causal drainage from
MENTAL CAUSATION IN THE PHYSICAL WORLD 69
the macro to micro-levels, the supervenience relation would have to apply
to properties belonging to objects at different levels. But this is not possible.
So, the worries raised by his supervenience argument are intra-level con-
cerns that do not cut across levels, in particular micro–macro boundaries.
Kim writes:
In general, supervenient properties and their base properties are instantiated by
the same objects and hence are on the same level . . . So the microphysical, or
mereological, supervenience does not track the micro–macro hierarchy: the
series of supervenient properties, one mereologically supervenient on the next,
when we go deeper and deeper into the micro, remains at the same level in the
micro–macro hierarchy . . . This means that the supervenience argument, which
exploits the supervenience relation, does not have the effect of emptying macro-
levels of causal powers and rendering familiar macro-objects and their properties
causally impotent. (1998, 86)
What is to be said in response to this? First, whatever is true of Kim’s
supervenience argument, it is not true of his own exclusion argument that
the properties that compete for causal efficacy belong to the same object.
His exclusion argument starts from the supposition that an organism’s
having a mental property causes its physical behaviour; and then, invoking
the physical causal closure principle, posits a simultaneous physical state that
is causally sufficient for this behaviour. The principle doesn’t require that
this physical state should be an instantiation of a physical property by the
very organism with the mental property: the physical state may consist in a
configuration of more basic physical objects having certain properties and
bearing certain relations to each other. So, whatever may be true of Kim’s
supervenience argument, his exclusion argument can cut across levels to
render higher-level properties and states causally redundant.
Secondly, this defence turns on the fact that the standard definition of
property supervenience requires that supervening and base properties
belong to the same object. But this is simply a peculiarity of the definition
of this type of supervenience. What’s actually crucial to the new exclusion
argument is the assumption that a higher-level state supervenes on, or is
realized by, a lower-level state. It is possible to define the supervenience of
one kind of state on another in a way that doesn’t require that the super-
vening and subvening states should be instantiations of properties by the
same object. For example, here is one such definition: a state S1 is realized
70 PETER MENZIES
by a state S2 in the actual world if and only if in the set of minimal physical
duplicates of the actual world, all worlds in which S2 holds are worlds in
which S1 holds. This definition doesn’t require that S1 and S2 should consist
in the same object instantiating different properties.
Thirdly, even assuming supervening and subvening properties must belong
to the same object doesn’t stop the causal drainage from non-fundamental to
fundamental properties. For Kim’s own notion of a micro-based property
(Kim 1998, 114) can be invoked to allow property supervenience to cut across
micro–macro boundaries. For an object to have a micro-based property is just
for the object to be decomposable into non-overlapping proper parts, each of
which has certain properties and all of which bear a certain relation to each
other. In other words, an object’s having a micro-based property is simply
constituted by the complex state of its proper parts being configured in a
certain way. So the property of instantiating a temporally ordered sequence
of neural firings may be a micro-based property, since a person has this
property just when the neurons of his brain enter into a certain complex
spatiotemporal relations with each other and have distinctive properties. So
a person may have a mental property and also this micro-based property,
meaning that the supervenience and realization relations can hold between
these properties. Clearly then the generalization of the new exclusion
argument can appeal to this kind of micro-based property to generate its
unacceptable conclusion that the causal powers of all macro-properties
drain away to the fundamental physical level.
So, I conclude that nothing that Kim says against the generalization
objection can assuage the worries about causal drainage.
4. Causal Relevance4
Where does the new exclusion argument go awry? I suggest that the error of
the new exclusion principle lies in its claim that one state’s causal sufficiency
for an effect excludes the causal efficacy of any state supervening on the first
state. The fundamental error of this principle is that it mistakes causal
sufficiency for causation. The fact that causal sufficiency doesn’t amount
4 This section and the next report on work done in collaboration with Christian List in List and
Menzies (2009).
MENTAL CAUSATION IN THE PHYSICAL WORLD 71
to causation has been known for some time. Wesley Salmon (1971) pointed
out that causal explanation requires information of the right kind and in the
right amount, or in other words, causal explanations must cite causally
relevant factors. However, while a man’s taking a contraceptive pill is
causally sufficient for his not getting pregnant, there is no causal relevance
here, as the man’s taking a contraceptive pill makes no difference to his not
getting pregnant. Even if had not taken the pill, he wouldn’t have got
pregnant.
As several philosophers have noted, the man’s taking the contraceptive
pill does not satisfactorily fit the role of a cause because it is overly specific
and involves extraneous detail. To be sure, the man’s taking the contracep-
tive pill is causally sufficient for the effect, but causal sufficiency is not the
same thing as causation, which requires causal relevance. To illustrate the
difference, consider an example of Stephen Yablo’s (1992) concerning a
pigeon that has been trained to peck at all and only red objects. The pigeon
is presented with a red target and she pecks at it. As it happens, the target is a
specific shade of crimson. What caused the pigeon to peck? Was it the fact
that the target was red or the fact that it was crimson? The exclusion
principle would say that since being red is realized by being crimson and
being crimson is causally sufficient for the pigeon’s pecking, the redness of
the target is not the cause. But this seems wrong, as Yablo points out: the
target’s being red is of the right degree of specificity to count as a cause of
the pigeon’s action. In contrast, the target’s being crimson is too specific to
count as the cause: citing it as the cause of the pecking might give the
erroneous impression that the pigeon would not peck at anything non-
crimson.
Many philosophers have sought to capture the idea of causal relevance in
terms of the dictum that causes make a difference to their effects.5 How are
we to cash out this dictum in more precise terms? I agree with those
philosophers (Pearl 2000; Spirtes et al. 2000; Hitchcock 2001; Woodward
5 Only the simplest form of causal relevance can be captured in terms of difference-making, as
explained below. The full complexity of the concept of causal relevance requires an account that goes
well beyond the simple outline I sketch below. For example, Woodward’s (2003) interventionist
account provides detailed explanations of a range of type-level causal concepts including direct, total,
and contributing causes that involve sophisticated elaborations of the basic notion of difference-making.
It is also important to note that the account below is not intended to handle the complications involved
in pre-emption and overdetermination examples. Such examples require a more sophisticated treatment.
72 PETER MENZIES
2003) who interpret causal claims as claims about relationships between
variables, and so interpret the dictum, quite literally, as requiring that
changing the value of the cause variable changes the value of the effect
variable. Applied to binary variables representing the presence or absence of
some state, the dictum says that changing the causal state from being absent
to being present (or vice versa) changes the effect state from being absent to
being present (or vice versa). Formally, I suggest that one state makes a
difference to another just when the following conditions are satisfied:
Truth conditions for causal relevance (or making a difference): The state S1 makes a
difference to the state S2 in the actual world just in case (i) if in any relevantly
similar possible situation S1 holds, S2 also holds; and (ii) if in any relevantly
similar situation world S1 does not hold, S2 does not hold.
For example, the target’s being red makes a difference to the pigeon’s
pecking because in any relevantly similar situation in which the pigeon is
presented with a red target it pecks, and in any relevantly similar situation in
which it is not presented with a red target it does not. The relevantly similar
situations in this example are ones in which the pigeon has received the
same training, the targets are presented to the pigeon in the same experi-
mental setting, there are no confounding influences on the pigeon and so
on. But under this construal of the relevantly similar situations, the target’s
being crimson does not make a difference to the pigeon’s pecking. Condi-
tion (ii) is not met: in a relevantly similar situation in which the pigeon is
presented with a non-crimson but red target, it still pecks. These observa-
tions confirm the conjecture that the requirement that causes make a
difference to their effects captures the crucial notion of causal relevance.
Further confirmation of this conjecture comes from examining how the
suggested truth conditions constrain the specificity of causes: satisfaction of
these conditions ensures that causes are specific enough for their effects, but
no more specific than needed. This is revealed most clearly in the case of
many-valued causal variables. Suppose, for example, there is a drug that
causes patients to recover from an illness. The effect variable is a binary
variable whose values are recovery or non-recovery. But the cause variable
is many-valued, with possible values 0mg, 50 mg, 100mg, 150 mg, and 200
mg. Suppose that any regular dose at or above 150 mg cures a patient, but
any lower dose does not. Suppose a patient has taken a regular dose of 150
mg and has recovered from the illness. What made the difference to the
MENTAL CAUSATION IN THE PHYSICAL WORLD 73
patient’s recovery? According to the truth conditions above, the answer is
‘Giving the patient a dose of at least 150 mg’. It satisfies both conditions (i)
and (ii): all relevantly similar patients who take a regular dose at or above 150
mg recover and all those who take a lower dose don’t. Other answers are
either too specific, or not specific enough. For example, the cause cannot be
‘Giving the patient a dose above 50 mg’ because that does not meet
condition (i): some relevantly similar patients who are given a dose above
50 mg, say 100 mg, do not recover. Similarly, it cannot be ‘Giving the
patient a dose of exactly 150 mg’ because that does not meet condition
(ii): some relevantly similar patients who are not given a dose of exactly
150 mg—say they are given 200 mg—nonetheless recover. In this way,
condition (i) rules out causes that are not specific enough to account for the
change in the effect variable, while condition (ii) rules out causes that are
too specific to account for it.
The truth conditions for making a difference can be expressed more
formally using counterfactuals, as understood in a possible-world semantics.
Specifically, let’s replace the notion of a relevantly similar situation with that
of a relevantly similar possible world, and thus rewrite the conditionals in
the truth conditions above as counterfactuals:
Truth conditions for causal relevance (making a difference): The S1makes a difference to
S2 in the actual world if and only if it is true in the actual world that (i) S1 holds
&? S2 holds; and (ii) S1 doesn’t hold &? S2 doesn’t hold.
I interpret the counterfactuals in accordance with the standard possible-
worlds semantics of David Lewis (1973), which provides truth conditions
for counterfactuals in terms of a similarity relation between possible worlds.
The similarity relation, which may vary with context, is represented by an
assignment to each possible world w of a system of spheres of worlds centred
on w. The system of spheres conveys information about the similarity of
worlds to the world w at the centre. The smaller a sphere, the more similar
to w are the worlds in it. So whenever one world lies in some sphere around
w and another lies outside it, the first world is more similar to w than the
second. In terms of this system of spheres, I now state the truth conditions
for counterfactuals as follows: P &? Q is true in world w if and only if Q is
true in all the closest P-worlds to w.
By adopting this semantic framework, I follow Lewis rather than Stalna-
ker, in allowing that there may be more than one closest P-world to w.
74 PETER MENZIES
Although there may sometimes be just one such world, this is not the
general rule. However, I diverge from Lewis in imposing only a weak
centring requirement on the systems of spheres. I allow the smallest sphere
around w to contain more than one world. Lewis imposes the stronger
requirement that the smallest sphere around w contains only w. This corres-
ponds to a constraint on the similarity relation whereby no world is as
similar to w as w itself. It also corresponds to the inference rule from the
premise P & Q to the conclusion P &? Q. In other words, if P and Q are
true in some world so is P &? Q. Lewis’ strong centring requirement, the
corresponding constraint on similarity and the corresponding inference rule
may appear plausible. But I cannot accept them. If the counterfactual
formulation of the truth conditions for causal relevance is to match the
earlier formulation, clause (i) of the counterfactual formulation must capture
the idea that every relevantly similar situation in which S1 holds S2 also
holds. In the original formulation, this condition is non-trivial: it rules out
insufficiently specific causes, provided the set of relevantly similar situations
in which S1 holds includes more than one such situation. To match this
condition, the counterfactual formulation must require that even if S1 and
S2 hold in the actual world, the smallest sphere around it also contains some
other worlds in which S1 holds.
Before I apply the difference-making account of causation to the exclu-
sion principle, I note an implication of the account. Several philosophers
(Hitchcock 1993; Woodward 2003; Schaffer 2006) have observed that causal
statements are contrastive in character. They have pointed out that descrip-
tions of both cause and effect seem to involve reference to a contrast
situation, or set of contrast situations. Sometimes the contrasts are made
obvious by the use of contrastive focus. For example, asserting a sentence
such as ‘Giving the patient a 150 mg dose of the drug caused his recovery’
highlights the fact that the 150 mg dose was one in a range of doses and not
all doses within this range cause recovery. But often the contrast situations
are left implicit. The rule for reconstructing the contrast situations is
straightforward in the case of causal claims involving binary variables.
Here the contrast situation is simply the opposite value to the actual one.
So the causal claim ‘The state S1 caused S2’ is to be understood as ‘S1’s
holding rather than not holding caused S2 to hold rather than not hold.’ All
these observations are predictable based on the account of causation as
difference making. If causal statements convey information about how
MENTAL CAUSATION IN THE PHYSICAL WORLD 75
variation in one variable is associated with variation in another, as explicated
by a pair of counterfactuals, it is no surprise that they can be expressed
contrastively.
The various examples discussed above—Andersen’s monkey and Yablo’s
pigeon—can be seen as counterexamples to the exclusion principle. In each
case, the exclusion principle leads us to identify the intuitively wrong
property as the cause of the given effect. In Yablo’s example, the causally
relevant cause of the pigeon’s pecking is not the crimson, but the redness of
the target, contrary to what the exclusion principle implies. This is sup-
ported by the truth of the counterfactuals:
Target is red &? pigeon pecks.
Target is not red &? pigeon does not peck.
In contrast, the following counterfactuals are not both true:
Target is crimson &? pigeon pecks.
Target is not crimson &? pigeon does not peck.
It is natural to interpret these counterfactuals in terms of a similarity relation
that makes the closest worlds in which the target is not crimson ones where
it is some other shade of red.6 Given this assumption, the second counter-
factual is false: in the closest worlds in which the target is not crimson it is
some other shade of red, in which case the pigeon will still peck.
A similar treatment can be given for the example of the monkey. The
causally relevant cause of the monkey’s reaching actionAi is not its particular
neural state Ni1, but its intention Ii. The following counterfactuals are true:
Monkey has intention Ii &? monkey performs Ai.
Monkey doesn’t have intention Ii &? monkey doesn’t perform Ai.
Whereas the following counterfactuals are not both true:
Monkey has neural property Ni1 &? monkey performs Ai.
Monkey doesn’t have neural property Ni1 &? monkey doesn’t perform Ai.
Assuming that the closest worlds in which the monkey doesn’t have neural
property Ni1 are ones in which it has another neural property realizing the
intention Ii, one can see that the second counterfactual is false: in any such
6 I discuss this assumption further in the next section.
76 PETER MENZIES
world, the monkey has another neural property that realizes Ii and so
performs Ai.
In summary, requiring causes to be causally relevant, one can see that the
exclusion principle is false. Even when some state S1 is causally sufficient for
another state S2, a state that supervenes on S1 can nonetheless be a cause of
S2. The monkey’s intention Ii to reach for a specific target is the cause of its
reaching action Ai even though it is realized by the neural property Ni1,
which is causally sufficient for the action.
5. Revised Exclusion Principle
We have seen that within the framework of a difference-making account of
causation there are some persuasive counterexamples to the exclusion
principle. A central feature of this principle is that it is couched in terms
of causal sufficiency: it states that a property that is causally sufficient for some
effect excludes certain other properties from being causes of that effect. But
one might ask: “Why talk of causal sufficiency rather than causation?”
Naturally, this raises the question of what happens if we reformulate the
exclusion principle, replacing the reference to causal sufficiency with one to
causation in a more adequate sense, understood as difference-making. So
let’s consider the following revised principle:
Revised exclusion principle: For all distinct states S and S* such that S* is realized by
S, S and S* do not both cause state T.
Here the truth conditions for causation are those for difference-making
introduced above. The principle can also be formulated in two different,
but logically equivalent ways. The first is the counterpart of the original
principle, whereas the second is seldom explored in the debate about the
exclusion problem:
Revised exclusion principle (upwards formulation): If a state S causes a state T, then no
distinct state S* that supervenes on S causes T.
Revised exclusion principle (downwards formulation): If a state S causes a property T,
then no distinct state S* that realizes S causes T.
Although logically equivalent, the two formulations draw our attention to
two different ways in which the exclusion principle can apply. An instance
MENTAL CAUSATION IN THE PHYSICAL WORLD 77
of upwards exclusion occurs when there exists a subvenient difference-making
cause that excludes a supervenient one; and an instance of downwards
exclusion, usually overlooked, occurs when there exists a supervenient
difference-making cause that excludes a subvenient one.
Is the revised exclusion principle true or false? Let’s focus on the instance
of the principle that concerns the causal relationships between a mental state
M, a neural state N, and a behavioural state B. Throughout the discussion,
I assume that N realizes M in the actual world. We are interested in the
logical relationship between the following two propositions:
(1) M is a difference-making cause of B.
(2) N is a difference-making cause of B.
Using the truth conditions introduced above, each of these propositions is
equivalent to a conjunction of counterfactuals:
(1a) M holds &? B holds.
(1b) M doesn’t hold &? B doesn’t hold.
(2a) N holds &? B holds.
(2b) N doesn’t hold &? B doesn’t hold.
The revised exclusion principle dictates that propositions (1) and (2), or
equivalently (1a), (1b), (2a) and (2b), are never simultaneously true. But is
this claim actually correct? One benefit of formulating the difference-
making conception of causation in terms of counterfactuals is that it
makes this question logically tractable. One can prove that these four
counterfactuals hold only under very special conditions. To state this result,
call a causal relation betweenM and B realization-sensitive if B fails to hold in
all thoseM-worlds that are closest �N-worlds (i.e., whereM has a different
realizer from the actual one). The result is the following:
Compatibility Result (List andMenzies 2009): IfM causes B, thenN causes B if and
only if the causal relation between M and B is realization-sensitive.
Rather than prove this result here, it is more instructive to describe a
situation that exemplifies the result. So consider the situation represented
in Figure 3.1. The concentric spheres represent sets of more andmore similar
worlds to the actual world; the innermost sphere contains the actual world,
labelled w, and the other worlds deemed maximally similar to it. The set of
N-worlds is represented by the convex region with light shading, and the set
78 PETER MENZIES
of M-worlds by the larger convex region that includes the set of N-worlds.
The regionwith darker shading represents the set ofB-worlds and this region
coincides with the innermost sphere. In this situation, it is easy to see thatM
causes B. First, since M holds throughout the innermost sphere, that sphere
picks out the closest M-worlds, and since B also holds in it, counterfactual
(1a) is true. Second, since B does not hold in any�M-worlds, it fails to hold
in all the closest�M-worlds and thus counterfactual (1b) is true. Further, the
causal relation between M and B is realization-sensitive: since B does not
hold in any�N-worlds, it follows a fortiori that it does not hold in any of the
closest �N-worlds that are M-worlds. And finally, N does indeed cause B:
counterfactuals (2a) and (2b) can easily be verified to be true.
It is important to note, however, that the conditions under which the
counterfactual pair (1a)–(1b) implies the pair (2a)–(2b) are very special.
Figure 3.1 illustrates this point nicely. Although both M and its actual
realizing state N are difference-making causes of B here, the realization-
sensitivity of the causal relation betweenM and Bmeans that small perturb-
ations in the way in whichM is realized would result in the absence of B. In
other words, if M were realized by any neural state other than N then B
would cease to hold. When might we expect the conditions for realization-
sensitivity to obtain? If the mental property M were identical to the neural
property N, then we would certainly expect instances of M to stand in
realization-sensitive causal relations with respect to instances of N. The
fact that M-instances had certain effects when and only when N-instances
are present would simply reflect the identity of the properties. However,
M
B
N
.w
Fig. 3.1
MENTAL CAUSATION IN THE PHYSICAL WORLD 79
I do not rule out the possibility of other explanations of the existence of
realization-sensitive causal relations.
At this point it is useful to consider a logically equivalent formulation
of the Compatibility Result that is relevant to Downwards Exclusion. In
analogy with the earlier definition, call a causal relation between M and B
realization-insensitive if B holds in someM-worlds that are closest�N-worlds
(i.e., where M has a different realizer from the actual one). The following
proposition is an immediate corollary of the Compatibility Result:
Downwards Exclusion Result (List and Menzies 2009): If M causes B, then N does
not cause B if and only if the causal relation between M and B is realization-
insensitive.
Again let’s consider a schematic example that exemplifies this proposition,
focusing on the situation represented in Figure 3.2. As before, the system of
spheres represents sets of worlds with greater or lesser degrees of similarity to
the actual world, labelled w. The set ofN-worlds is represented by the convex
region with lighter shading, and the set ofM-worlds by the larger vex region
that includes the set of N-worlds. The region with darker shading, which
coincides with the innermost sphere, represents the set of B-worlds. This
figure shows thatM causesB, sinceB holds in all the closestM-worlds and fails
to hold in all the closest�M-worlds, i.e., counterfactuals (1a) and (1b) are both
true. It is also easy to see that this causal relation is realization-insensitive:
B continues to hold in some, indeed all, of the M-worlds that are closest
�N-worlds. Finally, it is easy to see thatN does not causeB: the counterfactual
(2b) is false, since B holds in all the closest �N-worlds.
B
M
N
.w
Fig. 3.2
80 PETER MENZIES
6. Some Implications for Compatibilism
To highlight the significance of these results I want to examine some of
their implications for a non-reductive physicalist solution to the mental
causation problem that has come to be called compatibilism. This has
recently become a popular position among philosophers. (See Thomasson
1998; Crisp and Warfield 2001; Pereboom 2002; Bennett 2003, 2008.)
This solution says that any piece of intentional behaviour has two causes: a
mental state and the neural state that realizes it. These causes are not partial
causes in the way that a short circuit and the presence of oxygen are each
partial causes of a fire since the mental state and its realizing neural state are
each causally sufficient by themselves for the physical behaviour. Nor are they
standard overdetermining causes in the way that two assassins who fatally
shoot their victim at the same time are each causes of the victim’s death. For
themental state and its realizing neural state aremetaphysically connected in a
way that the overdetermining causes usually are not. In short, the compatibi-
list solution says, very roughly, that a mental state and its realizing neural state
are non-standard overdetermining causes of a piece of intentional behaviour.
Sydney Shoemaker (2007) has provided the most fully developed version
of the compatibilist solution. Shoemaker argues that in any given world a
property’s identity is determined by its causal profile—the set of its forward-
looking causal powers to cause other properties and its backward-looking
causal powers to be caused by other properties. In other words, whether
properties count as the same or different depends on whether they have the
same causes and effects.7 In his discussion Shoemaker gives equal weight to a
property’s forward- and backward-looking causal powers. But it will serve
my exposition to focus just on properties’ forward-looking causal powers.
Further, talking about a property’s forward-looking causal powers is short-
hand for talking about the causal powers of instances of the property. An
instance of a property F has the causal power to produce an instance of the
property G just in case the first instance can in suitable conditions cause the
second. The causal powers of properties are to be understood in terms of
generalizations about the causal powers of their instances.
7 While Shoemaker actually believes that a property’s causal profile defines its inter-world identity, his
account of the realization relation depends only on causal profiles providing an intra-world criterion of
identity.
MENTAL CAUSATION IN THE PHYSICAL WORLD 81
Shoemaker (2007, Chap. 2) explains realization in terms of the inclusion
of causal powers. He states that one property G realizes another property F
just in case the forward-looking causal powers of F are a subset of the causal
powers of G. So a neural state N realizes a mental state M just in case the
forward-looking causal powers of the mental property are a subset of the
causal powers of the neural property. This definition is supposed to avoid
the problem that the causal role of the realized property is pre-empted by
the causal role of the realizer property. He remarks that, on the contrary, his
account starts with the supposition that the realized property has a causal
profile, and nothing in the account takes this assumption back. Moreover,
his account avoids appealing to a problematic kind of overdetermination.
While it is true that a mental state and a neural state are both causes of
behaviour, the neural state has its causal role in virtue of the causal role of the
mental state. He explains this by analogy with an example. Suppose a firing
squad fires a salvo of shots at Smith but the only shot that hits Smith is one
fired by Jones. In this case the salvo killed Smith, but it did so in virtue of a
particular shot, Jones’ shot, that killed Smith. As a further illustration, he
cites Yablo’s example about Sophie, the pigeon:
Now we take scarlet as a realizer of red. The forward-looking causal features of a
red are a subset of the forward-looking features of scarlet . . . This instantiation of
red was realized in an instantiation of scarlet, and the instantiation of scarlet was
of course causally sufficient (in the circumstances) for the occurrence of Sophie’s
pecking. But it seems right to say that it was the instantiation of red, not the
instantiation of scarlet, that caused Sophie’s pecking. (2007, 14)
The instance of scarlet caused Sophie’s pecking but it did so in virtue of the
realizing the instance of red.
Crucial to the feasibility of the compatibilist solution is acceptance of two
principles that are implicit in Shoemaker’s discussion. The first principle is that
causal sufficiency is the same thing as causation. Shoemaker is tacitly committed
to this principle. In the quoted passage above, for example, Shoemaker moves
back and forwards from the claim that an instance of a colour is causally sufficient
for the pigeon’s pecking to the claim that it is a cause. The second principle is
what I earlier called the transmission of causal sufficiency across the realization
relation: if a property is causally sufficient for another property then any
property that realizes the first property is also causally sufficient for the second.
This principle holds as an analytic consequence of Shoemaker’s definition of
82 PETER MENZIES
realization in terms of inclusion of causal powers and his assumption that causal
sufficiency is the same thing as causation.8 Acceptance of these principles
allows one to argue that if a mental property is causally sufficient for some
piece of intentional behaviour, then both the mental property and its realizing
neural property are causes of the behaviour and that the neural property is a
cause in virtue of the fact that the mental property is. Though there are other
routes to compatibilism, many philosophers who offer compatibilist solutions
to the mental causation problem seem to be committed to these principles.
Despite its initial plausibility, the compatibilist solution does not, I think,
stand up to critical scrutiny. The solution entails that in Yablo’s example the
red and the scarlet properties are both causes of the pigeon’s pecking; and
that a mental property and its realizing neural property are both causes of
intentional behaviour. If this looks like one too many causes, the multipli-
cation of causes does not end with two causes. In section 3, I argued that the
new exclusion argument generated a causal drainage problem in that causal
powers of higher-level properties drain away to the fundamental physical
level. The principles used in this argument generate a descending sequence
of properties, one for each level in the hierarchy of levels, each of which is
causally sufficient for some effect. To generate such a sequence all that is
required is the assumption that some higher-level property, say a mental
property, is causally sufficient for some effect, say some behaviour. Then by
successive applications of the supervenience principle and the principle of
the transmission of causal sufficiency across realization, one can show there
is a descending sequence of properties, each of which (bar the first) is a
realizer of the one above and each of which is causally sufficient for the
effect. The upshot of this argument is that an effect of a higher-level causal
property such as a mental property has many causes: besides its mental cause,
it has a neural cause, a biochemical cause, a molecular cause, an atomic
cause and so on, one for each level in the hierarchy of levels. Positing
two piggybacking causes is somewhat implausible, but positing a whole
sequence of piggybacking causes for any effect strains credibility.
In any case, a central assumption of compatibilist solutions to the mental
causation problem, namely, that causal sufficiency is the same thing as
8 The principle of the transmission of causal sufficiency across the realization relation does not require
the specific doctrines of Shoemaker’s framework. A compatibilist who held that properties are sets of
possibilia and that realization is a matter of the inclusion of one set of possibilia in another would also be
committed to the principle.
MENTAL CAUSATION IN THE PHYSICAL WORLD 83
causation, is seriously mistaken, as I have been at pains to argue. Replacing the
causal sufficiency conception of causation with a more sensible one like the
difference-making conception does not increase the plausibility of the com-
patibilist solution. Indeed, the downwards-exclusion result, described in the
last section, implies that when a mental state causes some behaviour in a
realization-insensitive way, then the neural state that realizes it cannot also
cause the same behaviour. So when the realization-insensitivity condition is
met, the downwards-exclusion result rules out as impossible the kind of dual
cause solution entailed by compatibilism. There is good reason to think the
realization-insensitivity condition is met in the kinds of cases that compatibilist
take as their standard models. For instance, it is reasonable to think about the
Yablo example that the causal relation between the target’s being red and the
pigeon’s pecking is realization-insensitive in that the causal relation doesn’t
depend on the particularway inwhich the redness of the target is realized by its
being scarlet. More precisely, it is plausible to think that the similarity ordering
of possible worlds relevant to this example is such that in some of the closest
worlds in which the target is not scarlet but still red the pigeon pecks. Given
the reasonableness of this assumption, it follows from the downwards-exclu-
sion result that the target’s being scarlet can’t also be a cause of the pigeon’s
pecking along with the target’s being red. The easiest way to see this is to
consider the counterfactual ‘If the target had not been scarlet, the pigeon
would not have pecked.’ The realization-insensitivity condition implies that
in some of the closest worlds inwhich the target is not scarlet but is nonetheless
red the pigeon pecks, which then implies that this counterfactual must be false.
More generally, it follows from the downwards-exclusion result that if the
causal relation between a mental state and some piece of physical behaviour is
realization-insensitive in theway specified above, then it’s logically impossible for
any state that realizes themental state to be a cause of the same behaviour. So any
kind of compatibilist solution to themental causation is ruled outwhere there is a
reasonable assumption that the mental causation is realization-insensitive.
7. Conclusion
I began this paper by formulating a new exclusion argument that is similar to
Kim’s exclusion argument in attempting to show that a physicalist who
accepts certain principles including a weak causal exclusion principle is
84 PETER MENZIES
committed to the view that mental states and other higher-level states
are causally inefficacious. The argument differed from Kim’s argument,
however, in being directed at all forms of physicalism, both reductive and
non-reductive. My aim was not to endorse the new exclusion argument,
but rather to highlight the absurd implications of the exclusion principle it
employs. I subsequently showed that, within the framework of a difference-
making conception of causation, there are straightforward counterexamples
to this exclusion principle. My intermediate conclusion was that both the
new exclusion argument and Kim’s exclusion argument are unsound to the
extent they rely on an exclusion principle that implies that a mental state is
causally inefficacious with respect to some behaviour if it supervenes on a
physical state that is causally sufficient for the behaviour.
In the second half of the paper I explored a version of the exclusion
principle that is formulated in terms of difference-making rather than causal
sufficiency. The principle states that if two states are related by super-
venience, it cannot be that both states are difference-making causes of
another state. It turns out that while this principle is not a logical truth,
there are nonetheless certain contingent conditions concerning the realiza-
tion-insensitivity of difference-making counterfactuals under which the
principle is true. Moreover, the principle can be applied in an upwards
direction to support exclusion reasoning about upper-level states and, less
familiarly, in a downwards direction to support exclusion reasoning about
lower-level causes. The application of the principle in a downwards direc-
tion has several significant implications, one of which I explored in con-
nection with Shoemaker’s compatibilist solution to the mental causation
problem. The downwards exclusion result implies that the compatibilist
view that a mental state and its underlying realizer are both causes of some
piece of behaviour can’t be correct when causation is understood in a
plausible way in terms of difference-making and when the relevant assump-
tion of realization-insensitivity holds.
References
Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just
Maybe, to Tract It’. Nous, 37: 471–97.
MENTAL CAUSATION IN THE PHYSICAL WORLD 85
Bennett, K. (2008). ‘Exclusion Again’. In J. Hohwy and J. Kallestrup (eds.), Being
Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford
University Press: 280–305.
Block, N. (2003). ‘Do Causal Powers Drain Away?’ Philosophy and Phenomenological
Research, 67: 133–50.
Crisp, T. and T. Warfield (2001). ‘Kim’s Master Argument’. Nous, 35: 305–16.
Fodor, J. (1974). ‘Special Sciences, or the Disunity of Sciences as a Working
Hypothesis’, Synthese, 28: 97–115.
Hitchcock, C. (1993). ‘The Role of Contrast in Causal and Explanatory Claims’.
Synthese, 107: 395–419.
—— (2001). ‘The Intransitivity of Causation Revealed in Equations and Graphs’,
Journal of Philosophy, 98: 273–99.
Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis.
Oxford: Oxford University Press.
Kim, J. (1998). Mind in a Physical World: An Essay on the Mind–Body Problem and
Mental Causation. Cambridge, MA: MIT Press.
—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-
sity Press.
Lewis, D. (1973). Counterfactuals. Oxford: Basil Blackwell.
List, C. and P. Menzies (2009). ‘Nonreductive Physicalism and the Limits of the
Exclusion Principle’. Journal of Philosophy, 1006: 475–502.
Menzies, P. and C. List (2010). ‘The Causal Autonomy of the Special Sciences’. In
G. and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University
Press: 108–28.
Mussallam, S., B. Corneil, B. Greger, H. Scherberger, and R. Andersen (2004).
‘Cognitive Control Signals for Neural Prosthetics’. Science, 305: 258–62.
Pearl, J. (2000). Causality: Models, Reasoning, and Inference. New York: Cambridge
University Press.
Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,
99: 499–531.
Putnam, H. (1975). ‘The Nature of Mental States’. In Mind, Language, and Reality:
Philosophical Papers, Volume 2. Cambridge: Cambridge University Press: 429–40.
Salmon, W. (1971). ‘Statistical Explanation’. In Salmon (ed.), Statistical Explanation
and Statistical Relevance. Pittsburgh, PA: University of Pittsburgh Press: 29–87.
Schaffer, J. (2006). ‘Contrastive Causation’. Philosophical Review, 114: 297–328.
Shoemaker, S. (2007). Physical Realization. Oxford: Oxford University Press.
Spirtes, P., C. Glymour, and R. Scheines (2000). Causation, Prediction, and Search.
Cambridge, MA: MIT Press.
Thomasson, A. (1998). ‘A Nonreductive Solution to Mental Causation’. Philosoph-
ical Studies, 89: 181–95.
86 PETER MENZIES
Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation.
New York: Oxford University Press.
—— (2008). ‘Mental Causation and Neural Mechanisms’. In J. Hohwy and
J. Kallestrup (eds.), Being Reduced: New Essays on Reduction, Explanation, and
Causation. Oxford: Oxford University Press: 218–62.
Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.
MENTAL CAUSATION IN THE PHYSICAL WORLD 87
4
Mental Causation: Ontologyand Patterns of Variation
PAUL NOORDHOF
Physicalism was initially motivated by its ability to deal with the problems of
mental–physical interaction. The most attractive version of physicalism,
though, is one which allows the mental some degree of autonomy with regard
to the physical. Few physicalists feel driven to defend the claim that mental
properties are identical with those which are identified by some suitably refined
version of current physics. Unfortunately, as is only too familiar, non-reductive
physicalism—that which denies such an identification—seems to have signifi-
cant problems withmental causation of its own. In this paper, I begin by setting
out the challenge to its efficacy due to Jaegwon Kim. I shall do this briefly
because I am sure the reader has, by now, tired of seeing this argument stated.
I just want to make a couple of comments upon it for the discussion ahead.
I then discuss two over-reactions to it—one which seeks to understand mental
efficacy non-ontologically in terms of patterns of variation, the other of which
uses the problem to motivate a particular ontology—trope metaphysics.
I explain why I consider these over-reactions, identify what is unsatisfactory
about them, and then take elements of each to motivate my own approach
which, you may not be surprised to learn, captures what is best in both. It also
dealswith an issue aboutmental causation untouched byKim’s initial challenge.
1. Kim’s Exclusion Argument
I’m going to set out Kim’s argument making certain assumptions to fix ideas.
These assumptions don’t change the import of the argument, nor affect the
responses to it I am going to consider. They just simplify presentation.
Specifically, I will assume that the non-reductive physicalist is committed
to holding that there is more than one arrangement (A1, A2, A3 . . . ) of
narrowly physical properties (P1, P2, P3 . . . )—those properties identified
by current physics or a future development of it which suitably resembles
it—such that, for each of them, it is metaphysically necessary that if they are
instantiated, then a certain broadly physical property (BP) is instantiated
(henceforth I will use upper case letters to designate type-properties and
lower case letters to designate specific instances). One subclass of broadly
physical properties is that of mental properties and behavioural properties. It
is because all properties instantiated in the world are either narrowly physical
or broadly physical, that non-reductive physicalism is true.
The appeal to metaphysical necessity is required to capture the fact that
non-reductive physicalism is committed to a tighter connection than mere
nomological necessity between arrangements of physical properties and
mental properties. The latter type of connection would be acceptable to
the emergent dualist. Debate has raged over whether appeal to metaphysical
necessity is sufficient to capture what is required. I have defended this
conclusion (Noordhof 2003, 2010). Nevertheless, all that matters is that it
is stronger than the relation allowed by emergent dualists (bracketing an
issue I touch on in section 5 about a powers ontology). If it is not, not only
do we not have a version of physicalism but the issue set aside in the
comment below about other events in the causal chain or causal circum-
stances becomes salient.
Appeal to metaphysical necessitation may appear too strong (e.g., Kim
2005, 49). Consider the relationship between O, a property occupying a
certain causal role R, and the property of having role R. On some accounts
of the connection between properties and laws, the relationship between
O and R is one of merely nomological necessity. Laws independent of O,
but governing its causal relations, give O the R-role. Nevertheless, the
thought runs, the instantiation of the property of having role R is explained
by the presence of O and the laws which hold relating to O. Instead of
metaphysical necessity, we have nomological necessity plus explanation.
This issue can be set aside by allowing that narrowly physical laws—those
identified by physics—can be part of the metaphysical necessitation-base for
a property. Thus we do have metaphysical necessitation still in play between
Os and laws on the one hand, and R on the other.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 89
With these assumptions in place, the argument against the efficacy of
those broadly physical properties recognized by non-reductive physicalism
runs as follows.
(1) A1(p1, p2, p3 . . . ) is causally sufficient for, or fixes the probability of,
A2(p100, p101, p102 . . . ) (necessitation-bases for, but not identical to,
bp1, bp2, respectively).
(2) bp1 is a cause of bp2 (Assumption).
(3) bp1 causes bp2 either directly or by causing A2(p100, p101, p102 . . . ).
(4) If bp1 causes bp2 directly, then either A1(p1, p2, p3 . . . ) is insufficient
for bp2 by causing A2(p100, p101, p102 . . . ) or bp1 is an overdetermin-
ing cause.
(5) If bp1 causes bp2 by causing A2(p100, p101, p102 . . . ), then the same
choice holds regarding A2(p100, p101, p102 . . . ).
(6) There is no systematic overdetermination in this way.
Therefore,
(7) bp1 is inefficacious (see, e.g., Kim 1998, 41–7; Kim 2005, 39–52).
The argument does not claim that if A1(p1, p2, p3 . . . ) is causally sufficient
for A2(p100, p101, p102 . . . ) then there can be no other sufficient cause
without overdetermination. There may be other sufficient causes which
are part of the causal circumstances, or further up or down the causal chain.
The focus is just on the efficacy of A1(p1, p2, p3 . . . ) and bp1 for target effects
A2(p100, p101, p102 . . . ) and bp2 standing in the same relationship. The
question is whether, at that point in the causal network, there is any
contribution for bp1 to make given A1(p1, p2, p3 . . . )’s presence.
Talk of position in a causal network may raise alarm bells because of
putative difficulties in fitting conditions, in particular negative conditions,
into the framework (Steward 1997, 135–40). It should not. The argument
does not require an exhaustive causal network. All that is required is that we
can make sense of the idea that token events, or property instances, stand in
a causal network against a backdrop of assumed causal conditions and that, as
a result of this, we can see two or more events, or property instances, as in
potential competition for efficacy at a certain position in this network.
Although I have dubbed this argument Kim’s exclusion argument, the
appeal to the causal exclusion principle is implicit. The principle holds that
90 PAUL NOORDHOF
No single event can have more than one sufficient cause occurring at any given
time—unless it is a genuine case of causal overdetermination (Kim 2005, 42).
(4) and (5) each claim that the choices are insufficiency of one of the
putative causes or overdetermination. This is what the causal exclusion
principle claims. I do not appeal to the causal exclusion principle explicitly
because it is inadequately formulated given the first point I made about what
the argument does not claim. At a given time, there may be two or more
sufficient causes each of which is sufficient, given causal circumstances that
include the other of the causes.
The argument involves a simplification relating to Jaegwon Kim’s dis-
tinction between supervening and micro-based properties. My appeal to
metaphysical necessitation does not distinguish between these two cases.
Nevertheless, Kim holds that the kind of argument I rehearse works against
the former and not the latter. Since the argument only seems to need to
appeal, at the crucial point, to the idea that A1(p1, p2, p3 . . . ) is sufficient for,
or fixes the probability of, bp2 by being sufficient for, or fixing the prob-
ability of, something which is sufficient for bp2, it is hard to see how to
justify the distinction between the cases (for more detailed discussion, see
Kim 1999; Noordhof 1999b, 2010).
The argument also works at a certain level of abstraction that may seem to
reduce its threat or make its application uncertain. Candidate BPs will include
those we attribute by attributing the belief that . . . where ‘ . . . ’ is filled in by
some specification of content, sensation of . . . where ‘ . . . ’ might be filled in by
‘burning feeling in the foot’, and so on. Thosewho put forward the argument,
and those who discuss it, oftenworkwith the standard picture that Ai(Pj, Pj + 1,
Pj + 2 . . . ) refers to some arrangement of narrowly physical properties in a
subject’s brain. It may well be plausible that the following is true:
If S has BP1 and BP2, and bp1 causes bp2 (where these are the particular instances
of the properties attributed to S), then there are some arrangements of narrowly
physical properties in S’s brain, say A1(p1, p2, p3 . . . ) and A2(p100, p101, p102 . . . ),
which are part of the metaphysical necessitation-bases of BP1 and BP2, and A1(p1,
p2, p3 . . . ) causes A2(p100, p101, p102 . . . ).
That is, corresponding to mental efficacy, there is related efficacy at the
narrowly physical level in the brain. However, it is no part of the argument
that this assumption is written in. All it needs is the idea that, however
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 91
extensive the metaphysical necessitation-base for mental properties needs to
be, putative causal relations between them imply corresponding causal
relations between these bases (where a metaphysical necessitation-base for
a property is one whose instantiation metaphysically necessitates the instan-
tiation of the property in question).
Two reactions to Kim’s argument are popular but I will argue are over-
reactions. The first says that, in fact, bp1 is efficacious because it stands in
different patterns of variation to bp2 than A1(p1, p2, p3 . . . ) as a result of
which it plays a distinctive causal explanatory role. I say that this relies on an
understanding of causation that is, at once, too strong and too weak. We
don’t have to adopt an account of causation with such counterintuitive
consequences (as we shall see) to have an answer to the exclusion argument.
We should not take distinct patterns of variation, and the inferential conse-
quences which flow from this, as fully capturing the reality of causation.
The second says that bp1 is efficacious because it is identical to A1(p1, p2,
p3 . . . ) even though BP1 is not identical to A1(P1, P2, P3 . . . )—we have a
property instance identity without an identity of properties. One version of
the latter proposal—attractive because it provides a prima facie answer to an
immediate objection—is formulated in terms of a trope metaphysics. Here
I will argue that instance identification is, in itself, questionable, inadequate
to support the whole weight of the response and leads one to a dubious
metaphysics. Most importantly, it conflates property causation with prop-
erty instance causation in its attempt to provide a defensible position.
I consider these responses in turn in the next two sections as preliminaries
to my own preferred approach, with, of course, nary a hint of over-reaction
to be found.
2. Different Patterns of Variation
The first line of response to the argument appeals to, in the limiting case,
different patterns of absence. For example, it is noted that the following
are true.
(PA1) If bp1 had not occurred, then no necessitation-base of bp1 would have
occurred.
Hence there would be no bp2.
92 PAUL NOORDHOF
(PA2) If A1(p1, p2, p3 . . . ) had not occurred, then another necessitation-base of
bp1 might have.
Hence, it is not the case that there would be no bp2 (e.g. List and Menzies
2009, 487–9; Menzies 2008, 210).
Truths such as this have been used in various contexts. Sometimes it is
said that causation is a contrastive matter. The basic form of causation is that
c rather than c’ causes e rather than e’. Different ways of describing what
might be thought to be one property instance (e.g., A1(p1, p2, p3 . . . )/bp1),
or identifications of distinct property instances, set up different comparison
classes. Contrastive accounts of causation typically take property instances
(or events) to be coarsely individuated—so that a property instance involves
the co-instantiation of multiple properties—because the case for more finely
individuated property instances (or events) can be answered if causation is
contrastive (e.g., Schaffer 2005, 347). Nevertheless, this is not mandatory.
Taking causation to be contrastive is often accompanied by the claim that
causal statements are context-sensitive. Context-sensitive statements
convey different propositions in different contexts of use. In the case of
causation, the context-sensitivity concerns what is the foil to the target
property instance or event. If there is variation in this—because, in some
contexts, the foil is absence of the target event, in others a specific alternative
event—then causal statements would be context-sensitive in this respect.
(PA1) and (PA2) take the foil to be the absence of a property instance
satisfying a particular description. Thus, describing a property instance as
bp1 determines the comparison class to be the absence of any property
instance correctly described as an instance of BP1. This will include lots of
other necessitation-bases of BP1. Whereas, describing a property instance as
A1(p1, p2, p3 . . . ) determines the comparison class to be the absence of any
property instance correctly described as an instance of A1(p1, p2, p3 . . . ) (see,
e.g., Menzies 2008, 206–8). By focusing simply on the case of absence, as
(PA1) and (PA2 do), we bracket the question of context-sensitivity. Never-
theless, the considerations offered below with regard to the case of absence
may be generalized.
Counterfactual theories of causation promise an immediate explanation
of the relevance of (PA1) and (PA2) though, as we shall see, this promise is
not kept. They are generally formulated as contrastive theories in which the
contrast is always with the absence of the target cause. Non-counterfactual
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 93
theories have to generate (PA1) and (PA2) either from taking their truth as a
constraint—their approach is geared to make such counterfactuals true—or
by writing in the contrastive component as an additional element, for
example, by holding that in a layered world of natural kinds ‘same level
causation is the norm’ (Gibbons 2006, 88, where the talk is of systematic
difference-making rather than explicitly of counterfactuals such as these
which express difference-making).
In any event, the claim is that the mental is shown to be efficacious, by
identifying the right difference-making as plausibly revealed in such coun-
terfactuals. The reasoning runs as follows. ‘If c were not the case, then
e would not be the case’ is a plausible sufficient condition for causation.
A1(p1, p2, p3 . . . ) fails to satisfy this condition if (PA2) is true. That is, if there
might be some other necessitation-base of bp1 so that bp2 may still be the
case. By itself, this doesn’t show that A1(p1, p2, p3 . . . ) fails to be a cause if
the counterfactual dependence of e upon c were merely a sufficient condi-
tion. So it seems that it is being taken as a necessary condition too, in the
circumstances.
The first thing to note is that we don’t allow that the possible occurrence
of replacements to a cause to discredit that cause from being efficacious, on
pain of making the world’s causal processes very gappy affairs. For example,
suppose my head of department comes to me and points out that I have done
very little administration for the department recently and other folks have
done lots of stuff. I don’t undermine what they have done by saying that,
since, if they hadn’t done it, I would have done it in their place, they cannot
be credited with having done anything. Yet, the situation seems analogous.
There were two, or doubtless more, possible undertakers of these adminis-
trative tasks. Undertaking these tasks was just realized in them rather than
me. Causes are those things which are actually involved in the process which
led to a certain target effect. Otherwise, at every point in the process at which
there might have been a replacement, we would have a causal gap. Of
course, you could decide to call the gap ‘a gap of causation’—a gap which
is filled by the occurrence of actual determination—but the decision to talk
this way has no particular utility and, as we shall shortly see, would not
alleviate the worries about mental causation in any case.
A second, and related point is that the counterfactual reasoning which is
meant to support bp1’s efficacy over A1(p1, p2, p3 . . . )’s claims is the same
reasoning that is judged inappropriate in all cases of redundant causation,
94 PAUL NOORDHOF
especially pre-emption. Recognizing the existence of pre-emptive caus-
ation precisely turns on supposing that the possible occurrence of replace-
ments does not undermine the pre-empting cause’s entitlement to be called
a cause. It is a significant cost to appeal to a pattern of reasoning which
would discredit all pre-emptive causes in order to discredit the causal claims
of A1(p1, p2, p3 . . . ), and yet this is what is required to answer the exclusion
argument without rejecting a version of the exclusion principle. Perhaps
Peter Menzies will say that it is appropriate to appeal to this pattern of
reasoning given that redundant causation has been tacitly ruled out in this
kind of case. However, it is hard to see this move as legitimate bearing in
mind that no characterization of the difference between this type of case and
redundant causation has been provided and the latter is very much an option
which is under consideration in discussions of this issue.
An unfortunate consequence of the patterns of variation approach is that it
makes the resolution of Kim’s argument turn upon brain plasticity, in our
terminology, the plasticity of arrangements of narrowly physical properties
supporting the causal relationship between mental properties. If it is the case
that no replacement arrangements of narrowly physical properties would
subserve the relationship, if the actual arrangement of physical properties
were absent, then we would be back with causal competition once more
with mental properties the potential losers. This may not, in fact, be an issue
because neuroscientists have observed that, as a result of damage, different
parts of the brain can be used to play the same function.Nevertheless, it would
be surprising if the efficacy of mental properties turned onwhether or not this
held on a case by case basis. Furthermore, since brain plasticity reduces with
age, this proposed response seems stuck with the potential consequence that
subjects’ mental properties may lose efficacy during the course of their lives.
The counterfactual reasoning with which I began this section has been
taken to express another feature of causes which, thereby, provides a
motivation for taking the previous points I’ve made to be inconclusive.
This is the idea that causes should be proportional to their effects and not
contain lots of redundant elements (Yablo 1992; Menzies 2008; List and
Menzies 2009, 488–9). This is alleged to be the difference between bp1 and
any of the A( . . . )s.
As things stand, this last claim is susceptible to a deflationary response. The
objector to the efficacy of bp1 can concede that talk of bp1 has a causal
implication that talk of A1(p1, p2, p3 . . . ) does not: bp1’s absence ensures the
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 95
absence of sufficient causes/chance-fixers, that is, any of the A( . . . )s; talk of a
particular A( . . . ) does not.Nevertheless, it can be argued, it is not that, by these
means, bp1 is revealed to be the cause itself. bp1 is not, by anybody’s lights, a
cause of any of its necessitation-bases, rather its absence entails the absence of
any of them. In brief, we have causal explanatory impact without causation.
The claim of proportionality is, plausibly, overstated in any case and,
thus, doesn’t get past the difficulty raised by the exclusion argument.
Considerations of proportionality entitle something to be counted a cause
in the following sense.
If c and c’ are putative competitor causes of e at the same point in the causal
network, and c is more proportional than c’ for e, then if c’ is a cause, c is a cause.
In brief, the reason for this is that more proportional causes are specified in
terms of properties which enable us to capture a generality that less propor-
tional causes miss. So if the latter is a cause, the former will be too. This will
become clearer on the development of my own approach in section 5. For
the moment I observe, first, that the attempts to discredit the efficacy of
A1(p1, p2, p3 . . . ) have not been successful and, second, there is no motiv-
ation for adopting a distinction between causing on the one hand, and causal
sufficiency for, or determination of, the probability of the target effect on
the other. Causing something may involve additional features than simply
being causally sufficient for, or a determinant of, the probability of the target
effect. Nevertheless, the latter is plausibly a necessary condition for the
former and, as a result, an exclusion argument run in terms of causal
sufficiency or determination would appear almost as damaging, if not as
damaging (for more discussion, see Noordhof 1999c, 374–5).
Thus, we are left with an apparent causal explanatory difference that we
must evaluate to see whether we have a corresponding difference in causal
reality. To conclude that difference in causal reality just falls out of the causal
explanatory difference identified is the first of the two over-reactions
I promised to identify.
3. Identity of Property Instances
An alternative fashionable approach to Kim’s argument is to argue that bp1 is
identical to A1(p1, p2, p3 . . . ). Although the properties, BP1 and A1(P1, P2,
96 PAUL NOORDHOF
P3 . . . ) are not identical, when it comes the instances—bp1 and A1(p1, p2,
p3 . . . )—they are. It cannot be denied that, if these property instances are
identical, then this particular problem is resolved. bp1 is a cause given
agreement that A1(p1, p2, p3 . . . ) is, and so on for all other instances of
broadly physical properties so identical. Evaluation of this approach doesn’t
focus on whether it works so much as whether, and indeed how, the
identification can be justified. On the positive side, there is the satisfactory
result for the problem of mental causation the exclusion argument raises.
But are there things to be said on the negative side? Identifications need to
be justified by more than the fact that they offer a convenient simplification
of our problems. They need to be independently plausible or, at least, not
implausible.
Some will take this challenge as unfair. They will remark that identity is a
primitive relation so one cannot expect any justification of it. To the extent
that we need a reason for recognizing the identity, resolution of the problem
of mental causation in this vicinity supplies us with one. However, this is
mistake. First, there is the slide from metaphysics to epistemology. Identity
may be a primitive relation but that doesn’t mean that justification for
supposing it to hold must be taken to be primitive. We may have complex
reasons to believe simple things. Second, the combination of views pro-
posed is that two properties may be distinct yet have identical instances. We
need an account of why this combination is coherent. Third, within a
metaphysical framework which makes this combination of views possible,
the considerations in favour of taking instances of mental properties to be
identical with instances of arrangements of physical properties must have
general application. We can’t have mental property instances as a special
case. That would be unmotivated.
It is no surprise, then, that sophisticated proponents of this strategy
address these issues. It is convenient to divide the approaches into those
which take properties as universals to be the fundamental element and those
which take property instances or tropes to be the fundamental element.
I shall consider these in turn.
The apparent problem for the first approach—which takes properties as
universals to be the fundamental element—is how one instance could involve
the instantiation of two distinct fundamental elements. Instance identity and
distinctness, it would seem, must follow universal identity and distinctness
(Ehring 1997, 462–3). The following makes the connection explicit.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 97
An instance of F is identical to an instance of G only if F = G (where ‘F’, ‘G’ are
universals).
A sufficient condition for instance identity will draw upon additional factors
that serve to distinguish between instances, e.g. spatiotemporal location.
This problem seems overstated. The, by now, standard, subset, approach
to property co-instantiation seems available to those who take properties to
be universals. The subset view of property instance identity holds that bp1 is
identical with A1(p1, p2, p3 . . . ) if the causal powers of BP1 are a subset of the
causal powers of A1(P1, P2, P3 . . . ). A property instance with a set of causal
powers {CP1, CP2, CP3, CP4 . . . } will count as a property instance of a
property with that set, and also of a property with, say, {CP1, CP2 . . . }
alone (Whittle 2007, 68–9, who doubts that the subset view can be used in
the straightforward fashion recommended here). The apparent distinctness
of the instantiation of F and the instantiation of G is shown to be mistaken
because the causal powers of the former are a subset of the causal powers of
the latter. Co-instantiation as partial coincidence in causal powers is not
ruled out to those with universals in their ontology.
Some prefer to say that bp1 is realized by, but is not identical to, A1(p1, p2,
p3 . . . ) when the causal powers of the former are a subset of the causal
powers of the latter (Shoemaker 2007, 17; for other grounds for resisting
identity, see ibid. 48–9). They rightly point out that instances cannot be
identical if the causal powers of one stands in the subset relation to the causal
powers of the other (or, with qualifications, more precisely, causal profile
which includes the ways in which an instantiation may be caused too, as
well as its causal powers, see Shoemaker 2007, 11–12, 16–17). Their strategy
is not strictly speaking an example of the identity of instance strategy;
however, it resembles it in important respects. They take it that there is a
state of affairs A1(p1, p2, p3 . . . ) which realizes bp1, by having its causal
powers as a subset, and bp1 is efficacious when the subset of causal powers
relating to it are in play. Talk of states of affairs allows them to resist
characterizing A1(p1, p2, p3 . . . ) itself as a property instance—as opposed
to an arrangement of property instances—while retaining the subset picture.
They can, of course, allow that there is a property of being a certain kind of
state of affairs if they wish (e.g., Shoemaker 2007, 32–4). The difference
between instance-identity and realization just mentioned brings out the
slipperiness of the term ‘co-instantiation’. It can either mean identity of
98 PAUL NOORDHOF
instances or coincidence of instances in a particular instantiation. My
remarks below apply equally to both characterizations.
The basic problem is that the subset view is not true if you take the
existence of BP properties seriously. To fix ideas, consider a case of pain.
A part of me hurts and my experience of it—my pain experience—is of a
type that causes writhing and general unhappiness in humans and, we may
presume, similar writhing and unhappiness in sentient robots. The hallmark
of this activity is that the writhing is because of heat damage done to an arm.
So the writhing stems from there and is, in some way, directed towards
alleviating what is going on there.
In humans, this property is necessitated by particular kind of A@-fibre or
C-fibre firing (let us suppose), which kind depending upon the type of
hurting involved. Let’s focus on the famous C-fibre firing and take the
particular hurting to be necessitated by C-fibre firing in way W. In robots,
the pain will be necessitated by something different, let’s call that C-circuit
activity in way V. Now the question is whether the causal powers of pain
are a subset of C-fibre firings’ causal powers and of C-circuit activity’s causal
powers. The answer seems to be no. Pain experiences have the capacity to
cause pain behaviour in humans and robots whereas C-fibre firing in way
W can only cause such pain-behaviour in humans, and C-circuit activity in
way V can only cause such pain-behaviour in robots. So, the subset view
would deny that instances of pain experience of the kind specified are
identical with, or for that matter realized by, either.
One move would be to claim that pain-experiences-in-humans and pain-
experiences-in-robots, rather than simply pain experiences, are to be identi-
fied with C-fibre firings and C-circuit activations. This was Kim’s proposed
response to the existence of variable realization in defence of reductive
physicalism (i.e., type-type identity theory) (Kim 1992, 330–5). It rested
upon the claim that the nomic relationships, and hence causal powers, of
instances of mental properties or states are to be explained in terms of the
nomic relationships, and causal powers, of narrowly physical properties (Kim
1992, 322). As we shall see later, this claim is susceptible to a number of
different interpretations. Its use in this context, though, is questionable.
Denying that there are pain experiences, as opposed to species-specific
pain experiences, is, quite obviously, refusing to take them seriously.
The existence of a cross-species psychology reflecting general truths
about the causal implications of having pain experiences requires more
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 99
than merely species-specific pain states. Information about one type of
species-specific pain experience would, if pain experiences failed to exist,
imply nothing about the role of pain experience in other creatures. Yet,
when we reflect upon how a creature would respond if something hurt like
this—thinking about a particular kind of pain experience—we think that
there are general psychological commonalities between how they would
respond, and how we would respond, independent of variation of physical
constitution. There may be differences too, as a result of our differences in
physical constitution, but recognizing that there may be psychological
differences is quite compatible with also recognizing commonalities. This
is a phenomenon with which we are familiar for individuals too. Recogniz-
ing individual differences does not imply that all that is possible are individ-
ual psychologies and individual-relative states.
Let me state a bit more precisely how this might work in the face of
Jaegwon Kim’s scepticism (e.g., Kim 1992, 323–5). Suppose that BP1 is
metaphysically necessitated by each of the following arrangements of nar-
rowly physical properties A1(P1, P2, P3 . . . ), A1(P11, P21, P31 . . . ), A1(P12,
P22, P32 . . . ) . . . Then, BP1 can be related to radically disjunctive narrowly
physical conditions—A1(P1, P2, P3 . . . ), A1(P11, P21, P31 . . . ), A1(P12, P22,
P32 . . . ) . . .—in that the corresponding causal powers of these conditions
are {CP1, CP2, CP3, CP4, CP5, CP6 . . . }, {CP1, CP2, CP31, CP41, CP51,CP61 . . . }, {CP1, CP2, CP32, CP42, CP52, CP62 . . . }. The vast majority of
the causal powers are disparate.
According to the subset approach, instances of A1(P1, P2, P3 . . . ), A1(P11,
P21, P31 . . . ), A1(P12, P22, P32 . . . ) . . . are each instances of BP1 because a
subset of their causal powers are {CP1, CP2}, the causal powers of BP1.
Some of the other causal powers listed may be responsible for psychological
differences but there is a significant psychological commonality. The prob-
lem I’m raising for this approach with regard to the case of c-fibre and
c-circuit activity is that the arrangements of narrowly physical properties
that we find it plausible to count as instances, or realizations, of mental
properties do not seem to have the causal powers that would constitute the
commonality. That is, {CP1, CP2} are not part of the sets of powers,
contrary to how they have been represented above.
A second move in response to this case may seem to help with the
difficulty just identified: the distinction between core realization and total
realization. The core realization of a BP is that arrangement of narrowly
100 PAUL NOORDHOF
physical properties which play the causal role associated with BP’s instanti-
ation. The total realization is the arrangement of narrowly physical proper-
ties together with the context in which they occur which, taken together,
necessitate that an instance of BP is presently typically playing the role in
question. As Sydney Shoemaker puts it, to motivate the distinction, firing
C-fibres in a Petri dish is not a case of pain (Shoemaker 2007, 21; the
distinction goes back to Shoemaker 1981). For mental properties to be
instantiated, it seems plausible that not only must properties with a certain
causal profile be instantiated but, in addition, key elements of the profile
must be typically manifested.
With this distinction in place, it might be argued that, if C-fibre firing in
way W occurs in a robot, then it fails to cause writhing because the total
realization of the relevant experience of pain is not present. So an instance of
BP does not display a causal power that an instance of C-fibre firing in way
W fails to have. This response is mistaken. For BP to have a causal power
A1(P1, P2, P3 . . . ) fails to have, BP does not have to display that causal power
in circumstances in which A1(P1, P2, P3 . . . ) is present and does not display
it. BP can display the causal power when it is realized in a different way and
causes something that A1(P1, P2, P3 . . . ) would not in those circumstances.
Appeal to the idea of total realization explains how individual differences
are compatible with cross-species psychological laws. The total realizations
of mental properties ensure that certain causal relations typically hold across
differences of constitution. Compatible with this, other differences in
constitution may result in differences of psychology which disrupt these
typical causal relations. Psychological laws may have written into them
conditions under the typical causal relations don’t hold and these conditions
may imply that the regularities do not hold at all in some other species. This
is no more exceptionable than our appreciation that, for example, if we
were more secure, a rejection would have less of a significant effect than it
does in our case.
The distinction between core and total realization raises a question mark
over the efficacy of instances of BP. If the conditions under which BP is
necessitated include the circumstances, then can it be attributed causal
powers with regard to those circumstances? An answer to this will come
in the development of my own proposal in section 5. However, in brief, the
response is that BP is efficacious in virtue of the fact that a part of its
realization is efficacious.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 101
A third move in support of the subset approach to mental causation is to
claim that the causal powers of an instance of pain experience are no more
than that of the property with which it is co-instantiated even though the
property of being a pain experience has causal powers which exceed those
properties with which it is co-instantiated. Two ways in which this might
be achieved are either indirectly from a claim about the individuation of
instances or directly from a claim about what is required for instantiation.
The indirect method would be to say that, suppose that a particular pain
experience is co-instantiated with an instance of C-fibres firing in way W,
then that instance of pain experience could not be co-instantiated with
something else. There is no possible world in which, say, my instance of
pain experience could be co-instantiated with C-circuit activity in way
V. So, this particular instance of pain experience cannot have causal powers
which exceed those of that with which it is co-instantiated.
Obviously, if the instance of pain experience is identical with an instance
of C-fibre firing in way W, then it is plausible that they share modal
properties. However, it would be illegitimate to appeal to instance identity
by itself to establish that the causal powers of one of the instances don’t
outstrip those of the other. That’s supposed to be the conclusion of the
subset approach. Instance identity is not meant to establish the correctness of
the subset approach. Furthermore, an actual future case, rather than appeal
to a possible case, might put this under pressure. Suppose that I have a
throbbing pain and I receive prosthetic neural fibre replacement without
anaesthetic in the hope of stopping the pain and the pain persists. It is very
plausible to say that that instance of pain experience—and not just a pain
experience of the same type—is continuing although realized by different
neural fibres.
It could be argued that the instances of prosthetic C-fibres have the same
causal powers as the instance of pain experience in the case described. So
they present no problem. By contrast, the instance of my pain experience
could not be co-instantiated with C-circuit activity because I could not be a
robot. The object in which a property is instantiated is, the claim would
run, one of the essential features of an instantiation.
The success of this response partly turns on whether it is plausible to insist
that I could not have been constituted from the same material as a robot. If
I am essentially a particular human animal, then the response receives
support. Those friendly to a psychological characterization of personal
102 PAUL NOORDHOF
identity will resist the claim that I could not be made from the same material
as a robot and even those who insist the psychological view is false, and that
we are animals, don’t have to conclude that we are animals essentially
(Olson 1997, 125, for the claim I could not become a robot; for the claim
that I am an animal is compatible with my becoming a robot, see Olson
2007, 27). If I might be a robot, then it is possible that my life could have run
a different course so that my pain experience now, in fact co-instantiated
with C-fibre firing in way W, could have been co-instantiated with
C-circuit activity firing in way V. In which case, the causal powers of the
instance of pain experience threaten to outstrip that which is co-instantiated
with it. So the instance of my pain experience is not identical with instances
of either C-fibre firing or C-circuit activity.
Nor is this the extent of the problems the response faces. Certain cases of
causation seem to involve the transfer of a property instance from one object
to another. For example, there is a difference between an object being sticky
on one side, it ceasing to be, and another object being sticky on one side,
and the stickiness of the first object wholly transferring to the second object
by contact. In the latter case, it is plausible to say that the same instance of
the property of being sticky has moved from one object to the other. In
which case, the object in which the property is instantiated cannot be
essential to the identity of the instance (Ehring 1997, 123–4). It is usually
thought that only a trope metaphysics can account for this kind of fact but
this does not seem correct.
The reasoning goes like this. Exemplifications are essentially momentary.
For suppose otherwise, then we should allow the same in the case of spatial
location. If an exemplification of squareness on one side of my office is not
distinct from an exemplification of squareness on the other side of my office,
then we have an exemplification which is wholly present in two places. We
have lost the distinction between exemplifications and universals. We
would do the same if we recognized non-momentary exemplifications
(Ehring 1997, 87–9).
This line of reasoning can be resisted. Exemplifications do not have to be
momentary. They can have duration. Thus if an object remains square over
a period of two hours, 11 a.m. to 1 p.m., there is a single exemplification of
the universal of squareness. What should we say about the object’s square-
ness at noon? Rather than recognize momentary exemplifications of square-
ness which, then, have to constitute the exemplification of squareness over
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 103
two hours, we should deny that there is an exemplification of squareness at 12
noon. There is an exemplification of squareness between 12 and 2 in virtue
of which it is true that the object is square at 12 noon. Corresponding to
every timely predicate, there does not have to be an exemplification.
The same applies to spatial extent. An expanse of red involves a single
exemplification of red. We do not have to suppose that exemplifications of
red at much smaller regions make up this single exemplification otherwise,
amongst other difficulties, we would need to consider the smallest extent in
which red could be realized and question whether that smallest extent could
be described as red at various points within it. If it is allowed that it can be so
described without there being an exemplification of red at the points with
the smallest extent, then we might as well accept that there is a single
exemplification of red across the extent with it still being true that portions
of that single exemplification are also described as red.
The difference between what it is plausible to say in this case, and what
I said previously concerning two spatially separated exemplifications of
squareness, derives from the fact that the latter property has defined bound-
aries. So an exemplification of squareness falling outside the boundaries
cannot constitute the same exemplification of squareness. Whereas, in the
case of an instance of redness, the extent is not fixed. So we don’t have to
concede that a lesser extent must also count as an exemplification of redness.
The claim that that lesser extent is red can be true simply in virtue of the
larger extent which is an exemplification of redness.
Since exemplifications of properties are not individuated by the objects
which possess them, there is nothing to rule out an instantiation of pain in
me being instantiated in another creature with a different constitution. Even
if they were so individuated, it is an additional step to hold that such
individuation of exemplifications requires that the means of individuation
is in terms of essential properties. We individuate objects by their spatio-
temporal position. From that, it does not follow that their actual spatio-
temporal position is essential to them.
The second version of the third move I identified in support of the subset
approach was to tinker with what is required for instantiation. So it may be
suggested that A1(p1, p2, p3 . . . ) is identical to bp1 by having truncated causal
powers of BP1. This will still not take the existence of BP1 seriously unless
you also allow that the same would hold for A1(P1, P2, P3 . . . ). Thus, an
instance of the latter may be identical to A1(p1, p3, p4 . . . ) because, whatever
104 PAUL NOORDHOF
causal powers it loses from not having p2 as part of the arrangement are not
sufficient to undermine the instance identity. You can’t resist this by
claiming that p2 must be part of the instantiation of A1(P1, P2, P3 . . . )
because instance identity is meant to be determined by the subset of causal
powers relation—or now the truncated subset relation—and not what
properties constitute the property to be instantiated. After all, if the latter
were in play, we would have grounds for denying that instances of pain
were identical with instances of arrangements of narrowly physical proper-
ties and the argument would centre around providing a justification for
thinking that narrowly physical properties were constituents of pain. How-
ever, since it is unacceptable to hold that A1(p1, p3, p4 . . . ) is an instance of
A1(P1, P2, P3 . . . )—otherwise how would be distinguish between instances
of A1(P1, P2, P3 . . . ) and instances of A1(P1, P3, P4 . . . )—this must be
because having a truncated set of causal powers is not enough.
The second problem with the truncated powers proposal is that our
evidence about bp1 in us would provide no grounds for supposing that pain
in robots would make silicon creatures writhe. The only causal powers that
my pain would reveal to me would concern how it affected me. I would not
be able reasonably to assert that that pain—in me—is so bad that, if it were
instantiated in a robot, they would be writhing about too unless they had
much greater powers of pain control to me and could focus their attention
away from it. I’d have no idea at all—by the truncated powers view—what
powers pain would have in robots. But this not correct. The reason why we
know how others would behave if they had pain—even if they had a different
constitution—is that we know the causal powers that pain would have in
them from our own case (Gibbons 2006, 95–7, also emphasizes this).
Our starting position was that there were two distinct properties under-
stood to be universals—being a pain experience of a certain type and being
c-fibre firing in way W—which, therefore, would naturally be thought to
have distinct instances. We were looking for a justification for concluding
that some of their instances are identical. The subset proposal fails to provide
it. This may not be altogether surprising. Proponents of trope metaphysics
take the situation to be different within their framework. In fact, many of
the points travel across and trope metaphysics has problems of its own with
providing what is needed.
Proponents of a trope metaphysics take property instances as fundamental
and construct physical and mental properties from these elements. This
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 105
gives them a motivated way of avoiding the question of how distinct
properties could be identical in instances because this will be built into the
construction. In addition, by insisting that property instances are fundamen-
tal, they have a way of resisting the claim that they have a structure which
will raise, once more, the question of whether it is the mental or physical
component of the instance which is causally relevant. Thus, a response to
Kim’s argument would then be a full response to the question of causal
relevance of properties. A further question cannot be raised about in virtue
of what features of an instance, is that instance efficacious. Unfortunately, a
substantial motivation in favour of their metaphysics tells against this solu-
tion to the problem of mental causation.
Trope metaphysicians construct properties from exact resemblance classes
of tropes. This only works if two properties don’t share the same instance. If
two properties do share the same instance, and are not coextensive, then we
cannot appeal to exact resemblance. We must appeal to rough resemblance
and, indeed, that is what those who hold that mental properties share
property instances with physical properties do.
Just in case, the issue isn’t obvious, let me explain why they need to move
to rough resemblance. Let m1, m2, m3 . . . mn be a particular class of mental
tropes (e.g., each of which is a pain experience) and p1, p2, p3 . . . pn be a class
of physical tropes. Remember that the physical class and the mental class
cannot be coextensive because, according to non-reductive physicalism,
mental properties are not identical to physical properties. Let p3 = m3. Then
if the classes of M and P were constructed from exact resemblance, then m3
would exactly resemble all the ms and exactly resemble all the ps too. But in
that case, each of the ps would exactly resemble each of the ms (because
exact resemblance is transitive) and we would just have one class after all
(Gibb 2004, 471–2). Or, put it another way, m1 and m2 must exactly
resemble each other. But if they are identical to different physical properties
(that is, if they are variable realized), then those physical properties cannot
exactly resemble each other. Again, to be able to construct a mental
property class, trope metaphysicians need to appeal to rough resemblance.
Members of a class roughly resemble each other to a certain degree which is
greater than any non-member.
The appeal to rough, rather than exact, resemblance undermines the
motivation for trope metaphysics in the first place. A principal reason for
adopting a trope metaphysics, rather than resemblance nominalism, is
106 PAUL NOORDHOF
because of the problem of imperfect community (Campbell 1990, 32–4, 72–3).
The difficulty identified for resemblance nominalism under this heading is
that if you try to construct universals from resemblance classes of objects,
then you will be committed to surrogates for universals which are not
united in a resemblance but rather may be united by different resemblances
between objects in the specified class. Moreover, it is unclear that even
some of these classes will be surrogates for universals because a class bound
together by a resemblance may be legitimately made a more extensive class
in virtue of other resemblances if all that is required is rough resemblance
(Manley 2002, 77–9, who raises the latter difficulty for worlds with limited
members but it is not clear that more populous worlds avoid the difficulty).
Trope metaphysicians claimed to avoid the problem of imperfect com-
munity by appealing to the notion of exact resemblance as opposed, simply,
to resemblance. Appeal to exact resemblance only works, though, for the
trope metaphysician, if each property instance is a property instance of only
one property (Campbell 1981, 134–5; Campbell 1990, 66, 72–3; abandoning
his Campbell 1981 position on p. 137). Once you allow that a property
instance may be a property instance of two or more properties the problem
reasserts itself.
The result is that proponents to the trope metaphysics solution to mental
causation face a trilemma depending upon whether they appeal to exact
resemblance, rough resemblance, or resemblance in a certain respect. If they
appeal to exact resemblance to construct classes of properties, then they
must either concede that there are no mental properties (in which case, non-
reductive physicalism is false) or that mental property instances are distinct
from physical property instances (in which case, they have no solution to the
problem of mental causation). That is, either they have no solution to the
problem or the doctrine for which they sought to provide a solution cannot
be formulated.
On the other hand, if they appeal to rough resemblance, or resemblance in
a certain respect, they face one of two difficulties. If the appeal is to rough
resemblance, then they face the problem of imperfect community and under-
mine the motivation for adoption of a trope metaphysics. That is not to
presume that the problem of imperfect community cannot be resolved.
Perhaps it can. Indeed, Resemblance Nominalists, or friends on their behalf,
have suggested solutions which we do not have to (and don’t have the space
to) evaluate here (e.g., Hirsch 1993, 58–9; Rodriguez-Pereyra 2002, Ch. 9).
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 107
The point is that unless it turns out these solutions work only for tropes—
which is unlikely—there is no particular reason for adopting a trope meta-
physics rather than resemblance nominalism.
I guess it can be argued that, if a solution were available, then the appeal
of resolving the problem of mental causation may make trope metaphysics
independently attractive. A lot would turn on whether causation by
events—say—rather than causation by property instances—preserves less
of our intuitions concerning whether the mental is efficacious. I have
argued elsewhere that proponents of the trope solution have to make
versions of the same moves that they criticize in those who claim that
mental causation just involves events and not properties (or their instances).
They have to suggest that apparently intuitive claims about efficacy reveal
something, instead, about the pragmatics of explanation (Robb 1997;
Noordhof 1998, 225–6). In section 5, I will explain how the claims have a
plausible ontological basis instead.
If trope metaphysicians appeal to resemblance in a certain respect—
perhaps even exact resemblance in a certain respect—then they needn’t
face the problem of imperfect community but the respects (be they mental
or physical) allow the problem of mental causation to be raised once more.
It can be legitimately asked, are mental property instances efficacious in
virtue of their mental respect or their physical respect? The trope metaphys-
ician cannot deny this structure because they have appealed to it to resolve
the problem of imperfect community (Gibb 2004, 473–5).
We must conclude that proponents of the trope solution fail to establish
that a successful response to Kim’s argument constitutes a successful defence
of the claim that mental properties are causally relevant. Nor does their
proposal sidestep the problem I raised with regard to the application of the
subset approach. It indicates that we need to recognize the existence of
mental tropes in addition to narrowly physical tropes (and their arrange-
ment) to capture the additional causal powers that mental properties possess.
In this section, I have examined how a certain ontological response to
Kim’s argument—identity of property instances—leads to distortion of
what we should say about the causal powers of mental property instances,
implausible theses concerning instance identity, or abandonment of the
advantages of trope metaphysics. This is the second of the two over-
reactions I identified. Before I turn to my own approach, let me briefly
discuss an approach which self-consciously does not take BP seriously.
108 PAUL NOORDHOF
4. Challenge from the Unilevellers
Unilevellers deny that the world is layered, the idea that there are different
levels of properties. One unileveller position would be to adopt a trope
metaphysics which takes properties to be fully determinate tropes in exact
resemblance classes and concludes that, for example, while there are con-
cepts of mental properties, and their instances, there are, in fact, no such
properties or instances of them. Instead, ourmental concepts capture families
of similar tropes (see, e.g., Heil 2003, 140–3, 153; Heil’s modes differ from
tropes in that the objects they characterize are essential to their identity).
If there are no mental properties, then they cannot be causally relevant.
Nevertheless, unilevellers suppose that our mental discourse picks out
something which is causally relevant. Take the case of pain experience of
a particular type. The picture is captured by a negative and positive claim.
First, the negative:
There is no single respect R (exact resemblance in some way) in virtue of which
{A1(p1, p2, p3 . . . ), A1 (p11, p12, p13 . . . ) . . . } . . . } are all members of the class of
property instances that fall under my concept of a particular type of pain
experience (hereafter, pain experience instances) (cf. Heil 2003, 153).
Second, the positive claim:
‘A particular instance of pain experience caused a particular instance of writhing’
is true in virtue of there being some member of the set of pain experience
instances, say a particular instance of C-fibre firing in way W, which caused
some member of the set of writhing instances.
Our talk of pain experience of a certain type is causally relevant because,
although there are no such properties, our concept of it has conditions of
application which pick out a class of property instances one of which was
efficacious in the circumstances.
I have two related objections to this position. The first is that it turns
inferences about how creatures behave as a result of being in pain—or being
in other mental states—into relatively shaky inferences. We are inclined to
assert that if a pain experience of a certain type were instantiated in a silicon
creature (a sentient robot), it would writhe. But exactly how it would
behave is open to question if the chain of resemblances that bundle all
pain experiences together allow for significant differences. Perhaps my pain
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 109
experience and a silicon creature’s pain experience resemble each other in
being distracting, but not in giving rise to certain bodily responses. Identical
causal powers are not guaranteed across subjects.
Unilevellers may dissipate this initial worry by insisting that the required
amount of similarity will include the robot version of writhing in circum-
stances in which c-circuit activity occurs. This is not guaranteed because,
once you go approximate, there is always the possibility that there will be
sufficient similarity without this element. However, let me for the sake of
argument concede it.
That will still not deal with the particular case. Suppose I am currently
having a pain experience. Then I can reasonably think, if my friend Robbie
the Robot was experiencing this, he would be writhing about in agony.
However, in thinking about this pain experience, I am thinking about
(according to the unileveller picture) C-fibre firing in way W. This
C-fibre firing would not cause Robbie to writhe about in agony. Unile-
vellers deny that there is any mental property apart from this upon which
my belief may be grounded. So they are committed to holding that I have
no grounds for the belief in question.
It might be argued that this upshot is intuitive. Humans do respond to
pain in different ways. Different creatures are likely to do so even more. But
these observations are compatible with shared causal profile (as I noted
before). The causal profile of a property will, in different contexts, manifest
itself in different ways. The unileveller position is more radical than this.
According to them, there is no shared causal profile—at best, just an
approximate similarity in causal profiles of different property instances.
This brings me tomy second objection. As we shall see in the next section,
what is required for the causal relevance of mental properties is not simply
the efficacy of an instance of a mental property (nor for that matter the
efficacy of an instance of something picked out by a mental concept) but a
condition-relative general relationship between the instantiation of mental
properties and their target effects. By denying that such properties exist,
unilevellers give up on this requirement. As such, this is a point against them.
The unileveller position derives much of its motivation from an appeal to
truthmaking. The basic idea is that if A1(p1, p2, p3 . . . ) putatively metaphys-
ically necessitates bp1, then it counts as the relevant part of a truthmaker of
sentences with terms putatively referring to bp1. There is no need for bp1 to
exist. There is, however, another dimension which A1(p1, p2, p3 . . . ) seems
110 PAUL NOORDHOF
less well suited to provide. That is, the generality that we associate with bp1expressed in the inferences we are inclined to make. The precise nature of
those will occupy us in the next section of this paper. But, to summarize the
concern in a phrase, unilevellers mistakenly emphasize truthmaking at the
expense of inference-basing.
5. Property Causation
Kim’s original challenge to non-reductive physicalismwas that itmademental
property instances inefficacious. An answer to his argument is a necessary
condition for a defence of non-reductive physicalism against the charge of
epiphenomenalism but it is not sufficient. In addition, we need an explanation
of how mental property instances are efficacious in virtue of being mental
property instances. The appeal to different patterns of variation seemed to
draw on material that might be helpful in this regard but at the expense of
neglecting the detail needed to defend the claim that mental property
instances were efficacious. The instance identity strategy was more focused
on the latter but the problems with the subset view started raising issues about
the efficacy of the mental instances qua being mental, which were revealed in
cross-subject judgements about the effects of a particular pain instance.
My proposal is an attempt to satisfy both requirements. It focuses both on
what is required for a particular instance of a mental property, or indeed any
broadly physical property to be efficacious, and also on the element of
generality that shows that the instance is efficacious in virtue of being a
mental property. It runs as follows.
F is a property cause of G if and only if
Particularity: part of the (minimal) necessitation-base for the instance of F causes
part of the (minimal) necessitation-base for the instance of G.
Generality: (part of ) each (minimal) necessitation-base of F is such that all its
instantiations would cause (or in the case of indeterminism, raise the probability
of ) an instantiation of one of the (minimal) necessitation-bases of G if they were
in some causal circumstances C—where C may vary for each kind of necessita-
tion-base.
Let me comment on various elements of this proposal.
First, the appeal to necessitation-base is meant to be understood in terms of
metaphysical necessity—just as with the characterization of non-reductive
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 111
physicalism. The insistence on a minimal necessitation-base is addressed to
the following difficulty. If T is the necessitation-base for F, then so is T plus
the whole world apart from T. But we wouldn’t want to conclude that F is
efficacious because of the efficacy of some feature of the world unrelated to
T (or, indeed, some feature of T unrelated to the minimal necessitation-base
for F if T is also not the minimal necessitation-base for F). I have character-
ized the minimal supervenience-base in a previous publication; talk of
minimal necessitation-base just focuses on the key element (see Noordhof
1999a, 307).
The intuitive idea is straightforward. The minimal necessitation-base for
F is all that needs to be instantiated for a particular way of instantiating F. It is
meant to capture, in some sense, the thought that broadly physical proper-
ties are constituted from arrangements of narrowly physical ones, although
these ways may vary. We know that a certain understanding of property
constitution cannot be right. Complex universals (if they exist) cannot be
composed from other universals. The classic example to illustrate this is
Lewis’ case of methane (Lewis 1986). It is not composed of four hydrogen
universals and one carbon universal because there are not four hydrogen
universals. We can’t understand property constitution simply in terms
of property instance constitution either. Instances of methane may be
composed from one instance of the property of carbon and four instances
of the property of hydrogen but even this does not work for variably
realized properties. A single universal cannot be constituted in various
ways, even if its instances can be. So variably realized universals can’t be
said to have other properties as constituents.
For properties, rather than property instances, it is better to appeal straight
to the idea of minimal metaphysical necessitation—if it can be defended
against counterexamples as I urge. When it holds, it seems to follow that
there is nothing over and above arrangements of instances of properties
needed to constitute an instance of the target broadly physical property.
If there were something over and above arrangements of instances of
these properties, then there would always be the possibility that the add-
itional element could fail to occur, even if it wouldn’t (given the physico-
psychological laws).
Of course, particular analyses of minimal necessitation-base—and the
background idea of property instance constitution—may fail. But since
the idea is natural and, more importantly, does not implicitly draw on
112 PAUL NOORDHOF
claims concerning the efficacy of the target broadly physical properties, we
could safely take it as a primitive without concern that it vitiates the
substance of the account of property causation.
Second, the appeal to ‘part of ’ is to allow for the possibility that some-
thing may count as efficacious only in virtue of an element of it being
efficacious. The fire burned because of the presence of air in virtue of the
fact that oxygen is part of air. I shall discuss this no further here but it is
relevant to the issue of the efficacy of mental properties for which extern-
alism is true (see Segal and Sober 1991; Noordhof 1999a).
So much for a preliminary understanding of the first element. Let me
now turn to the second element, that of generality: the every minimal
necessitation-base clause. Of course, I am not the first to recognize the
implicit generality. Anybody who has offered an account of causal relevance
in terms of law has also done so (e.g., Fodor 1989; Segal and Sober 1991, 15).
However, first, my proposal does not appeal to law because it is question-
able whether there is a law if the pattern I have identified holds, and, second,
those who offered such an account often failed to appeal to the idea of
minimal necessitation too. Yet an appeal to the latter is also required.
Appeals to law by themselves struggle to explain whether correlation
between broadly physical properties reveals that their instances are standing
in a causal relationship. Broadly physical properties which are nomically,
but not metaphysically, necessitated by arrangements of narrowly physical
properties will have a true general statement concerning their co-
occurrence even if the broadly physical properties are intuitively ineffi-
cacious (Segal and Sober 1991, 4–5). So something extra is needed. Either
this can be part of the conditions under which the generality would count as
a causal law, or it can be characterized independently. That these conditions
are needed is not in dispute.
Turning to the first point, my condition bears most resemblance to an
account which appeals to a ceteris paribus law relating F and G to capture
the generality involved in causal relevance. A preliminary analysis of ceteris
paribus laws is that there is a ceteris paribus law relating F and G, if and only
if, for all R, where R realizes F, there are some conditions C, such that,
whenever R & C, then G and it is nomologically possible that R without
C (Fodor 1991, 23–4; Schiffer 1991, 6–7). If the second condition were not
met, then the law would be strict. The possibility of R without C provides
conditions in which the correlation between F and G fails.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 113
An objection to this analysis is that ceteris paribus laws have what Fodor
has dubbed ‘absolute exceptions’: realizations for F for which there are no
circumstances C which, together with the realization, are sufficient for an
instance of G. One way in which conditions may be unequal is if F is
realized by a dud. Fodor accommodates this by allowing that F can figure in
a ceteris paribus law if most of the time, it is not realized by duds for G, and
for other properties, say H, with which it also stands in a ceteris paribus law,
the dud realization does have circumstances in which it yields an instance of
G (Fodor 1991, 27–8). Others respond to this objection by denying the
existence of ceteris paribus laws (Schiffer 1991).
Whichever way one goes, the characterization of my generality condition
does not, then, involve an appeal to laws. However, its motivation remains
intact. If two properties are co-instantiated, then the effects of this instanti-
ation may be due to one or the other of the properties. One famous
illustration is the soprano’s singing of ‘my love’, at a certain pitch and
loudness, causing the glass to crack. It is plausible that the soprano’s singing
is an instance of that pitch, that loudness, and those words. Yet we would
not conclude that the glass cracking occurred in virtue of those words. So
how should we differentiate?
According to the generality condition, the property of involving the
words ‘my love’ does not serve to explain the pattern of causal relations
concerning glass crackings, taking into account different ways in which the
property of involving the words ‘my love’ may be realized. If the generality
condition holds for a certain property for a target effect, then we have such
an explanation. The causal relevance of a property, and not just one of its
instances, is hard to deny if, for every type of minimal necessitation-base of a
property, there are circumstances in which an instance of that property
always causes the target effect.
Consider the property of being rickety. If something is rickety, it is likely
to collapse. However, whether or not something will collapse depends
upon the precise circumstances in which it is located and the precise way
in which the property of being rickety is realized. If for every way of being
rickety, there are circumstances in which collapse follows, then we can say
that a collapse followed in virtue of being rickety. Otherwise, talk of
ricketiness, at best, figures in a ceteris paribus law. If it does not meet the
generality condition, then, while various ways of being rickety may be
causally relevant, ricketiness is not.
114 PAUL NOORDHOF
Closer to our interests to begin with, consider the case of pain. If pain is
realized by C-fibres firing in me, together with, perhaps, the laws which
govern their causal role, and these firing fibres are transplanted into our
silicon friend, Robbie the Robert, we would expect no writhing to ensue.
Now it might be urged that, in that case, what is shown to be efficacious is
the way in which the pain is realized and not pain itself. Nevertheless, there
is a fact unaddressed by this suggestion. It is that all of the realizations of pain
(e.g., in the case of Robbie the Robert C-circuit activity) have conducive
circumstances in which they give rise to writhing behaviour. This is not the
case for other properties with regard to that kind of behaviour. So it is
plausible that there is something about pain, and not just pain in such and
such a type of creature, which is responsible for the link between pain and
writhing. It is this which grounds the claim that pain is causally relevant for
this behaviour.
Suppose that there is a lone species in which instances of pain do
not cause writhing. Would that imply that human pain does not cause
writhing?1 It would not because, by limiting the question to human pain,
the generality condition would only apply to necessitation-bases of human
pain. Nor would it even imply that pain is not a property cause of writhing.
That would depend on whether the way in which pain is realized in this
lone species might also be realized in other creatures in which it did cause
writhings, or whether there were conditions in the lone species in which
writhing might be so caused.
The generality condition is also related to, but importantly distinct from,
a distinction drawn recently between sensitive and insensitive causation.
Often, the latter distinction is made within the context of taking causation
to be difference making, something I discussed in the second section of this
paper (e.g., Woodward 2006, 7). A causal relation is relatively insensitive—
between particulars, or types of things—if the counterfactual dependence
between the causal relata holds in a variety of different background condi-
tions. It is sensitive if this dependence is easily disrupted. Christian List and
Peter Menzies extend this idea to include sensitivity, or otherwise, to the
way in which the properties standing in the putative causal relation are
realized. Cases of sensitive causation in this sense are taken to be counter-
examples to the exclusion principle I mentioned earlier. List and Menzies
1 This question was asked by an anonymous referee.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 115
hold that both the way in which a property is realized, and the instance of
the property, are to be counted as causes in such cases (List and Menzies
2009, 491–2, 497–9; in Menzies 2008, he seems committed to an exclusion
principle to which this kind of case is a counterexample).
To illustrate, suppose that a certain kind of pain, Pa, has four necessitation-
bases N1, N2, N3, and N4 and let Bg be the utterance ‘That hurts!’ Suppose
that, further, the following counterfactuals were true.
If Pa were not instantiated in S, then S would not utter Bg.
If N1 were not instantiated in S, then S would not utter Bg.
For the latter to be true, the closest worlds in which N1 is not present are
ones in which Bg wouldn’t occur even though there is a replacement, N2,
and Pa is, thus, present. In those circumstances, List and Menzies conclude
that both Pa and N1 are causes of Bg.
I can see why it is plausible to suppose that N1 is a property cause in that
situation. It is far less clear why it is plausible to suppose that Pa is. Givenwhat
has been previously been argued, we are allowed the question: Does N1
cause Bg partly in virtue of necessitating Pa? Evidence that it is not in virtue of
Pa is that, when a substitute realization, N2, is present, Bg does not occur. List
and Menzies suggest that the relationship between Pa and B is sensitive,
depending upon the precise way in which Pa is realized. Instead, the sensi-
tivity supplies evidence that it is N1 rather than Pa that is the causally relevant
property. If the sensitivity were just the result of a failure of the right causal
circumstances, then the case List and Menzies cite would not be a problem.
The verdicts of the two approaches would coincide. The difference stems
from the decision to count as one source of sensitivity the way in which Pa is
realized. It is here that I think their account yields counterintuitive verdicts.
Sensitivity is not compatible with causal relevance.
A consequence of my favoured account is that it delivers the verdict that
there are causal relations between broadly physical properties. Kim’s argu-
ment may be viewed as questioning this on the grounds that all the work is
being done by the arrangements of narrowly physical properties. Since my
proposal does not make instances of broadly physical properties identical to
instances of arrangements of narrowly physical properties, I don’t have an
immediate response to this worry. True my proposal may get the right
verdict in the sense of what we want to believe but the charge is that it
shouldn’t.
116 PAUL NOORDHOF
Part of my response to this objection is contained in my reply to the
challenge of the unilevellers. At this point, I emphasize another issue. The
debate in this area begins by conceding that there are broadly physical
properties but then challenges their existence by arguing that they are
inefficacious. However, the initial concession undermines the challenge.
Either you don’t think broadly physical properties exist, in which case
I draw your attention to the way in which they back inferences we want
to make about how things will behave in different circumstances. Or you
accept that they do, barring an argument to the contrary. Appeal to causal
considerations will not provide such an argument because, in allowing that
broadly physical properties exist, you must also allow that broadly physical
causal relations exist. They are just one more species of property whose
existence we have allowed as a result of their necessitation by arrangements
of, in this case, causal relations between narrowly physical properties. There
seems an entirely unmotivated asymmetry in the debate whereby causal
relations are treated differently to any other kind of property. One illustra-
tion of this last point is that, just as other properties seem to stand in relations
of determinable to determinate, so do various types of causal relations, for
example, 6 inch diameter ball depression, ball depression, depression, spe-
cify causal relations at different degrees of generality. These are determin-
able causal relations in which determinable properties may stand.
There might be other reasons to resist the claim that broadly physical
causal relations exist. My point is simply that these considerations had better
not take the same form as considerations, independent of causation, for
rejecting the existence of broadly physical properties in general. We were
supposed to be provided with a consideration from causation against the
latter, not just a blanket favouring of the narrowly physical. My account of
property causation is an attempt to identify when these broadly physical
property causal relations are present, and how they capture something in
addition to particular arrangements of narrowly physical properties, through
the generality condition.2
Another objection to the proposal discussed recently derives from the
possible truth of a powers ontology. A powers ontology takes the causal
profile of a property to be internal to it. By that I mean that the causal profile
of the property does not depend upon laws which hold, in addition, but
2 This paragraph was written in response to an objection by an anonymous reviewer.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 117
rather given that the property is instantiated, certain laws hold. Suppose that
emergent dualism is true. Then one element of the causal profile of an
arrangement of narrowly physical properties is that they cause the presence
of an emergent non-physical property. If arrangements of narrowly physical
properties in such an ontology could not fail to have their causal profiles,
then it follows of metaphysical necessity that, if the arrangement of narrowly
physical properties is present, then the emergent dualist property is instanti-
ated. Nevertheless, it could still be the case that it is not part of the causal
profile of the emergent dualist property that it cause some target effect
which is part of the causal profile of the arrangement of narrowly physical
properties. Indeed, that is what epiphenomenal emergent dualists assert.
The objection to my proposal is that it suggests a certain account of how
broadly physical properties can inherit the efficacy of arrangements of
narrowly physical properties that cannot allow for this possibility (O’Con-
nor 1994, 97; Wilson 2005, 436–47).
One response to the specifics of the objection is to say that if a powers
ontology were true, there would be no basis for being an epiphenomenal
emergent dualist. The grounds for being a dualist are usually the intrinsic
features of phenomenal states. If a powers ontology were true, there would
either be no intrinsic features, or the intrinsic features in question would not
be different for narrowly physical properties. I mention this last possibility to
take into account C. B. Martin’s position that every property has both a
qualitative and dispositional aspect (e.g., Martin 1997).
Nevertheless, this does not deal with the general structure of the objec-
tion. Suppose that there is a property C1 which has a causal role CR1 which
includes, if C1 is instantiated in S, then E1 and F1 is instantiated. Then C1 and
S metaphysically necessitate E1 and they also metaphysically necessitate F1.
Doesn’t my position have as the upshot that E1’s causal role ER1 must
include the instantiation of F1 when, intuitively, it need have nothing to do
with the instantiation of F1?
Here are two more general lines of response. First, my talk of metaphys-
ical necessitation was meant to capture the important characteristic of
previous talk of constitution, namely that if that which was necessitated
by the necessitation-base involved nothing more than what was in the base,
then given the base, the necessitated must also be instantiated. It might be
argued that, if a powers ontology is true, metaphysical necessitation cannot
suffice to capture our notion of constitution even given the assumption that
118 PAUL NOORDHOF
the entities it associates are contingent. In which case, we might take
property instance constitution as a primitive and note that it supports
metaphysical necessitation claims but is not the only possible support. The
proposal would be reformulated in terms of constitution. This is not a
particularly damaging adjustment because there is no reason to think that
the proper understanding of property instance constitution must appeal to
causation or kindred notions that I am seeking to illuminate by my proposal.
Second, we can deny that a powers ontology implies dispositional essen-
tialism, the view that the causal profile of a property is essential to it. In
which case, there is no reason to accept that C1 and S metaphysically
necessitate E1 and hence no grounds for supposing that my proposal must
accept the verdict that E1 has the instantiation of F1 as part of its causal
profile. The point is especially plausible with regard to the fundamental laws
relating arrangements of narrowly physical properties and the properties put
forward by epiphenomenal emergent dualists as mental properties. They
envisage that these stand in isolation from other narrowly physical proper-
ties and so it is perfectly conceivable that the same narrowly physical
properties may be instantiated without this part of their causal profile.
However, more generally, any particular aspect of the causal profile of a
property could be plausibly supposed to be absent with the rest still present.
One way to think of these possibilities is in terms of counterpart theory.
We can suppose that, in other possible worlds, there are properties with a
strong similarity to the causal profiles instantiated in our world—structurally
speaking—and yet some differences. The question arises whether it is
plausible to consider these properties counterparts of the property in our
world. It is hard to see why not. Properties with different causal profiles may
be counterparts and yet nothing that has been said rules out the possibility
that the causal profile is internal to the property. Indeed, counterpart theory
was introduced to, amongst other things, deal with the problem of acciden-
tal intrinsic properties of particulars. The suggested strategy just applies this
to the case of properties (for further discussion, see Noordhof 2010).
A second objection to the proposal follows from something I said earlier.
I remarked that the causal powers of many broadly physical properties
exceeded those of the arrangement of narrowly physical properties that
metaphysically necessitated them. This can seem wrong on one of two
counts. First, how can it be that novel causal powers are metaphysically
necessitated by arrangements of properties which, it is alleged, individually
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 119
or together don’t possess them? Second, if it is allowed that they do
metaphysically necessitate novel causal powers for the sake of argument,
then why doesn’t just admitting this make the causal powers accrue to the
arrangements of narrowly physical properties which do the necessitating?
The answer to the second question is that the powers of BP1 don’t
transmit to A1(p1, p2, p3 . . . ) because, first, the instance of BP1 is not
identical to, nor caused by, the instance of A1(p1, p2, p3 . . . ) and, second,
downward transmission of causal powers does not apply because A1(p1, p2,
p3 . . . ); the latter stops some of the powers of BP1 from being manifestable,
namely those associated with other physical realizations of BP1. Of course,
part of the causal profile of a property F need not be manifested in order for
the property to have that causal profile. However, F cannot have, as part of
its causal profile, the potential for causal relations it could not stand in while
remaining the property it is, given the laws which hold. I mention this
second point in case it is thought that a version of my proposal should
explain how efficacy of broadly physical properties transmits downwards
even if the attribution of the causal powers is not immediate in virtue of the
first point.
This response to the second question makes it harder to see how one
could provide an answer to the first. How can a particular arrangement of
narrowly physical properties necessitate a property which has causal powers
more extensive than it? An incomplete answer would be that, although the
powers of BP1 exceed particular minimal necessitation-bases of it, if we
consider all the various minimal necessitation-bases, then the complete set
of causal powers that these minimal necessitation-bases have is possessed by
BP1. There are two problems with this response. The first is that its
plausibility partially rests upon the assumption that all the possible minimal
necessitation-bases of BP1, which give it distinct causal powers, are instanti-
ated in a particular world. In the absence of this, upon what basis could we
conclude that the other elements of the causal power were present? This is
not merely a notional objection. Many candidate BPs actually allow for
physical and non-physical minimal necessitation-bases with the presump-
tion that there are none of the latter if physicalism is true. Second, even if we
have some explanation of why we may allow that all the powers associated
with BP1 are instantiated, it is unclear why we should conclude that they are
instantiated with regard to a particular instance of BP1 when necessitated by
a minimal necessitation-base that cannot have at least some of the powers.
120 PAUL NOORDHOF
The proper response is to distinguish constitution as co-ordination of the small
in making up the bigger from constitution as involving grounding, in which the
constituents are viewed as fundamental. These are clearly distinct notions
otherwise we would have a fast argument from something being a constitu-
ent to monism (the priority of the whole) being false. We should reject the
idea that arrangements of narrowly physical properties constitute broadly
physical properties and, more specifically, that the causal relations of the
narrowly physical properties so arranged constitute the causal relations of the
broadly physical properties in a metaphysically fundamental sense in which
the constituents are taken to be primary. Instead, the proper relationship
between narrowly physical and broadly physical properties is one of har-
monization (see Noordhof 2003, 105–6). The right metaphor is not of an
economical God who, if only he were to fix the arrangements of the
physical, he would have the broadly physical properties fixed, but rather
of a God subject to constraints. He is not allowed to instantiate some of the
first lot without instantiating some of the second lot too. Broadly physical
properties, and their causal relations, are no less fundamental than the
arrangements of narrowly physical properties with which they are closely
related.
From this alternative perspective, the relations of metaphysical necessita-
tion between the arrangements of narrowly physical properties and broadly
physical properties capture the constraints upon instantiation, and co-in-
stantiation, between these properties. If the constraints are not observed,
then the causal relations of the properties would literally be incompatible
with each other. We would have an impossible world. Arrangements of
narrowly physical properties only appear to be ontologically fundamental
because the causal relations identified at that level are more detailed than
those identified between broadly physical properties. Since there are various
ways more general causal relations may be realized by more detailed causal
relations, we have an asymmetry. Arrangements of narrowly physical prop-
erties fix what broadly physical properties there are but the latter only imply
that one or other of various arrangements of narrowly physical properties
are present. However, interpreting this asymmetry as implying that arrange-
ments of narrowly physical properties are fundamental is not mandatory if
constitution is just co-ordination of the small.
Of course it is true that instances of narrowly physical properties may be
present, and stand in causal relations, whether or not broadly physical
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 121
properties are present. That might suggest that they have some priority.
However, once the alternative picture is in play, this fact needs to be set in
the context of other observations. First, since broadly physical properties
may be related to different arrangements of narrowly physical properties and,
indeed, in some cases, to arrangements of non-physical properties, there is no
reason to take broadly physical properties to be dependent on their instances’
actual constituents. Second, arrangements of narrowly physical properties
are subject to constraints on co-instantiation stemming from their constitut-
ing, in the co-ordination of the small sense, broadly physical properties. If a
broadly physical property is to be instantiated with certain causal powers, and
certain constituent instantiations of narrowly physical properties are to be
instantiated, then certain other constituent instantiations of narrowly phys-
ical properties must be instantiated too, namely those implied by the instanti-
ation of the broadly physical property with those other narrowly physical
properties as constituents.
Recognition of novel causal powers, in the way that I have sketched,
does not constitute a rejection of a weak causal closure principle like ‘every
event with a cause has a narrowly physical cause’ for, at least, two reasons.
First, that principle is compatible with there being non-physical causes too.
But, second, and more important in the present context, denying that
arrangements of narrowly physical properties constitute the causal relations
of broadly physical properties, does not mean that there are events with
broadly physical properties as causes without arrangements of narrowly
physical properties as causes. Allowing that there are cases in which the
broadly physical properties are no less fundamental than narrowly physical
ones does not imply that they have causal consequences without arrange-
ments of narrowly physical properties being present.3
6. Concluding Remarks
The causal relevance of properties, or property causation as opposed to
property instance causation, turns on two issues: first, causal facts about their
instances; second, the causal significance of a generality captured in terms of
the properties in question. Focus on difference making, or patterns of
3 This paragraph was written in response to a question by an anonymous referee.
122 PAUL NOORDHOF
variation, are better at capturing the second element but they ignore the first
element at their peril. Even with regard to the required display of generality,
they fail to observe the conditions which should be met for a property to be
said to be causally relevant (rather than just its instances).My own proposal—
involving a particularity condition plus a generality condition—has themerit
of indicating how the two elements should be integrated. It suggests that the
concern about efficacy, within the context of non-reductive physicalism,
partly stems from an inadequate understanding of how this integration
should be undertaken given that there are different minimal necessitation-
bases for broadly physical properties together with an unmotivated asym-
metry in the treatment of causation itself, as just one kind of property
amongst others. The other root of the trouble is the focus on developing a
picture of the world which places the emphasis on truthmaking rather than
inference-basing. The recognition that non-reductive physicalists should
allow broadly physical properties to have causal powers which outstrip
their bases in a circumscribed sense and, as a result, resist the constitution-
as-grounding assumption, suggests a different understanding of how we
should see the relationship between broadly and narrowly physical proper-
ties. If this proposal is along the right lines, it provides support for the view
that the apparent problem of mental causation is only properly resolved if the
metaphysical picture, in which it might figure, is made much clearer: the
theme of the AHRC funded project to which this paper was a contribution.4
References
Campbell, K. (1981). ‘The Metaphysic of Abstract Particulars’. Midwest Studies in
Philosophy, 6: 477–86.
—— (1990). Abstract Particulars. Oxford: Basil Blackwell.
Ehring, D. (1997). Causation and Persistence. Oxford: Oxford University Press.
4 Many thanks to the audience of the Metaphysics of Mind conference in Durham to which an earlier
version of this was presented, S. C. Gibb, for inviting me to participate and, more recently, to the
members of the York Mind and Reason group who subjected my ideas to scrutiny: Keith Allen, Will
McNeill, Ema Sullivan-Bissett, and Rachael Wiseman. I would also like to thank the Leverhulme Trust
for supporting a period of research, of which this was an element, by a Major Leverhulme Research
Fellowship.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 123
Fodor, J. A. (1989). ‘Making Mind Matter More’. Philosophical Topics, 17(1): 59–79.
Repr. in Fodor (1990), A Theory of Content and Other Essays. Cambridge, MA:
MIT Press: 137–59.
—— (1991). ‘You Can Fool Some of the People All of the Time, Everything Else
Being Equal; Hedged Laws and Psychological Explanation’.Mind, 100(1):19–34.
Gibb, S. (2004). ‘The Problem of Mental Causation and the Nature of Properties’.
Australasian Journal of Philosophy, 83(3): 464–76.
Gibbons, J. (2006). ‘Mental Causation without Downward Causation’. Philosophical
Review, 115(1): 79–103.
Heil, J. (2003). From an Ontological Point of View. Oxford: Oxford University Press.
Hirsch, E. (1993). Dividing Reality. Oxford: Oxford University Press.
Kim, J. (1992). ‘Multiple Realization and the Metaphysics of Reduction’. Philoso-
phy and Phenomenological Research, 52: 1–26. Repr. in Kim (1993), Supervenience
and Mind. Cambridge: Cambridge University Press: 309–35 [page references in
text to latter].
—— (1998). Mind in a Physical World. Cambridge, MA: MIT Press.
—— (1999). ‘Supervenience Properties and Micro-Based Properties: A Reply to
Noordhof ’. Proceedings of the Aristotelian Society, 99: 115–18.
—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-
sity Press.
Lewis, D. (1986). ‘Against Structural Universals’. Australasian Journal of Philosophy,
64: 25–46.
List, C. and P. Menzies (2009). ‘Nonreductive Physicalism and the Limits of the
Exclusion Principle’. Journal of Philosophy, 106(9): 475–502.
Manley, D. (2002). ‘Properties and Resemblance Classes’. Nous, 36(1): 75–96.
Martin, C. B. (1997). ‘On the Need for Properties: The Road to Pythagoreanism
and Back’. Synthese, 112(2): 193–231.
Menzies, P. (2008). ‘The Exclusion Problem, the Determination Relation, and
Contrastive Causation’. In J. Hohwy and J. Kallestrup (eds.), Being Reduced: New
Essays on Reductive Explanation and Special Science Causation. Oxford: Oxford
University Press: 196–217.
Noordhof, P. (1998). ‘Do Tropes Resolve the Problem of Mental Causation’.
Philosophical Quarterly, 48 (191): 221–6.
—— (1999a). ‘Causation by Content?’. Mind and Language, 14(3): 291–320.
—— (1999b). ‘Micro-Based Properties and the Supervenience Argument:
A Response to Kim’. Proceedings of the Aristotelian Society, 99: 109–14.
—— (1999c). ‘The Overdetermination Argument versus the Cause-and-Essence
Principle—No Contest’. Mind, 108(430): 367–75.
124 PAUL NOORDHOF
—— (2003). ‘Not old . . . but not that new either: Explicability, Emergence and the
Characterisation of Materialism’. In S. Walter and H.-D. Heckman (eds.),
Physicalism and Mental Causation: The Metaphysics of Mind and Action. Charlottes-
ville: Imprint Academic: 85–108.
—— (2010). ‘Emergent Causation and Property Causation’. In C. Macdonald and
G.Macdonald (eds.),Emergence inMind. Oxford: OxfordUniversity Press: 69–99.
O’Connor, T. (1994). ‘Emergent Properties’. American Philosophical Quarterly, 31(2):
91–104.
Olson, E. (1997). The Human Animal. Oxford: Oxford University Press.
—— (2007). What Are We? Oxford: Oxford University Press.
Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:
178–94.
Rodriguez-Pereyra, G. (2002). Resemblance Nominalism. Oxford: Oxford Univer-
sity Press.
Schaffer, J. (2005). ‘Contrastive Causation’. Philosophical Review, 114(3): 327–58.
Schiffer, S. (1991). ‘Ceterus Paribus Laws’. Mind, 100(1): 1–17.
Segal, G. and E. Sober (1991). ‘The Causal Efficacy of Content’. Philosophical
Studies, 63: 1–30.
Shoemaker, S. (1981). ‘Some Varieties of Functionalism’. Philosophical Topics, 12(1):
83–118. Repr. in Shoemaker (1984, 2003), Identity, Cause and Mind. Oxford:
Oxford University Press: 261–86.
—— (2007). Physical Realization. Oxford: Oxford University Press.
Steward, H. (1997). ‘On the Notion of a Cause “Philosophically Speaking”’.
Proceedings of the Aristotelian Society, 97: 125–40.
Whittle, A. (2007). ‘The Co-Instantiation Thesis’. Australasian Journal of Philosophy,
85(1): 61–79.
Wilson, J. (2005). ‘Supervenience-based Formulations of Physicalism’. Nous, 39(3):
426–59.
Woodward, J. (2006). ‘Sensitive and Insensitive Causation’. Philosophical Review,
115(1): 1–50.
Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.
MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 125
5
Causation is Macroscopicbut Not Irreducible
DAVID PAPINEAU
1. Introduction
In this paper I argue that causation is an essentially macroscopic phenom-
enon, and that mental causes are therefore capable of out-competing their
more specific physical realizers as causes of physical effects. But I also argue
that any causes must be type-identical with physical properties, on pain of
positing inexplicable physical conspiracies. I therefore allow macroscopic
mental causation, but only when it is physically reducible.
2. Causation is Macroscopic and Not Physically
Fundamental
It is widely supposed, at least among philosophers, that causation is grounded
in basic dynamical processes. At bottom, according to this supposition, causal
relations consist in the way that basic dynamical laws govern the temporal
evolution of precise arrangements of fundamental physical particles and fields.
This supposition is often on display in contemporary debates about mental
and other ‘higher-level’ causation. Suppose you think that some mental state
M—your wanting to hail a taxi, say—supervenes on physical state P without
being reducible to it. (P is more specific than M: it metaphysically necessitates
M, but M can be realized by physical states other than P.) Now consider some
further physical effect P*—your arm moving—that apparently results from
M. Can M really be the cause of P*? Some philosophers deny that it can, on
the grounds thatMwill always be ‘out-competed’ as a cause by P.Others assert
that M can be such a cause, arguing that there is nothing wrong with both
M and P ‘overdetermining’ the effect P* in such cases. But scarcely anyone
queries whether the realizing P will itself qualify as a cause of P*.1
Thus Jaegwon Kim, in discussing just this kind of case, insists that ‘The
question is not whether P should be considered a cause of P*; on anyone’s
account, it should be’ (Kim 1993, 207, my italics). (Kim then continues with
his familiar query: ‘What causal work is left over for M, or any other mental
property, to do?’) However, I myself am very doubtful about the claim that
Kim takes to be agreed on all sides.
Why does Kim take it to be obvious that P is a cause of P*? Presumably
he is reasoning from the above supposition that causation is constituted by
the way maximally precise physical arrangements evolve in accord with
basic dynamical laws. If P is a full specification of physical initial conditions,
and these evolve in line with basic dynamic laws into P*, then of course—or
so Kim assumes—P must cause P*.
However, there is strong reason to doubt that causation is constituted by
basic dynamical processes. The objection is simple. Causation is asymmetric
in time, but basic dynamics is not. So it seems that causation must involve
something more than basic dynamics.
The basic laws of dynamics determine no direction in time. Take a
specification of what happens at each point of spacetime in some closed
physical system. Then you can view the ‘initial’ conditions as evolving into
the ‘final’ conditions in accord with the basic dynamical laws. But you could
equally well think of time as ‘flowing’ in the opposite direction, with the
‘final’ conditions evolving into the ‘initial’ ones, again in accord with the
basic dynamical laws. In this sense, the basic laws of dynamics don’t care
which direction is ‘earlier’ and which ‘later’. Accordingly, if you are given
a basic dynamical description of a physical system, but not told which
temporal direction is which, you won’t be able to read this off from the
description.
But causation is different. If you can discern the causes and effects within
a physical process, then this alone will tell you which way time is directed.
1 I take causes and effects to be facts or states of affairs, paradigmatically consisting of some particular
possessing some property. Given this, it will sometimes be natural to talk as if properties themselves are
causes or effects; but strictly what is meant by such usages are the facts involving those properties.
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 127
Causes always come earlier than their effects, and so a specification of causal
structure will tell you which temporal direction is which.
Of course, this would be trivial if the difference between ‘causes’ and
‘effects’ simply consisted in the former occurring earlier than the latter. Thus
suppose that the relationship between ‘causes’ and ‘effects’ were just like
that between the sets of conditions at the two temporal ends of a basic
dynamic process, save that the direction of time was independently given,
and it was specified that the ‘cause’ was the earlier set of conditions, and the
‘effect’ the later set. Then it would scarcely be surprising that we could read
off temporal order from information about ‘causes’ and ‘effects’—for such
talk would simply add this temporal information explicitly to the temporally
neutral dynamic facts.
But it is arguable that the difference between causes and effects lies deeper
than this, and can be discerned prior to any independently given infor-
mation about temporal order. Causal relations have a characteristic prob-
abilistic signature which is asymmetric in time. It is this signature that the
recent tradition of ‘Bayesian nets’ exploits in order to infer causal structure
from probabilistic correlations. It is noteworthy that the techniques
exploited by the Bayesian net tradition do not need to assume temporal
order in order to distinguish causes from effects. Sufficiently rich correla-
tional information on its own will always determine a causal order among
related variables. (This is not the place to go into details. But, to get a
flavour, note that the correlation between the joint effects of a common
cause will disappear when we ‘control’ for the common cause—that is,
consider separately cases where the cause is absent and where it is present.
By contrast, any correlations between the joint causes of a common effectwill
not disappear when we ‘control’ for that effect. For more on the asymmetric
probabilistic dimension of causation see Hausman 1998; Spirtes et al. 2000;
Papineau 2001.)
There is room here to debate the precise metaphysical relationship
between the underlying causal structure and the temporally asymmetric
correlational structure which manifests it. But the very possibility of epi-
stemologically distinguishing causes from effects without assuming temporal
order suggests that there must be something in the nature of causation that
orientates it in time. If so, this means that there must be more to causation
than the temporally symmetric structures of basic dynamics.
128 DAVID PAPINEAU
3. Thermodynamics and Causation
It is illuminating to compare causation with thermodynamics in the above
respects. As is well known, the second law of thermodynamics is also inexplic-
able in terms of basic dynamics alone, precisely because it refers to a specific
direction in time: later entropy is always greater than earlier entropy within a
closed physical system. So an explanation of the second law needs to invoke
assumptions that go beyond basic dynamics. In particular, such an explanation
needs to posit, in addition to basic dynamics, first, that entropy was low in the
past, and, second, that there is a certain probability distribution over all the
precise physical microstates consistent with given ‘macrostates’ of tempera-
ture, energy, entropy, and so on (see Albert 2000).
I take causal asymmetry to have an analogous basis. There is no estab-
lished way of relating causation to thermodynamics. But the asymmetric
correlational structures displayed by causal relationships suggest that causal
processes are akin to thermodynamic processes. In particular, it suggests that
causation is also is an essentially macroscopic phenomenon, constituted by
the nature of past facts together with probability distributions over the
maximally specific microstates that can realize given macrostates.
Some readers might be puzzled by the suggestion that thermodynamic
processes, and causal relationships alongwith them, are essentiallymacroscopic
phenomena. I alluded above to the way that thermodynamic processes like
entropy increase can be explained in terms of particle physics, together with
past facts and probability distributions over microstates. But if such explan-
ations are possible, then don’t they show that themacroscopic thermodynamic
phenomena can all be reduced to microscopic processes, and so aren’t really
macroscopic after all?
But it does not work like that. The explanation of thermodynamic
phenomena by particle physics does not eliminate macroscopic features,
but makes essential use of them. Take a volume of gas that is hot in one half
and cold in another. Thermodynamics tells us that in a while the tempera-
ture will almost certainly be uniform throughout. Now, you could in
principle have analysed this particular system by applying basic dynamics
to the precise initial conditions of all the particles involved, and this would
no doubt have told you that the later temperature would be uniform. But
this microanalysis would owe nothing to the general principle that almost
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 129
any system in that initial macrostate would end up at a uniform temperature.
(After all, you could have applied an entirely analogous microanalysis to
predict the evolution of one of the very unlikely ‘rogue’ microstates that
would not end up with a uniform temperature.) To bring out the general
principle, you need to ‘throw away’ the information about the precise
microstate, and note instead that the system is in a macrostate which is
overwhelmingly likely to (be realized by a microstate that will) end up
with a uniform temperature. This is why thermodynamics is essentially
macroscopic. Without probabilistic information about the way in which
given macrostates get realized by microstates, you cannot infer any thermo-
dynamic patterns from microphysics.
Similarly, I suggest, causation is an essentially macroscopic phenomenon.
If you focus on the precise microstate of some physical process, you will lose
sight of causation. The causal structure of the world depends on probabil-
istic facts about the ways in which given macrostates are realized at the
micro-level, rather than on the actual micro-realizations themselves.
4. Intuitions are Irrelevant
Of course, this is not howwe think about causation intuitively. The intuitive
paradigm of a causal interaction is of one physical object bumping into
another and the latter’s motion changing. We humans are naturally prone
to judge without further ado that in such cases the impact of the former
caused the new motion of the latter (see Michotte 1946/1963). In line with
this, our intuitive conception of causation contains no mention of probabil-
istic distributions over the microstates that realize different macrostates, and
correspondingly our concept of causation sees no contradiction in the idea of
causation existing even when such probabilistic distributions do not.
In this connection, consider Ernest Sosa and Michael Tooley’s objection
to the way David Lewis analyses causation in terms of the ‘asymmetry of
overdetermination’. Lewis’s analysis is in the spirit of theories that account
for causal asymmetry in terms of probabilistic asymmetry. True, Lewis does
not put this analysis explicitly in probabilistic terms, but the appeal to
probabilistic facts is not far beneath the surface. (Thus note how Lewis
appeals to the way that causes typically issue in many independent chains of
effects, yet typically only stem from one chain of causes.)
130 DAVID PAPINEAU
Sosa and Tooley object to Lewis’s theory on the grounds that
it is not a necessary truth that any world containing causally related events is one
where events typically have more effects than causes. The world . . . could have
been a very simple one, where there were no causal forks . . . Lewis’s analysis cannot
be sound, therefore, since there are logically possible causal worlds for which it
yields the wrong results . . . (Sosa and Tooley 1993, 27)
Well, no doubt there are conceivable scenarios which contain causal relations
but lack the asymmetric probabilistic structure to which Lewis appeals in
explaining causation. And if Lewis’s theory were put forward as a piece of
conceptual analysis, then such scenarios would suffice to refute it.
But there is no reason to read theories like Lewis’s in this conceptual way.
Rather, they are better understood as synthetic metaphysical theories,
which aim to uncover the nature of causation, not via analysis of our
concepts, but through a posteriori investigation of the world we live in.
You can’t argue against theories of this kind by appealing to merely conceiv-
able scenarios, any more than you can argue against orthodox chemistry by
appealing to the conceivability of a world with water but no H2O. Of
course, if it could be established that worlds with causation but no probabil-
istic structure were metaphysically possible, then this would indeed defeat
probabilistic accounts of the nature of causation. But the mere conceiv-
ability of such worlds does not show that they are metaphysically possible. If
causation is indeed constituted by temporally asymmetric probabilistic
structure, then there is no metaphysical possibility of the one without the
other, however much this may be conceivable.
5. Autonomous Mental Causes
So far I have argued that it is a mistake to think of causal relationships as
being determined by some maximally specific level of physical facts. Rather
causation depends on general patterns essentially involving macroscopic
properties, where these macroscopic properties will be realized by different
arrangements of fundamental physical facts on different occasions.
Over the last couple of decades a number of philosophers have argued
that mental facts M are no less causes of subsequent physical results P* than
their physical realizers P. On this view, the result P* can be attributed to
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 131
both of the ‘parallel causes’ M and P. Of course, nobody wants to view all
mental causation as overdetermination by two ontologically distinct causes,
like the death of the man who is shot and struck by lightning at the same
time. But defenders of the ‘parallel causes view’ can observe that M and
P are not so ontologically distinct, in that M metaphysically supervenes on
P. True, M is not identical to P, and so in a sense a kind of ‘benign
overdetermination’ is being posited. But precisely because M supervenes
on P, and is not ontologically independent, it is not obvious that there is
anything wrong with such benign overdetermination. (See Shoemaker
2001; Pereboom 2002; Bennett 2003.)
This ‘parallel causes view’ has the virtue of recognizing macroscopic
mental facts as causes in their own right. But the points made in the last
section open the way to a more radical position. Why shouldn’t the mental
state M be the cause of P* rather than the physical state P? If causation derives
from patterns essentially involving macrostates, then perhaps it is the mental
M that figures in these patterns, not the physical P.
Just this possibility has been explored by some recent writers. (See
Menzies 2008; List and Menzies 2009; Menzies and List 2010. Also relevant
are LePore and Loewer 1987; Yablo 1992.) Their standard form of argument
appeals to plausible counterfactual requirements on causation. Let us sup-
pose that, if C causes E, then
(1) E wouldn’t have occurred if C hadn’t occurred, and
(2) E would still have occurred if C had occurred differently.
Take the case where you are waving for a taxi. Let the physical effect P* be the
movement of your arm. M is your wanting to hail a taxi. P is the definite
neuronal arrangement which realizes this mental state. Now, both M and
P satisfy clause (2)—your arm would still have moved as long as either M or
P occurred, even if they had occurred in a different way. But only M satisfies
clause (1)—your arm wouldn’t have moved if you hadn’t wanted a taxi—
where P does not—the absence of just that precise neuronal arrangement
wouldn’t have stopped your arm moving, for you would still have wanted to
wave even if your desire had been realized by a slightly different neuronal state.2
2 Laurie Paul has queried (in correspondence) whether M would have been different realized, rather
than simply being absent, if P had not occurred. My reading does sound natural to my ear, but in any case
let me simply stipulate that the suggested counterfactual requirement on causation be understood this way.
132 DAVID PAPINEAU
So here M is the cause and not P. P is too specific. This analysis is in line
with Stephen Yablo’s thesis (1992) that causes should be proportional to their
effects. Causes must be specific enough for their effects, but no more
specific than this requires. In the above example, the neuronal arrangement
P is too specific for the effect, but the wanting M is just right.
Note how the example comes out differently if the effect P* is not your
arm moving as such, but your arm moving in the precise way that it did on
this occasion. Now both M and P satisfy clause (1)—if you hadn’t wanted to,
you wouldn’t have waved at all, and so a fortiori not just as you did; and if
you hadn’t had just that precise neuronal set-up, you also wouldn’t have
waved just as you did. But, with this precise effect P*, only P satisfies clause
(2)—while you would still have moved just like that if P had occurred a bit
differently, you wouldn’t have moved just like that if you’d still wanted to
move your arm but this desire had been realized with some different
neuronal arrangement.
So now P is the cause and not M. The wanting M is not specific enough
to account for your moving just like that, but the neuronal arrangement P is
just right. Does this last example not run against the points made in my first
section? There I said that causation is an essentially macroscopic phenom-
enon, and disappears at the level of maximally specific physical processes.
Now I am saying that your neuronal arrangement P can be the cause of your
particular movements, rather than your mental state M. However, these
claims are not inconsistent. The reason is that the neuronal arrangement
P needn’t constitute a maximally specific physical state. Just as a given mental
state M can be realized by different neuronal arrangements, so too can a
given neuronal arrangement be different realized at the maximally specific
level of precise fundamental particles and fields. So in both cases there is
room for the kinds of probabilistic facts which I say are essential to asym-
metrical causal relationships. It is only at the level of fully specific physical
arrangements that causation disappears.
Let us return to the analogy between causation and thermodynamics.
Suppose we have a volume of gas in a container with a safety valve. If the gas
is heated, there is a temperature T at which the valve will open. This
temperature can be realized by the many different sets of particle move-
ments which would yield the requisite mean kinetic energy. Which causes
the valve’s opening on some given occasion, the temperature T or the
specific particle movements which there realize that temperature? Intuition
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 133
might suggest that it is the particle movements. But if the effect at issue is the
opening as such, the approach I am defending argues that it is the tempera-
ture and not the particle movements that cause this effect. The particle
movements are too specific. We would still have had the opening even if
the temperature T had been realized by different particle movements.
Does thismean that the specific particlemovements cannot cause anything?
That would be an undesirable conclusion. There are more fine-grained
effects, such as the precise trajectory of the valve’s opening, which will surely
be the results of the specific particle movements. Perhaps the valve’s opening
in that specific manner is due to precise sequence of high-energy particles that
impact its inner surface. With the more fine-grained effect, it is the more
specific particle movements and not the generic temperature T which is
proportional to the cause: if the temperature T had been differently realized,
then the valve would not have opened in just that manner.
Note how the causal efficacy of the particle movements does not under-
mine my claim that causation is essentially a macroscopic phenomenon. Even
after we have focused on the definite particle movements, there will be yet
further features of the set-up—such as the bonding properties of the gas’s
molecules, the molecular structure of the valve’s inner surface, and so on—
that will still be variably realized at the level of fully specific physical arrange-
ments. And it is still probabilistic facts about the distribution of such further
realizers that underpin the asymmetric causal relationship between the par-
ticle movements and the manner of opening. If we descend to a level where
all physical facts are fully determinate, then I say that we lose sight of any
asymmetric causal relationships. But this leaves plenty of room for relatively
definite physical facts like given particle movements to function as genuine
causes of relatively fine-grained effects. We can descend to particle move-
ments without descending to the level of fully specific physical states.
I alluded earlier to the idea that a mental cause M and a physical realizer
P might both be the cause of some physical effect P*, via a sort of benign
overdetermination. The points made so far in this section argue that this is
not a possibility—not on the grounds that such overdetermination per se
would generate any unacceptable consequences, but simply because the
requirements of proportionality rule out two such causes. Once we have
fixed on a specific effect P*, then it can’t be that some M and some more
specific realizer P are both causes of P*. If M is proportionally ‘just right’, then
the more specific P will violate requirement (1), in that we would still have
134 DAVID PAPINEAU
had P* without P. And if P is proportionally ‘just right’, then the less specific
Mwill violate requirement (2), in that we wouldn’t still have had P* if M had
been realized differently. It can’t be the case that a more and less specific
state are both causally proportional to a given effect. (Cf. LePore and
Loewer 1987; Menzies and List 2010.)
Suppose that some mental state M out-competes its more specific phys-
ical realizer P as the cause of some physical effect P*. What then is the
relation of the realizer P to the effect P*? Many of those who defend the
causal status of M nevertheless retain the idea that the physical realizer P is
‘causally sufficient’ for the physical result P*. Their thought is that P still
causally determines P* even though it is too specific to count as ‘the cause’
of that result. But from the perspective being defended here, even this seems
to concede too much to the intuition that causation is grounded in basic
physical processes. Of course we might wish to allow that such specific
physical antecedents are nomologically sufficient for the subsequent physical
results. But there is no reason to think of this sufficiency as a causal matter,
in cases where the precise physical detail omits any mention of the macro-
scopic pattern that constitutes the causal relationship.
6. Proportionality and Reduction
Does the fact mental states can eclipse their physical realizers as causes of
certain effects vindicate the possibility of non-reductive physicalism in the
philosophy of mind? This conclusion is typically drawn by those philoso-
phers who stress that proportionality requirements can favourmental states as
causes over their physical realizers. But it is by no means clear that it follows.
Reductive physicalism requires type identity of mental properties with
physical properties. Non-reductive physicalists maintain that no such type
identities are available. It is important to realize that, in order to establish
non-reductive physicalism, it is not enough to show that there are some
physical differences present on the different occasions where M is realized.
Rather we need to show that there is no distinctive physical commonality
present on all those occasions.
When Putnam and Fodor introduced the idea that mental and other
special science properties might fail to reduce to physical properties, they
weren’t just making the weak claim that different instances of these properties
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 135
will display some physical differences. Rather their idea was that there would
be no common physical feature of different instances. There would be
nothing physically in common between the different computers that can
run a given program, or between the different organisms across the universe
that can think a given thought.
So far in this paper we have been dealing with cases where some mental
M is realized by different more specific physical Ps on different occasions.
That is, we have been dealing with cases where distinguishable physical Ps
can metaphysically determine the same mental M. This by itself fails to
establish the anti-reductive thesis that there is no further physical feature
Q which is type identical to M.
To see that this stronger anti-reductive thesis does not follow from the
fact that M is determined on different occasions by distinguishable Ps, we
need only consider a thermodynamic example once more. Take the case
where the gas reaching temperature T opens the valve. Temperature T can
be realized by many distinct arrangements of specific particle movements.
But it does not follow that there is no further physical property which
characterizes all the instances of T. And of course in this case there is. All the
instances of T involve arrangement of particles with the same mean kinetic
energy. And it is precisely this common physical feature which allows the
possibility of a uniform thermodynamic explanation of why the valve will
open at that temperature. The probability distribution over the possible
microstates that realize that mean kinetic energy implies that the valve is
overwhelmingly likely to open at that temperature.
This is surely the paradigm of a type–type reduction. We identity tem-
perature with some common physical feature specifiable in terms of particle
movements, namely a given mean kinetic energy, and thereby explain
patterns involving temperature in terms of particle physics.
But as well as being the paradigm of a type–type reduction, this is also a
case where proportionality considerations point to the macroscopic tem-
perature as the cause of the valve opening, rather than the more specific
particle movements which realize it on given occasions. I infer that there is
nothing in the idea of macroscopic facts being proportionate causes to rule
out fully reductive physicalism.
Recall Putnam’s famous example of the square peg and the round hole
(1975). Putnam argued that the properties of squareness and roundness will
be a much better explanation for why the peg does not fit in the hole than
136 DAVID PAPINEAU
any detailed specification of the quantum mechanical arrangement and
properties of relevant bodies’ molecules. Quite so. It is the squareness and
roundness that are proportional to the peg’s failure to fit, not the very
specific molecular arrangements that realize these properties.
But this does not mean that squareness and roundness are not physically
reducible. We can still specify features of their molecular arrangements that
will be common to all square pegs and round holes, and we can appeal to the
so-specified features to explain at that level why square pegs don’t go into
round holes. Here again we see that it does not follow from the causal
dominance of macroscopic facts over their more precise realizers that those
macroscopic facts must be physically irreducible.
These examples manifest a typical set-up in physics. Some macroscopic
property common to many microscopically distinguishable states can be
identified with some common feature of those microscopic realizers, and
this common feature then accounts for the way that the macroscopic
property features essentially in some general pattern.
7. Against Unreduced Causes
The last section showed that macroscopic causes are one thing,
non-reducibility another. There are plenty of cases where macroscopic
properties can feature as proportionate causes of certain physical effects,
and thereby causally eclipse their more specific microphysical realizations,
and yet these macroscopic properties are fully reducible to some common
physical feature of their microscopic realizations.
I now want to argue that macroscopic causation is not just consistent with
physical reducibility, but that it positively requires this.
I have argued in this paper that asymmetric causal relations derive from
probabilistic facts about the way in which macrostates are realized at the
micro-level. This picture assumes that each macro-cause corresponds to
some constraint specifiable at the micro-level, in the way that temperature
corresponds to mean kinetic energy. The probabilistic facts about the
different ways this condition can be satisfied by precise microstates then
accounts for the asymmetric causal patterns involving macrostates.
If this is the right general story about causation, then it is hard to see how
macroscopic causes can fail to be physically reducible. Their very nature as
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 137
causes will derive from their type-identity to some physically specifiable
constraint, for it will only be in virtue of this identity that they systematically
generate their effects.
What options are open to non-reductive physicalists here? There seem to
be two ways they might go. First, they might argue that, when a given
macroscopic cause is variably realized, it generates its effects via different
causal processes at the physical level. Alternatively, they might argue that
there is no need to invoke any casual processes at the physical level to
explain how a variably realized macro-cause generates its effect. However,
neither of these options seems at all attractive.
To bring out the difficulties here, note that proportionate causation
involves an element of generality. Recall that our two requirements for
C to cause E were that
(1) E wouldn’t have occurred if C hadn’t occurred, and
(2) E would still have occurred if C had occurred differently.
Clause (2) here tells us that in other similar circumstances where C occurs,
E will occur too. On other similar occasions where I want to hail a taxi, my
arm still moves. And clause (1) tells us that when C doesn’t occur in similar
circumstances, E will fail to occur too. On other similar occasions where
I don’t want to hail a taxi, my arm doesn’t move. In short, there is a general
co-variation of C and E in similar circumstances.
Now, the problem facing non-reductive physicalists is to explain why
C and E should so co-vary if there is no uniform physical condition
corresponding to C which can account for this. The answer is obvious if
C can be identified with some physical condition which systematically
generates the result E. But in the absence of any such identification, non-
reductive physicalists seem to face a challenge.
The first non-reductive response to this challenge would be to hold that
the different realizations of C give rise to E via different causal processes. This
is probably how non-reductive physicalism is normally understood. On
different occasions when people want to hail a taxi, their desire is realized
by different physiological arrangements—but each of these different physio-
logical arrangements has the causal power to produce an arm movement.
The trouble facing this option is that we have been given no account of
why the different causal processes that realize C should all alike be ones that
give rise to E. For all that has been said so far, this looks like a mystery. If the
138 DAVID PAPINEAU
processes at the physiological level are all so different on different occasions
of desiring to hail a taxi, why ever should they all be followed by E?
To focus this issue, it will be helpful to consider an inorganic example, as
there are features of mental phenomena that can obscure the difficulty at
hand, in various ways to be considered below. Let us imagine that the water
from a certain lake seems to have a distinctive power to destroy rubber. But
when we look into the mechanism, we find no common causal process. In
one case, the water contains rubber-eating bacteria. In another, the water
turns out to be highly acidic. In yet another, there are high levels of ozone in
the water and this produces a rubber-destructive agent. And so on. In each
of the cases that we examine, we find a physical explanation for the rubber’s
deterioration, but the explanation is different in each case.
I take it that this story does not hang together. If we really came across a
case like this, and discovered a different mechanism in each case, we would
surely conclude that it wasn’t a genuine causal relationship after all, and that
the observed pattern was just a coincidental feature of the cases so far
observed. I think that we should have the same reaction to the suggestion
that some mental C can produce a physical effect E via different causal
mechanisms on different occasions. In the absence of any further infor-
mation, it seems incredible that nature should work like this.
Let me now consider the alternative non-reductive response to the
challenge of explaining why some C and E should co-vary if there is no
uniform physical condition corresponding to C which can account for this.
The alternative non-reductive option would be to deny that we need any
causal accounts at the physical level for macroscopic causal processes. Now
the idea is not that different causal processes account for the C–E link in
different cases, but that there are no further causal stories to be told at the
physical level at all.
But this too looks like mystery-mongering. Remember that we are
exploring the possibility of non-reductive physicalism. It is not as if we are
positing some ontologically independent realm of mental causes with brute
powers to produce physical effects. Rather, we are taking it that mental and
other macroscopic causes metaphysically supervene on the physical facts—
nothing more is required for their presence than those physical facts. But
then causation without a physical explanation looks like a conspiracy. On
different occasions C is realized by different physical microstates, and
somehow these all evolve into later microstates that determine E. But
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 139
there are no conditions satisfied by the initial microstates that might account
for their all evolving into states that determine E. I see no reason to accept
that there are macroscopic causal patterns which correspond to no causal
patterns at the physical level in this way. Imagine that temperatures super-
vened on molecular motions, and that certain temperatures produced
certain regular effects, but that there was no uniform story available at the
molecular level of why this should be so. This doesn’t seem the way that
things work in our world.
Note that I am not accusing non-reductive physicalism of any outright
inconsistency. There is nothing contradictory in the idea that the physical
realizers of some C should all just happen to eventuate in some E, either via
different causal paths at the physical level, or via no such casual paths at all.
But I take it that our experience shows us that the world just doesn’t work
like that. Macroscopic causal patterns do not depend on massive coinci-
dences at the physical level. Rather any macroscopic cause corresponds to a
common physical condition satisfied by its realizers, and there is a physical
story to be told about why this condition gives rise to the relevant effect.
8. Causes not Laws
In a number of previous papers (Papineau 1985, 1992, 2010) I have offered a
similar argument against non-reductive physicalism. However those earlier
papers focused on laws, not causes. My earlier arguments owed nothing to
the way that asymmetric causation depends on probability distributions over
the microstates consistent with a given macro-condition. Rather I simply
appealed to our supposed knowledge of the way general laws of any kind
depend on uniform physical processes, arguing that it would be incredible
that there should be a law involving physically supervenient properties, yet
no uniform physical account of the way that the physical realizations of the
initial condition evolve into physical realizations of the final condition.
However, this line of argument can be criticized for its appeal to the
unexplained notion of a ‘uniform’ physical account. Take the full range of
nomologically possible physical microstates that can realize the relevant
initial macrostate. Now suppose that those initial conditions evolve
according to the basic dynamical law (Newton’s second law in a classical
context, or Schrodinger’s equation in a quantum context). The upshot will
140 DAVID PAPINEAU
be that this range of initial microstates will be shown to end up in a range of
later microstates that determine the relevant final macro-condition. On
what grounds do I say that this is not a ‘uniform’ account of the original
macroscopic law? It can’t just be that the basic dynamic law is being applied
to a range of distinguishable microstates. That would rule out pretty much
anything as a uniformly explained law, including any laws of thermodynam-
ics, chemistry, or planetary motion.
Yet in my earlier papers I offered no other account of what might render
a physical account non-uniform. Given this, it is unclear what force there is
to my insistence that it is ‘incredible’ that there should be macroscopic laws
that lack a uniform physical account. In the absence of some further
explanation of what counts as uniform, what exactly is it that I say I find
incredible?3
Of course, this is not an objection that can happily be made by those who
want to define themselves as non-reductive physicalists, since they too will
need to appeal to a notion of ‘uniform’. This is because they want to insist
that the application of the basic dynamic law to the collection of initial
conditions in not a uniform physical reduction, but a derivation that covers a
physically heterogeneous range of cases. Still, the point remains that, with-
out some further account of what counts as a ‘uniform’ physical process, it is
unclear what substance there is to my dispute with non-reductive physical-
ism. It seems as if there may be nothing at issue when I claim that macro-
scopic laws must be physically reducible, and they deny this.
I take this paper to add substance to my position by focusing on causal
processes. I no longer wish to argue that all laws are physically reducible—I
concede that there may be no good sense in which all macroscopic laws
must have a uniform physical reduction. Rather my focus is now specifically
on asymmetric causal patterns, and my claim is that for any such causal
pattern, there will be a constraint specifiable at the physical level common to
all realizations of the cause, and that a probability distribution over the
microstates satisfying this constraint will play a part in explaining why the
effect follows. This is what I mean by a uniform physical explanation for a
causal pattern. So my present thesis is that there are no causal patterns in our
world that lack uniform explanations of this kind.
3 Barry Loewer has pressed me on this point in conversation on a number of occasions.
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 141
9. Does Functionalism Help?
It might have occurred to some readers to wonder whether a functionalist
account of mental states might not help to explain how a given mental cause
may produce its results via different physical-level causal processes on
different occasions. If mental states are defined functionally as states which
produce certain effects, then won’t such variable causal mechanisms be just
what we would expect? Suppose the mental state of wanting to hail a taxi is
defined as a state that will produce arm movements or similar signals. There
may well be lots of different physiological states that satisfy this requirement.
But it will scarcely be ‘incredible’ that they should all alike give rise to arm
movements—for it is just this tendency that qualifies them as realizations of
wanting to hail a taxi in the first place.
However, the appeal to functionalism does not help. As is well known,
there are two very different versions of the functionalist thought that mental
states can be ‘defined’ as states which produce certain effects. Once they are
clearly distinguished, we can see that neither of them helps non-reductive
physicalism to explain how one state can cause another via different causal
processes. The impression that functionalism helps with this problem only
arises if the two versions are run together.
The first version—realizer functionalism—is a thesis about how the
reference of mental terms is fixed. On this view, mental states are physical
states that are identified via their connection to certain causes and effects.
For example, the term ‘desire to hail a taxi’, applied to some person, is to be
understood as referring to that physical state which causes appropriate arm
movements in that person. This term might thus refer to different physical
states in different people, just as the term ‘your watch’ might refer to
different devices when different people are being addressed.
This view does nothing to explain variably realized causes, for the simple
reason that it does not trade in variably realized states of any kinds. The only
states it countenances are ordinary physical states—such as the physical state
which causes my arm to move, say—and these physical states cause their
effects in an ordinary uniform manner. True, different physical states may
well be picked out by the same mental word, in virtue of producing some
common effect—but there is nothing here to suggest that any given such
state produces its effect via different routes on different occasions.
142 DAVID PAPINEAU
Then there is role functionalism. This does recognize states which are
variably realized at the physical level. Role functionalism takes mental terms
to refer, not to the first-order physical states that have certain specified
causes and effects, but to the second-order states of having some first-order
state that plays that causal role. On this view, mental terms will have the
same referent even when applied to differently constituted beings: they refer
to the second-order state shared by all beings who instantiate the relevant
causal role.
We can usefully bring out the difference between realizer and role
functionalism by thinking of a term like ‘dormitive virtue’ as applied to
sleeping pills. The realizer option would take this term to refer to the
narcotic chemical constituent present in whichever sleeping pill is under
discussion. The role option, by contrast, would take the term to refer to the
property common to all sleeping pills, namely, their tendency to produce
sleep by whatever means.
Now, role functionalism does arguably give us variably realized states.
Just as a tendency to produce sleep can be realized by different chemical
processes, so can a tendency to move one’s arm be realized by different
physiological processes. The trouble is that, if mental states are like tenden-
cies to produce sleep, then surely they are disqualified as causes of the effects
that constitute them. A tendency to produce sleep isn’t sufficiently distinct
from the sleep itself to qualify as its cause. Similarly, if a desire to hail a taxi
constitutively requires appropriate arm movements, it isn’t distinct enough
from the arm movements to cause them.
So whichever way we turn functionalism, it doesn’t give us causes which
produce their effects via non-uniform physical processes. Realizer function-
alism gives us causes all right, but they operate in a physically uniform
manner. Role functionalism gives us variably realized states all right, but
they aren’t causes of the relevant effects.
10. Selectional Properties
Perhaps a different kind of functionalism can account for variably realized
causes. Rather than considering states that are defined or constituted by a
causal role, let us instead consider states that are functional in the sense that
they have been selected to play some causal role.
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 143
The puzzle I have been pressing so far is how some physically superven-
ient putative cause C can regularly co-vary with some putative effect E if
there is no common feature at the level of its physical realizations to account
for this. In the absence of any such commonality at the physical level, it
seems mysterious that E should generally follow.
But now suppose that the instances of C have been selected because they
produce result E. That is, they occur as the result of some selection process
that favours items that produce E. Then the puzzle would be explained.
There would be an explanation for why C generally leads to E even though
there is no uniform explanation at the physical level. E generally follows
because different instances of C have been selected to produce precisely
that result.
To illustrate, consider the simple example of thermostats in electrically
controlled domestic hot-water heating systems. Any such system contains a
thermostat which stops the heating once the water reaches some set tem-
perature. But these thermostats involve a range of different mechanisms at
the physical level, including bi-metallic strips, expansion gases, mercury
bulbs, and thermocouples. Yet there is clearly no puzzle here as to why
these different kinds of thermostat always produce the same effect of
stopping the heating. Their mechanisms have been selected by the heating
designers precisely in order to produce this effect.
So maybe this is a model for unreduced causes. Take the property, in a
heating system, of containing a thermostat. Let us suppose that this property
does not constitutively involve the effect of stopping the heating, and so is a
candidate for causing that effect. Won’t this now amount to a case where
this physical effect is caused by a variably realized property, namely the
property of containing a thermostat?4
Certainly the counterfactuals seem to vindicate this claim. It is the generic
presence of a thermostat per se, rather than the specific mechanism that
realizes it in a given case, that comes out as proportional to the effect of
4 The most natural way of construing selectional properties like being a thermostat is as constitutively
involving some past history of selection. So understood, selectional properties arguably won’t consti-
tutively involve their effects as role properties do—something can be selected to do F and yet have no
tendency to do F in the future. Even so, selectional properties might be held to be ineligible as causes on
the different grounds that historical provenance cannot matter to causal significance. I shall not press this
particular worry, however. Perhaps it can be avoided by construing selectional properties in some way
that disconnects them from their history. But even if it can be so avoided, the causal status of selectional
properties would still be open to the more fundamental objection made below.
144 DAVID PAPINEAU
stopping the heating. If (1) there hadn’t been a thermostat, the heating
wouldn’t have stopped. And if (2) the thermostat had been realized differ-
ently, the heating would still have stopped. By contrast, the specific mech-
anism does not seem proportional. While it is true (2) that a differently
realized bi-metallic strip would still have stopped the heating, it isn’t true (1)
that if there hadn’t been a bi-metallic strip, the heating wouldn’t have
stopped—because in that case a different design of thermostat would no
doubt have done the job instead.
Of course mental systems are not designed by intelligent engineers in the
way that heating systems are. But, to the extent that they are designed by
phylogenetic and ontogenetic selection processes, the same moral will
apply. These selection processes will ensure that there is some mental
component available to produce a given effect, but the precise mechanism
that does this may vary from case to case.
Thus consider pain across different species. Intergenerational genetic
selection will have ensured that all organisms have some mechanism that
responds to bodily damage by seeking to avoid the source of the damage.
But it may well have lit on different things to do this job in different species.
Nor is the point restricted to the way that the products of intergenera-
tional genetic selection can vary across species. Humans and other complex
animals are sophisticated learning machines that embody a hierarchy of
processes that operate to preserve items that produce such-and-such effects.
The items selected may well be physically different in different individuals,
or even in the same individual at different times, but this won’t matter to the
selection mechanisms, provided they produce the reinforcing effects. The
state which leads me to hail a taxi when one is needed may be quite
differently realized in me and in you, but we are both likely to possess
some such state.
Just as with the thermostats, proportionality considerations again suggest
that such selectional mental states can qualify as variably realized causes
of physical effects in their own right. Consider again the state of wanting
to hail a taxi, and the effect of my arm moving. If (1) I hadn’t wanted to hail
a taxi, my arm wouldn’t have moved. And (2) if this desire had been realized
differently, my arm would still have moved. But now consider the specific
brain state that realizes the desire in me. While it is true (2) that if this brain
state had been realized differently my arm would still have moved, it
isn’t true (1) that if I hadn’t had that brain state, my arm wouldn’t have
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 145
moved—because in that case a different brain state would have been
selected to move my arm instead.
11. Too Many Causes
This might all now look like good news for unreduced mental causes.
However, I think that appearances are deceptive. Despite the points made
above, there is a strong reason to doubt that selectedmental items, and indeed
selected items generally, can feature as non-reduced causes in their own right.
To the extent that the proportionately counterfactuals argue differently, I say
that these counterfactuals are misleading as to causal structure.
Let us ask why certain physical states are selected to play a certain role in a
cognitive structure or other designed system. The answer is that these states
are apt to cause some specific effect, and the relevant selection mechanism
favours items with this feature. However, if this is the reason why these
physical states are selected, it rules out the more generic variably realized
selectional state from also causing that effect.
Take the thermostat example again. If a heating engineer chooses to put a
bi-metallic strip into the electric circuit, this is because this itemwill cause the
circuit to break when the temperature rises. It is precisely the causal status of
this item that renders it suitable for the engineer’s purpose. But this then
undermines the thought that having a thermostat per se causes the circuit
breaking. This generic property is common to different kinds of circuits, and
in each of these the breaking is caused by a different mechanism. Having a
thermostat itself does not cause the result, for having a thermostat depends on
being in some more specific state which does cause the result.
The same point applies to selectedmental causes.Whyhave phylogenetic and
ontogenetic selection processes picked certain brain states for the wanting-to-
hail-a-taxi-role? Because those brain states get activated when a taxi is needed
and they then cause arm waving or similar movements. The relevant selection
mechanisms will favour just those brain states that have this causal profile. But
this again argues that the selectional property of wanting-to-hail-a-taxi cannot
itself cause anything. Wanting-to-hail-a-taxi involves being in a brain state
which itself causes arm waving and so on. It was because this brain state already
caused this result, so to speak, that it was selected. Given this, it makes little sense
to think of the generic selectional state as also causing the result.
146 DAVID PAPINEAU
Note that this analysis does not appeal to some unthinking intuition that
more specific physical states always casually out-compete any more generic
states that supervene on them. As the earlier sections of this paper will have
made clear, I regard this intuition as fundamentally misguided. Rather
I have a more particular objection to viewing generic selectional states as
causes. This objection derives from the structure of the selection processes
that account for such selectional states, and in particular which explain how
they can produce uniform effects despite being variably realized. Selection
processes operate on causal facts. Their workings hinge essentially on the
causal properties of the items selected. They preserve items that cause certain
effects (see Papineau 2003). This is why we are forced to accept that it is
these realizing items that cause those effects, and not the generic selectional
states that supervene on them.
What about the counterfactuals? As we saw, they do seem to indicate the
generic selectional states as causes, in preference to the more specific mech-
anisms that realized them. If we hadn’t had the generic state, we wouldn’t
have had the result. By contrast, it’s not true that the result wouldn’t have
occurred if we hadn’t had the specific realizing mechanism—since in that case
some other item would no doubt have been selected to produce the result
instead. So it looks as if the generic state is proportional to the effect, rather
than the specific realizer.
However, I take this to be analogous to the many familiar cases where the
counterfactuals fail match causal structure because of back-up arrangements.
When I make an assassination plan with a contingency arrangement (for
example, a back-up assassin lest the first one fail), it is the whole plan that is
proportional to the death of the prisoner, not the shooting by the first
assassin. Yet it is that first assassin that caused the death, and the back-up
assassin played no causal part. Similarly in the cases at hand: it is the selec-
tional state of being designed for some end that is proportional to the effect,
but the specific mechanism that fulfils the design that actually causes it.
12. Explanation is Different
Variably realized selectional states may not cause physical effects, but this
does not mean that they cannot be used to explain them. It will be worth
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 147
clarifying this issue, in order to forestall any inference from the explanatory
significance of variably realized states to their causal efficacy.
In this connection, note first that we can often refer to genuine causes
indirectly, by citing variably realized selectional states like wanting to hail a
taxi. This is possible because we can use descriptions involving the selec-
tional states to construct variable names for the genuinely causal physical
states. The way this works has already been discussed under the heading of
‘realizer functionalism’. We saw there howwe can read ‘dormitive virtue’ as
referring to the specific chemical property present in whichever sleeping pill
is under discussion. Similarly, we can understand ‘wanting to hail a taxi’ as
referring to the specific brain state that makes the relevant subject’s arm
move. So understood, claims like ‘he fell asleep because he took a pill with
dormitive virtue’, or ‘his arm waved because he wanted to hail a taxi’, will
state causal truths.
Not only will such claims state causal truths, but they can also be
explanatory. Explanations of particular facts need to name causes. But
they can do so indirectly, using descriptions involving the selectional status
of those causes. As long as this mode of reference shows us how those causes
fit into patterns that can be used to anticipate and control, the attribution of
causes will be explanatory.
Thus it can certainly be explanatory to say that someone fell asleep
because he took a pill with a dormitive virtue (as opposed to having had a
very tiring day, say). Similarly, it can be genuinely explanatory to say that
someone’s arm waved because they wanted to hail a taxi. (Not all indirect
references to causes are explanatory. It is not explanatory to say I fell asleep
because I was caused to fall asleep. We need to cite the cause in a way that
fits it into a practically significant pattern.)
Selectional states of all kinds are very commonly cited in explanations.
I might explain the high temperature in the room by the setting on the
thermostat, or the improved performance of my car by its new carburettor.
I may have no idea of the actual mechanisms in either case. But knowledge
of design properties tells me how the relevant items will work and so suffices
for explanatory purposes. Similarly with mental explanations. The states we
cite may be variably realized selectional states which are not themselves
causes, but they can be genuinely explanatory for all that.
148 DAVID PAPINEAU
13. Causal Closure
The principle of the ‘causal closure of the physical’ has played a significant
role in recent philosophy of mind. According to this principle, every
physical effect must have a physical cause. It is this principle that lies behind
the widespread modern acceptance of physicalism (Papineau, 2002). It
allows us to argue that any non-physical realm can only be epiphenomenal,
since it would generate an unacceptable overdetermination of physical
effects to attribute them to non-physical causes in addition to the physical
ones already guaranteed by closure.5
It is tempting to infer the falsity of the closure principle from the
possibility of macroscopic causes. If, as proportionality considerations
argue, macroscopic causes can out-compete the more specific realizers as
causes of certain physical effects, does this not show that those physical
effects at least will have macroscopic causes rather than physical ones, and
therefore that the physical realm is not causally closed? (Cf. Menzies and List
2010.)
Rejecting causal closure would have a cost. Without a principle of causal
closure, we would be left with no argument against interactive Cartesianism
and other strong forms of dualism. Fortunately, the possibility of macro-
scopic causes does not refute causal closure. This would only follow if
macroscopic implied non-physical. I have argued that it does not. To repeat
my standard example, temperature is a macroscopic property, but it can be
type-identified with the physical property of mean kinetic energy.
Moreover I have argued that, not only is macroscopicity consistent with
physicality, but that macroscopic causation positively requires physicality. If a
macroscopic cause cannot be type-identified with a physical property, we
can’t give a uniform explanation of why the same physical effect always
follows from its different realizations. Nor does it help to appeal to selection
processes to explain this, for it is built into the nature of selection that the
5 The literature displays different uses of this argument. Some use it only to rule out forms of dualism
on which the mental realm does not even supervene on the physical. But others, most prominently
Jaegwon Kim, also use it to argue against ‘non-reductive physicalisms’ that respect supervenience but
deny type identity. The latter form of argument assumes that any kind of overdetermination is unaccept-
able, even when one cause supervenes on the other. The former can allow such supervenient overde-
termination, and need assume only that overdetermination by metaphysically distinct causes is
unacceptable.
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 149
relevant effects are caused by realizing mechanisms, rather than by the
generic selectional states that these mechanisms determine.
So my overall analysis reinforces the causal closure of the physical. Cer-
tainly many physical effects should be attributed tomacroscopic causes rather
than their more specific realizers. But these macroscopic causes will still
always be physical, thus upholding the principle that every physical effect
must have a physical cause, and leaving the argument against dualism intact.
14. Mental Causes
One last point. In the latter half of this paper I have been arguing that
variably realized mental states cannot be causes. But this does not of course
mean that mental states as such can never be causes. For there remains the
possibility that some mental states can be type-identified with physical
states, in the way that temperature is type-identified with mean kinetic
energy.
I have paid little attention to this possibility so far, given that my main
concern has been to establish that variably realized states cannot be causes.
But the physical type identity of at least some mental causes is a serious
option. Remember that type identity does not require that there can be no
physical differences between the bearers of a given mental state, just that
there should be some physical commonality which might explain why the
state regularly produces certain effects.
It seems very likely that a wide range of mental states are so uniformly
realized within humans, and indeed across many of the other taxa to which
we belong. For example, there is every reason to suppose that the pain
mechanism is uniformly realized across humans and similar mammals.
Again, many sensory mechanisms can be expected to be physically uniform
in this way. Perhaps the basic mechanisms of learning and reasoning will also
be uniformly realized in all humans, even if not in other species. Provided
that we understand our mental terms for these categories as indexed to the
appropriate range of species, we can read them as referring to physically
reducible types, and hence to fully causal states.
On the other hand, I accept that many other mental states will be variably
realized across humans. These will be states which derive from ontogenetic
selection processes. For example, I would expect wanting to hail a taxi to be
150 DAVID PAPINEAU
variably realized within humans, and even perhaps within individuals. States
like these will thus not be causally efficacious, even though they can be
explanatory significant in the way explained above.
It is an intriguing question which states are which. For everyday explana-
tory purposes the difference may not matter much, given that both kinds
can equally be invoked in explanation. But the contrast will be significant
for cognitive science. Investigation of the physical nature of physically
reducible states could bring important scientific benefits, but a similar
investigation of variably realized states would inevitably be fruitless. Cogni-
tive science thus needs to know which mental states are causal in their own
right, and which play only an explanatory role.6
References
Albert, D. (2000). Time and Chance. Cambridge, MA: Harvard University Press.
Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just
Maybe, to Tract It’. Nous, 37: 471–97.
Hausman, D. (1998). Causal Asymmetry. Cambridge: Cambridge University Press.
Kim, J. (1993). ‘The Non-reductivist’s Troubles with Mental Causation’. In J. Heil
and A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 336–57.
LePore, E. and B. Loewer (1987). ‘MindMatters’. Journal of Philosophy, 84: 630–42.
List, C. and P. Menzies (2009). ‘Non-Reductive Physicalism and the Limits of the
Exclusion Principle’. Journal of Philosophy, 106(9): 475–502.
Menzies, P. (2008). ‘Causal Exclusion, the Determination Relation, and Contrast-
ive Causation’. In J. Kallestrup and J. Hohwy (eds.), Being Reduced: New
Essays on Reductive Explanation and Special Science Causation. Oxford:
Oxford University Press: 196–217.
—— and C. List (2010). ‘The Causal Autonomy of the Special Sciences’. In
G. and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University
Press: 108–28.
Michotte, A. (1946/1963). The Perception of Causality. English translation 1963 by
E. and T. Miles. New York: Basic Books.
Papineau, D. (1985). ‘Social Facts and Psychological Facts’. In G. Currie and
A. Musgrave (eds.), Popper and the Human Sciences. Dordrecht: Nijhoff: 57–71.
6 I would like to thank Eleanor Knox, Barry Loewer, Laurie Paul and David Yates for valuable
comments on earlier drafts.
CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 151
Papineau, D. (1992). ‘Irreducibility and Teleology’. In D. Charles and K. Lennon
(eds.), Reduction, Explanation and Realism. Oxford: Oxford University Press:
45–68.
—— (2001). ‘Metaphysics over Methodology—or, Why Infidelity Provides no
Grounds to Divorce Causes from Probabilities’. In M.-C. Galavotti, P. Suppes,
and D. Costantini (eds.), Stochastic Causality. Stanford: CSLI Publications: 15–38.
—— (2002). Thinking about Consciousness. Oxford: Oxford University Press.
—— (2003). ‘Causation as a Guide to Life’. In The Roots of Reason: Philosophical
Essays on Rationality, Evolution, and Probability.Oxford: Oxford University Press:
167–211.
—— (2010). ‘Can Any Sciences be Special?’ In C. and G. Macdonald (eds.),
Emergence in Mind. Oxford: Oxford University Press: 179–97.
Pereboom, D. (2002). ‘Robust Nonreductive Physicalism’. Journal of Philosophy,
99: 499–531.
Putnam, H. (1975). ‘Philosophy and Our Mental Life’. In Mind, Language and
Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University Press:
291–303.
Shoemaker, S. (2001). ‘Realization and Mental Causation’. In C. Gillett and
B. Loewer (eds.), Physicalism and Its Discontents. Cambridge: Cambridge University
Press: 74–98.
Sosa, E. andM. Tooley (1993). ‘Introduction’. In Sosa and Tooley (eds.),Causation.
Oxford: Oxford University Press: 1–32.
Spirtes, P., Glymour, C., and Scheines, R. (2000). Causation, Prediction, and Search.
Cambridge, MA: MIT Press.
Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.
152 DAVID PAPINEAU
6
Substance Causation, Powers,and Human Agency
E. J. LOWE
The powers of individual substances may, I believe, be distinguished in at
least the following two ways. First, some of these powers are causal powers,
while others are non-causal powers. Second, some of these powers are active
powers, while others are passive powers. But all powers, as we shall see, are
individuated by their manifestation types, that is, by the characteristic types of
activity that constitute their exercise. A causal power is one whose exercise
consists in the bearer of the power acting on one or more substances to bring
about a change in them. A passive power is one whose exercise is always
caused by one or more substances acting upon the bearer of the power. This
classification of powers leaves open the possibility of there being a type
of power that is at once active and non-causal: a power whose exercise is
(1) not caused by any substance acting upon its bearer and (2) does not
consist in its bearer bringing about a change in any substance. Such a power
may be called a spontaneous power. It seems clear that such powers do exist in
nature, the power of a radium atom to undergo radioactive decay being an
example. This fact shows that there need be nothing anti-naturalistic, or
incompatible with current physical science, in supposing that the human
will, as it is exercised in episodes of voluntary action, is another such power.
In saying that the will is a non-causal power, it is not being implied that the
will is causally inefficacious, only that its exercise does not consist in the
agent’s bringing about any sort of effect. Agent causalists who suppose that
agents cause their own volitions by exercising agent-causal powers are,
I believe, mistaken in this regard and mistaken too if they think that their
view explains the special sense in which free agents have control over their
voluntary actions. What, in my view, distinguishes the will from any other
kind of spontaneous power is (1) that it is a two-way power—a power either to
will or not to will a particular course of action—and (2) that it can be exercised
rationally, that is, ‘in the light of reason’. The possession of such a power
would, I believe, give human agents all the control that they could need or
want over their voluntary actions. And very arguably, as we shall see, we
cannot—on pain of undermining our entitlement to regard ourselves as
rational beings—deny that we have such a power.
Individual substances
Individual substances—as I propose to construe this term—are persisting,
concrete bearers of properties, including powers of various kinds, at least
some of which are causal in character. There cannot, in my view, be any
such thing as a causally inert individual substance. Furthermore, individual
substances are ontologically independent entities, in a sense of ‘independent’
that needs to be carefully defined. The sense in question is this: individual
substances do not depend for their identity on other entities of any kind.1 This
is quite consistent with saying that they may depend for their existence on
various other entities, such as their proper parts (if they have any). I shall
illustrate this point by means of examples in a moment.
Now, ostensibly at least, individual substances include both material objects
of certain kinds (‘bodies’, in a broad sense of that term) and also human
persons, such as ourselves. Items of these two types constitute what
P. F. Strawson called ‘basic particulars’—‘basic’ in the sense that it is ultim-
ately only by reference to items of these two types that we can in general
succeed in identifying concrete particulars of any other types, such as
particular events and particular states, including mental events and mental
states.2 Different kinds of individual substance are primarily distinguished,
moreover, by their distinctive identity conditions and their distinctive powers—
just as John Locke maintained.3 Thus, amongst material objects, mountains
have very different identity conditions and powers from, say, mice. The sorts
of changes through which a mountain can survive identically are mostly very
1 I defend this claim more fully in Lowe (1998), Chapter 6. 2 See Strawson (1959).3 See Locke (1975) [1690], Book II, Chapter 23.
154 E. J . LOWE
different from those through which a mouse can survive identically. And
what a mountain can characteristically cause is very different from what a
mouse can characteristically cause. (By a ‘concrete’ particular, incidentally,
I simply mean a particular that exists in space and time, as opposed to an
abstract particular, such as the number 7.4)
Returning, however, to the point that individual substances do not
depend for their identity on other entities of any kind, consider, for example,
a particular mouse, Mortimer. Mortimer certainly depends for his existence
on other entities, such as the atoms and molecules that compose him at any
given time, in the absence of which he could not exist at all. But, since
Mortimer can survive a change of these atoms andmolecules, his identity does
not depend on their identity: that is to say, whichmouseMortimer is does not
depend on which atoms and molecules compose him at any given time. After
all, Mortimer’s identity is essential to him, but it is not essential to him that he
is composed at any given time of the particular atoms and molecules that do
happen to compose him then, for the latter is a purely contingent circum-
stance. Nor does Mortimer depend for his identity on any event in which he
‘participates’, such as his birth or death—since, on the contrary, any such
event depends at least in part for its identity on him, and identity dependence
is asymmetrical.5 Which death is the death of Mortimer is at least partly
determined by which mouse Mortimer is, not vice versa.
Powers: token and type
When we speak of powers it is important to distinguish carefully between
token powers and power types. By token powers, I mean the particular powers
of individual substances, such as a particular copper wire’s power to conduct
electricity, as opposed to the type of power of which this particular power is a
token, namely, electrical conductivity. Different copper wires share this
same power type, inasmuch as they are all electrically conductive, but each
has a different token power of that type—its own particular power to conduct
electricity. A token power belongs essentially to the individual substance that
is its bearer and cannot be ‘transferred’ to another individual substance.
4 I explain my conception of the abstract/concrete distinction more fully in Lowe (1998), Chapter 10.5 See again Lowe (1998), Chapter 6.
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 155
As I mentioned earlier, powers—both token powers and power types—
may be distinguished in two ways. Some are causal, some non-causal. And
some are active, some passive. Moreover, these two distinctions are mutually
independent, generating a fourfold division of powers to which I shall
return shortly. All powers, however, are individuated at least partly by
their manifestation types.6 This again applies both to token powers and to
power types. Every power, P, is essentially a power to ç, for some distinct-
ive type of manifestation, ç-ing. For example, electrical conductivity is a
power to conduct an electrical current. Again, water-solubility is a power to
dissolve in water. Now, it is important to recognize that there are higher-order
powers—powers to acquire lower-order powers. For instance, magnetiz-
ability is a power to acquire the power of being magnetic, which in turn is a power
to attract ferrous metals. However, in what follows I shall concentrate on first-
order powers, these being powers to act in a certain way.
Now, where power types are concerned, I am happy to say that they are
individuated entirely by their manifestation types—and I consider moreover that
each power type has only onemanifestation type.7 Inmy view, it makes no sense
to say that the very same powermight bemanifested in two quite different ways. If
we are sometimes tempted to think this, it may be because we are apt to confuse
what is properly speaking the manifestation type of a given power with certain
further effects that such a manifestation may give rise to. For example, we may be
inclined to say that the very same power—magnetism—ismanifested both by the
attraction of ferrous metals and, say, by the navigational behaviour of homing
pigeons, which relies on a sensitivity to the earth’s magnetic field. But, in fact, it
appears that homing pigeons can only navigate in this way because they contain
particles of ferrous metals in their sensory systems which are attracted by the
earth in virtue of its magnetism. Their homing behaviour is not amanifestation of
the earth’s magnetism.Rather, the earth’s magnetismmanifests itself by attracting
the particles of ferrous metals in the pigeons’ sensory systems and these systems have
evolved so as to adjust the flight behaviour of the pigeons in response to the
movements of those particles.
As for token powers, however, we need to say that they are individuated
not just by their manifestation types but in addition by their bearers and by
6 I defend this claim more fully in Lowe (2010).7 Here, then, I agree with George Molnar: see Molnar (2003), 195. For further discussion, see Lowe
(2010).
156 E. J . LOWE
their time of existence. This is because token powers, unlike power types,
are concrete particulars. Thus, this particular copper wire’s token power of
electrical conductivity is individuated by its manifestation type—conducting
an electrical current—together with its bearer, the copper wire in question,
and the time (which may be of long or short duration) at which that wire
possesses the token power. In the case of at least some token powers,
including this one, we have to be able to say that they can be manifested
more than once, but also that they need not be manifested even once. That is
why we cannot say that a token power is individuated by its manifestation
tokens, only by its manifestation type in conjunction with its bearer and time
of existence. (Other token powers, such as the token power of a stick of
dynamite to explode, can be manifested only once, since any manifestation
of the token power destroys its bearer. But even in this case there is no
necessity for the token power actually to be manifested at all.) Incidentally,
when I say that an item, such as a token power, is individuated by certain
other items—such as its manifestation type, bearer, and time of existence—I
mean to imply that it thereby depends for its identity on those other items.
Causal and non-causal powers
At this point it is appropriate to return to the topic of the fourfold division of
powers alluded to earlier. A causal power, as I shall construe this term, is one
whose manifestation or ‘exercise’ consists in its bearer’s acting on one or
more other individual substances (or sometimes on itself) so as to bring about
a certain kind of change in them (or it). Take, for instance, water’s power to
dissolve salt. Dissolving something is a matter of causing that thing to become
dissolved. And here, most importantly, we have a case of what I would call
substance causation. More particularly, we have a case of one substance, water,
exercising one of its distinctive causal powers by bringing about a certain
kind of change in another substance, salt.
My own view is that, fundamentally speaking, all causation is substance
causation, because only substances strictly and literally possess causal powers.8
If events and properties, for example, are sometimes said to ‘have’ causal powers,
this can in my view only be understood to be the case in a derivative or
8 For earlier defence of this claim, see Lowe (2008), Chapters 6–8.
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 157
secondary sense. For causing is a kind of action—abringing about of change—
and events and properties cannot literally act: only substances can do that.
Events may be the effects of action—what is brought about when an agent
acts—since they just are changes. But they are never literally agents themselves
and so can never literally cause anything. If we are often inclined to speak as if
they do, this should be understood as no more than a facon de parler.
We might say, for instance, that the explosion of the stick of dynamite
caused the collapse of the building. But really, in my view, this is just an
elaborate way of saying that the stick of dynamite, by exploding, caused the
building to collapse. It is the dynamite that literally possesses the destructive
power, not the explosion. To treat the latter as a powerful particular is
indulge in an illicit hypostatization: the treatment of a non-substance as if it
were substance. This is not, of course, to deny the relevance of the explosion
to the kind of change that was brought about. If the detonator had failed, the
building would have remained intact. Substances can bring about effects
only by acting in appropriate ways: but it is nonetheless the substances that
bring those effects about, not their actions.
Now, in contrast with a causal power, such as water’s power to dissolve salt,
we have also non-causal powers. An example would be the power of a spherical
object to roll down an inclined plane. The manifestation of this power—the action
of rolling down an inclined plane—does not consist in its bearer (the spherical
object) bringing about any distinctive kind of change in anything (not even in
itself). It simply consists in a certain kind of translational motion—a movement
from one location to another. Of course, by so moving the spherical object
might cause some change to occur in another object: for instance, itmight cause
an object in its path to be crushed. But the important point is that any such effect
is not part of themanifestation of the power in question, anymore than a pigeon’s
homing behaviour is part of a manifestation of the earth’s power of magnetism.
The proof of this is that the spherical object could roll without anything’s
actually being crushed by it. By contrast, water obviously could not manifest
its power to dissolve salt without some salt’s actually being dissolved by it.
Active and passive powers
Having explained the distinction between causal and non-causal powers,
I come next to the distinction between active and passive powers. A passive
158 E. J . LOWE
power, as I propose to use this term, is one whose manifestation or exercise
always needs to be caused by one or more substances acting on its bearer (and
thereby exercising their causal powers). For example, salt’s water-solubility
is a passive power—we might alternatively call it a ‘liability’—because its
manifestation type, which is dissolving in water, has to be brought about by
some water exercising its power to dissolve the salt in question, by causing it
to dissolve.
An active power, by contrast, is one whose characteristic manifestation
never needs to be ‘triggered’ in this way. In the current jargon, such a power
has a manifestation type, but no stimulus type. An example would be the
power of a radium atom to undergo spontaneous radioactive decay.
Radium has a characteristic half-life, implying that there is a certain object-
ive probability or chance of any given radium atom decaying within a
specified interval of time—a probability that is the same for all radium
atoms, no matter how long such an atom may have been in existence. No
external circumstances or conditions can affect this probability. When such
an atom decays, then, this isn’t a matter of probabilistic causation. That is to
say, it isn’t a case of anything’s raising or fixing the chances of the atom’s
decaying, since those chances are already fixed by the nature of the atom
itself, independently of any external conditions that it may happen to find
itself in. Rather, in such a case, there is no causation at all. This means, of
course, that causal determinism—the doctrine that every event is either
causally necessitated, or at least has its chances of occurrence fixed, by
antecedent events—is false. As we shall shortly see, this has important
implications in the domain of mental causation and voluntary action.
The fourfold classification of powers
It may be helpful if I present here in diagrammatic form the fourfold
classification of natural powers that I have just been describing, providing
in the case of each basic type of power a paradigmatic example. This is set
out in Figure 6.1.
A few words of further explanation are perhaps called for. Matter’s power
of gravitational attraction is clearly a causal power, since its manifestation
consists in the causal activity of attracting other matter. But it is also an active
power, in the sense defined earlier, since its manifestation does not need to
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 159
be ‘triggered’ or ‘stimulated’ by anything acting on the bearer of the power.
This is because all matter, by its very nature, is always exercising or mani-
festing this power.Water’s power to dissolve salt, by contrast—although it is
likewise a causal power—is a passive power, precisely because its manifest-
ation needs to be ‘triggered’ by the introduction of some salt into any body
of water that possesses the power.
The will as a spontaneous power
As we have just seen, there can be and in fact are powers in nature that are at
once active and non-causal—the radioactivity of a radium atom being an
example. This is a non-causal power because its manifestation just consists in
a change in the properties of the atom, not in the atom’s causing a change in
anything else (or even in itself). Now, I want to say that the human will is
another such active, non-causal power—in short, what I propose to call a
spontaneous power.9 Locke, of course, likewise held the will to be a power,
with volitions or acts of will as its manifestations or exercises.10 He took
volitions to be a species of ‘thoughts’, in the broadest sense of that term—a
kind of ‘inner command’ of the mind to itself. But thinking is not a causal
activity. It is not a matter of bringing about some effect, not even in one’s
own mind. Of course, this is not to deny that thoughts may have effects,
or (as I would prefer to put it) that by thinking we can sometimes cause
Causal Non-causal
E.g., matter’sgravitational power
of attractionActive
Passive
E.g., radium’spower of spontaneous
radioactive decay
E.g., water’spower to dissolve
salt
E.g., a sphere’spower to roll downan inclined plane
Fig. 6.1
9 I propose this in Lowe (2008), Chapter 8. 10 See Locke (1975) [1690], Book II, Chapter 21.
160 E. J . LOWE
something to happen. It is just to say that thinking does not consist inmaking
something happen. (Clearly, it would be absurd to say that thinking is a
matter of causing thoughts to happen or occur in our minds. A thought just is
an episode of thinking and hence to say that thinking consists in causing
thinking to occur would result in either circularity or an infinite regress.)
The same applies, more specifically, to willing. I may, by willing, cause my
arm to rise. But my willing in this instance does not consist inmy causing this
(or indeed anything else) to happen. That is demonstrated by the fact that
I may will to raise my arm and yet my arm may still fail to rise. If my willing
to raise it consisted in my causing it to rise, my willing could not occur
without my arm’s rising.
Here, however, it may be asked how it can be the case that, by willing, an
agent can cause something to happen, even though the will is not a causal
power. But we have already seen that something similar occurs in other
cases of substance causation. Thus, a spherical object’s power to roll down
an inclined plane is a non-causal power and yet, by so rolling, the object can
cause another object in its path to be crushed. This implies, indeed, that the
spherical object does also have a certain causal power, namely, the power to
crush objects in its path, for crushing something is causing it to be crushed.
However, the power to roll and the power to crush are distinct powers,
since they have distinct manifestation types. What is true, nonetheless, is
that sometimes an object exercises or manifests its power to crush by means of
manifesting its power to roll, when another object with a suitable passive
power lies in its path. I take something similar to be case where the will is
concerned. By willing an agent may exercise or manifest a causal power to
raise his or her arm, but the power to will is nonetheless distinct from the
power to cause one’s arm to rise. Whether by willing the agent succeeds in
raising his or her arm will depend on whether the arm possesses a suitable
passive power to rise, which can be ‘triggered’ by the agent’s act of will.
As for the will’s being an active power—one whose manifestation or
exercise is never caused by something acting upon the agent whose will it
is—this is a claim that I believe to be supported on both phenomenological
and metaphysical grounds. But I shall return to this matter later. At present
I merely wish to observe once more that one cannot object to such a claim
on the empirical grounds that no such powers exist in nature, since modern
atomic physics tells us otherwise, the power of spontaneous radioactive
decay being a case in point.
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 161
Agent causation
What I have just been saying about the will sets me apart from most self-
styled agent causalists, who typically say that human agents possess a special
power to cause their own volitions or intentions.11By this they mean a power
to cause such volitions or intentions not by doing anything—not, that is, by
acting in any manner—but simply in virtue of being a cause of them. Since
agents are, by almost anyone’s account, not events but substances, this means
that such agent causalists are committed to the notion of substance causation:
the causation of something by a substance. Where they differ from me,
however, is in supposing that a human agent can be a cause simpliciter of
something, namely, of a volition or intention. By contrast, I consider that
substances, including human agents, can only cause anything by acting in some
way. I can make no clear sense of the idea of an agent’s being a cause of
somethingwithout doing anything to cause it. Apart from anything else, there
is the familiar difficulty here of explaining why the effect should occur when
it does, if its cause is merely the agent as such.12 For the same agent may cause
many different effects which occur at different times, so that the agent’s time
of existence, which extends over the times of existence of all these different
effects, does not serve to explain why each of them occurs when it does.
The human power of will, as I understand it, is not at all like the sort of
agent-causal power that typical agent causalists subscribe to. As I see it,
when I exercise or manifest my power of will, the exercise consists in my
willing to do something, such as raise my arm. It doesn’t consist in my causing
anything, let alone causing myself to have a volition or intention to raise my
arm. As I say, I regard the will as a non-causal power, although I do want to
say that by exercising this power an agent may, in suitable circumstances,
succeed in causing something, such as the rising of his or her arm. Success or
failure will depend on whether some other thing has a suitable passive
power which can be ‘triggered’ by the act of will in question. If I have an
arm which has a passive power to respond to my volitions to raise it, then
I may succeed in causing it to rise by willing to raise it. This is the only
model of a powers-based conception of voluntary human agency that I can
11 Perhaps the best recent account and defence of agent causalism is O’Connor (2000).12 Perhaps the earliest formulation of this objection is to be found in Broad (1952). I acknowledge that
present-day agent causalists have replies to this objection, but I do not have space to discuss them here.
162 E. J . LOWE
understand. It has the advantage over standard agent causalism that it does
not invoke a special, sui generis type of substance causation which is confined
to the domain of human agency. Rather, it appeals only to types of
causation and powers that are found ubiquitously in the natural world.
Freedom of action
Standard agent causalists typically claim that their account of human agency
can alone provide a satisfactory sense in which voluntary human action can
be free, because by their account agents are originating causes of their own
voluntary actions, making those actions causally determined and yet at the
same time self-determined—that is, determined by the self or agent whose
actions they are, rather than by other agents or events. However, I believe
that my own account of human agency is no less able to explain the sense in
which human agents can freely determine their own voluntary actions. By
my account, they do this precisely by exercising their power of will, whose
manifestations—in the form of volitions or acts of will—help to causally
determine, for example, their bodily movements. To say that agents ‘deter-
mine’ their own volitions by causing them in a sui generis way is, in my view,
no advance on saying that they ‘determine’ them simply by exercising their
power of will, whose manifestations these volitions are. And saying the former
has all the disadvantages mentioned earlier. My account still allows us to say
that agents have an ‘originating’ role in their own actions, inasmuch as
causal chains leading to their voluntary bodily movements begin, by my
account, with an agent’s uncaused exercise of his or her will.
Of course, many philosophers opposed to ‘libertarian’ theories of free will
may be expected to reject both my account and that of standard agent
causalism, on the grounds that these accounts render our voluntary actions
mere ‘chance’ events. The idea is that if nothing determines the agent to
exercise his or her supposed agent-causal power or power of will in a
particular way, then any action determined by the agent in such a fashion
will be indistinguishable from a merely chance happening, such as the fall of
a die. Indeed, there is a notorious argument—the ‘roll-back argument’—
designed to demonstrate precisely this.13 If we hypothetically ‘re-run’ any
13 See van Inwagen (2002).
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 163
supposedly free course of action, as libertarians would characterize it, a
sufficient number of times, then we may expect a certain proportion of
these ‘re-runs’ to replicate what actually happened in the original case, but
also expect a proportion of them to turn out differently—for, after all, if
they all turned out the same way, that would suggest that the agent was not
really free, in the libertarian sense, to determine the outcome. But now it
seems that we may regard the ratio of same to different outcomes in a
sufficiently large number of hypothetical ‘re-runs’ precisely as a measure of
the probability or chance of the original action turning out as it actually did.
And this makes it a purely chance event, like the fall of a die or indeed the
spontaneous decay of a radium atom.
Now, since I have already likened the power of will to the spontaneous
power of a radium atom to undergo radioactive decay, it might seem that
I am particularly vulnerable to this objection. However, although I do
indeed classify the will as a spontaneous power—meaning thereby merely
that it is an active, non-causal power—I by no means want to say that it is in
every other respect just like the power to undergo radioactive decay. I shall
now try to explain in what ways it is crucially different from a power such as
the latter.
The will as a rational two-way power
The first way in which the will differs crucially from the spontaneous power
of radioactive decay is in being a two-way power. As Locke recognized, the
will is a power to will or to refrain from willing any particular course of action
that presents itself to the mind. Presented with the possibility of raising my
arm on a given occasion, I can either will to raise it or alternatively refrain
from so willing by willing not to raise it. In contrast, a radium atom cannot
in any coherent sense refrain from decaying on any given occasion: at most it
can simply fail to decay, because it happens not to manifest its power to
decay on this particular occasion. Its failing to decay is not a manifestation of
a power it has not to decay, since it has no such power. As a free human
agent, however, I do have a power not to will to raise my arm because I have
a power to will not to raise it. And this is the very same power that I can
alternatively exercise by instead willing to raise it. (This, incidentally, is not
to contradict my own doctrine that each power has only one manifestation
164 E. J . LOWE
type, since both in willing to raise my arm and in willing not to raise it my
will manifests itself in the form of acts of will or volitions, albeit volitions with
different intentional contents.) This, then, is why I say that the will is a two-
way power—and in this respect it appears to be utterly unlike any other
power to be found in the natural world.
The second crucial way in which the will differs from other spontaneous
powers is in being a rational power, as indeed Aristotle maintained in Book Ł
of the Metaphysics. By this I mean that it is a power whose exercises are
responsive to reasons, or which is exercised ‘in the light of ’ reasons. A reason for
action I take to be any consideration which speaks in favour of the agent’s
acting in a certain way in certain circumstances.14 When deliberating about
how to act, an agent reflects on such reasons and then exercises his or her
will in a manner that, typically, corresponds to his or her judgement as to
where the weight of reasons for or against any particular course of action
falls. This is not to imply that the agent’s judgement causes him or her to will
or not to will a particular course of action, since that would obviously be
incompatible with saying that the will is a spontaneous power whose
exercises are, accordingly, uncaused. Even so, by this account, we may say
that an agent’s voluntary action, as determined by the exercise of his or her
will, typically has a rational explanation, in the shape of the particular reasons
that the agent judged to merit him or her acting in that particular way.
Obviously, there is a great deal more that needs to be said about such
matters,15 but enough has been said already to set the will radically apart
from a power like that of a radium atom to undergo spontaneous radioactive
decay. When an atom so decays, there is evidently no rational explanation, in
the foregoing sense, for this occurrence. A radium atom can have no
‘reason’ to decay, since it makes no sense to say that there could be
considerations which ‘speak in favour’ of its decaying at any given time,
or indeed ever. Of course, a human agentmight prefer some radium atom to
decay or not decay at some particular time, but the atom itself can have no
such ‘preference’, nor can the inanimate natural world at large. Inanimate
things simply do not belong to ‘the space of reasons’.
These two crucial features of the will serve, I believe, to nullify the ‘roll-
back’ argument. Even if an agent might have willed to act differently in
14 Compare Dancy (2000) and see further Lowe (2008), Chapter 9.15 I say a good deal more myself in Lowe (2008), Chapter 9.
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 165
exactly the same circumstances, this by no means implies that his or her
willing in the particular way that he or she did is just a pure ‘chance’ affair.
For in either case the agent will, typically, have exercised his or her will in
the light of the reasons that were presented to his or her mind and since, in
many cases, it is simply a matter of judgement as to where the weight of
reasons falls—not a matter of incontestable fact—the agent will typically
have acted rationally either way, not just arbitrarily. Moreover, precisely
because this is largely a matter of judgement rather than of fact, there can be
no incontestable way of estimating the ‘probability’ of the agent’s exercising
his or her will in one way rather than another when presented with the same
reasons in the same circumstances. That being so, it is incoherent to suppose
that there is some measurable objective chance of his or her voluntarily acting
in this way or that in these circumstances, in the way that there is a certain
measurable objective chance of a radium atom’s decaying during any given
interval of time.
Agential control
Some may object that the account I have just given leaves us with no real
control over our actions, since it leaves us with no control over our will. The
thought here is that it is not enough for proper agential control over our
actions that we can determine them by exercising our will, unless we can
also determine the exercises of our will. But this then seems to lead us into
a vicious infinite regress of higher-order volitions, with second-order voli-
tions determining first-order ones, third-order ones determining second-
order ones and so on ad infinitum. It might seem, indeed, that standard
agent causalism gets around this difficulty by having the agent as such being a
cause of his or her own volitions and thereby determining them. But I, for
my part, certainly cannot allow that our volitions are ‘determined’, in sense
of being caused, by anything whatever, since I regard the will as a spontan-
eous power.
However, I have already implicitly answered this objection earlier, when
I said that an agent ‘determines’ his or her volitions or acts of will simply by
exercising his or her will. A volition is itself a kind of action—indeed, it is the
most primitive or basic kind of action that any agent can perform. To have a
power of will is to be able to will to do various things, such as raise one’s arm.
166 E. J . LOWE
One exercises this ability simply by willing. Because willing is a basic action,
however, one does not and cannot do it by doing anything else, and so a
fortiori not by willing to will. The idea that one should be able to ‘control’
one’s will, in the sense of doing something which will determine its oper-
ations, simply misrepresents the essential nature of the will, alienating it
from the agent whose will it is. The will is not some mechanism inside
me that I need to have power over. It is my power, for me to exercise as
I see fit, in the light of the reasons for action that present themselves to me.
When I do exercise it, I am demonstrating par excellence my control over
my actions: it is the very source of that control, not some means that I use
whereby to control my actions. For the same reason, it makes no sense to
speak of another agent controlling my will, even if we sometimes mislead-
ingly talk in these terms. Certainly, another agent can offer me strong
inducements to exercise my will in a certain way. But what another agent
cannot do is to determine my will for me. Saying this would be as absurd as
saying that someone else can literally make my mind up for me—indeed, it
really amounts to the same absurdity. Only I can make my own mind up
about some matter, however forceful the inducements that others may offer
me to make it up in a particular way. And it is in this fact that our true
freedom ultimately lies. (Of course, another agent may well be able to render
my will ineffective on a particular occasion, for instance by forcing my arm to
rise when I will not to raise it—but that is entirely different from making me
will to raise it by determining my will for me, which I hold to be impossible
because unintelligible.)
Incidentally, none of what I have just said is to deny that we can school
ourselves to be less impetuous in our decision-making. In that sense, we can
‘rein in’ our will and learn from unwise or hasty decisions taken in the past.
But this is not the same as having to ‘control’ our will by constantly having
to monitor and determine its operations, as though it were some homun-
culus inside us. That is just absurd and no consequence of the theory of the
will that I am advocating.
Causal closure and physicalism
Many contemporary physicalists, who almost always conceive of physical
causality in terms of event causation rather than substance causation,
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 167
maintain that a principle of causal closure reigns in the physical domain.
Although such a principle has been variously formulated by different philo-
sophers, a commonly favoured formulation would be this: at any time at
which a physical event has a cause, it has a sufficient wholly physical cause—
where, by a ‘sufficient cause’, is meant an event or conjunction of events
which causally necessitates the physical effect in question.16 It might appear
that this principle is logically consistent with everything that I have said so
far about voluntary human action. For one thing, nothing that I have said so
far is inconsistent with the view that volitions just are physical events of a
certain kind. For another, the principle as just stated allows that some
physical events may lack causes altogether, just as I say volitions do.
However, further reflection shows that the principle as just stated is in fact
false, quite independently of anything that I want to say about volitions. The
earlier example of spontaneous radioactive decay shows why, recalling that
any such decay is an event that has no cause whatever. Suppose a radium atom,
R, decays at a certain time t1, and that this event—call it e1—is part of a
sufficient physical cause, at t1, of a later physical event, e2, occurring at the
later time t2. To make matters more concrete, suppose that e2 is the
registering of e1 by a Geiger counter, in the form of a ‘bleep’ emitted by
the machine. Now consider an earlier time, t0, preceding that of the decay
e1. At t0 there will surely be some physical events that are causal antecedents
of the bleep e2. After all, amongst the physical events at t1 which jointly
constitute a sufficient physical cause of the bleep e2 there will be some that
have antecedent physical causes, even though the decay e1 does not.
(Clearly, it would be ridiculous to suppose that the decay e1 was, all by
itself, a sufficient cause of the bleep e2.) Hence, by the transitivity of
causation, the bleep e2 has some physical causes which occur earlier than t1,
such as at t0. However, we know that all of e2’s causes at t0 cannot jointly
constitute a sufficient cause of e2, because they cannot jointly causally neces-
sitate e2. This is quite simply because at t0 it is still causally possible for the
decay e1 not to occur and hence for the bleep e2 not to occur. Thus, t0 is a
time at which the bleep e2 has a cause, but not a sufficient wholly physical
cause. Consequently, the principle of causal closure as just stated is false.
16 Compare Kim (1993), 280. I discuss various forms of causal closure principle, including this one, in
Lowe (2008), Chapter 2.
168 E. J . LOWE
The significance of this finding, as far as I am concerned, is the following.
Physicalists often appeal to the causal closure principle, usually in conjunc-
tion with a principle ruling out systematic causal overdetermination, in
order to argue against dualists who claim that mental events and states are
not identical with, nor even ‘realized by’, physical events and states. And,
although I have not so far proclaimed myself to be in favour of dualism in
the present essay, I am in fact firmly of the opinion that it makes no sense to
think of volitions as being either identical with or ‘realized by’ physical
events, such as neural events in the brains of human agents. I do not think
that the human will is a physical power at all, even if human agents need to
have certain physical powers in order to have an efficacious will. Since,
however, we have seen that the principle of causal closure, at least in the
relatively strong form just described, is simply false, it does not present a
threat to the kind of dualism concerning volitions to which I have now
committed myself. I fully acknowledge that this kind of dualism is inconsist-
ent with the causal closure principle as just stated. After all, according to my
account, a volition may be part of a sufficient cause of a subsequent bodily
movement, so that at the time of the volition the bodily movement has a
cause, but typically not a sufficient wholly physical cause, since the volition
itself, by my account, is not a physical event. But since, as we have seen, the
causal closure principle in this form is false in any case, it need be of no
concern to me that my account is inconsistent with it. I strongly suspect that
the same may be said with regard to other variants of the principle which do
not simply beg the question against dualism by ruling it out in the very way
they are formulated. (Incidentally, although I have couched all these
remarks in terms of event causation, this is only because physicalists typically
speak in these terms. The same points could be rephrased in the language of
substance causation that I regard as preferable.)
The rational undeniability of free will
It is pertinent to ask at this point what reason we have for believing
ourselves to possess a rational, spontaneous, two-way power of will of
the sort described in this essay. I have tried to show that our possession of
such a will is metaphysically consistent with current physical science, since
current physical science already admits the existence of spontaneous powers.
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 169
Moreover, physical science has nothing to say, one way or another, about
matters of rationality, since these are normative matters, whereas physical
science is entirely factual or descriptive in its purview and content. I also
consider that my kind of account is phenomenologically plausible. We seem to
possess a rational will that we can freely exercise in deliberative action—
why else would such a doctrine have proved so popular for so many
centuries? Physical determinists will contend, of course, that this is just an
illusion. But we know that current physical science is not on the side of
physical determinism. Can we, though, say anything even more compelling
in favour of the current proposal? I believe we can. I believe we can show
that it would just be irrational to deny that we possess a will of the kind that
I have described. This is not to say that it is metaphysically impossible for us to
lack such a will, only that it is logically impossible for us to believe that we
lack it and still qualify as genuinely rational beings.
The reason for this is as follows. Our voluntary, deliberative actions are
not confined only to our bodily movements, but include also our processes
of thought and reasoning. Now, in any process of reasoning, the mind
draws conclusions from certain premises that it entertains. But for the mind
to draw a conclusion rationally from certain premises, it must draw that
conclusion in virtue of apprehending the support that the premises confer
upon the conclusion. That is to say, the mind—or, more properly speaking,
the personwhose mind it is—must perform the mental action of drawing the
conclusion ‘in the light of ’ the reasons that the premises supply in favour of
that conclusion. The mind must respond to those reasons precisely as reasons
favouring the conclusion. However, if the movement of the mind from
premises to conclusion were purely causally determined, then the mind
could not be said to embrace the conclusion in virtue of apprehending the
support that the premises confer upon the conclusion. For the notion of
‘support’ in this sense is a thoroughly normative one and hence not one that
can be captured by the content of any universal causal law supposedly
governing movements of the mind. Causal laws are merely descriptive of
the domain of non-normative facts. No such law can intelligibly be said to
regulate movements of the mind in accordance with the demands of
rationality, because what those demands are is itself an essentially contestable
matter for rational debate between rational beings. To put it another way: no
cause brings about its effects in recognition of some rational consideration’s
170 E. J . LOWE
favouring the obtaining of those effects.17 (Of course, standard agent caus-
alists will presumably deny this, but can do so only because they invoke a sui
generis notion of agent causation which has no connection with the notion
of a universal causal law.) Only if we are free to draw conclusions because
we apprehend the cogency of the arguments in which they figure can we
truly be said to be rational thinkers—as opposed to mere computing
machines, which may mechanically replicate certain patterns of reasoning
deemed valid by their designers, but without ever actually conducting such
reasoning by and for themselves.
Those who deny that we have a rational free will of the kind I propose are
therefore faced with the following dilemma.18 If what they say is true, then
the movements of their minds that have led them to say it are simply
consequences of certain causal laws governing those movements. Hence,
these movements of their minds may at most replicate valid reasoning but do
not and cannot constitute it. Consequently, their belief in the conclusion—
that we have no rational free will—is not a rationally held belief. On the
other hand, if what they say is false, and unsupported for the reasons that
I have advanced, then again their belief is not a rationally held one, because
they hold it in defiance of the reasons that count against it and only because
they are party to a deterministic dogma which has no foundation in
empirical science. Either way, then, their belief that we lack a rational free
will cannot be a rationally held one. But if it is not rational to believe that p,
then one should not believe that p. Consequently, we should not believe
that we lack a rational free will: we should, on the contrary, believe that we
do possess one.
I have no doubt, of course, that my opponents will want to challenge this
conclusion vigorously and seek every means available to question the
cogency of my argument for it. But, ironically enough, that merely serves
to emphasize its cogency. For in reacting in this fashion my opponents will
precisely be evincing their implicit conviction that they really are rational
thinkers, capable of responding to an argument on its rational merits, not
merely as a result of the playing out of causal laws governing the movements
of their minds.
17 As I have elsewhere expressed it, causation is ‘blind to reason’. See Lowe (2008), 9.18 Compare Malcolm (1968).
SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 171
References
Broad, C. D. (1952). ‘Determinism, Indeterminism, and Libertarianism’. In Ethics
and the History of Philosophy: Selected Essays. London: Routledge & Kegan Paul:
195–217.
Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press.
Kim, J. (1993). Supervenience and Mind: Selected Philosophical Essays. Cambridge:
Cambridge University Press.
Locke, J. (1975 [1690]).AnEssay Concerning HumanUnderstanding, ed. P. H.Nidditch.
Oxford: Clarendon Press.
Lowe, E. J. (1998). The Possibility of Metaphysics: Substance, Identity, and Time.
Oxford: Clarendon Press.
—— (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford
University Press.
—— (2010). ‘On the Individuation of Powers’. In A. Marmodoro (ed.), The
Metaphysics of Powers: Their Grounding and Their Manifestations. London and
New York: Routledge: 8–26.
Malcolm, N. (1968). ‘The Conceivability of Mechanism’. Philosophical Review, 77:
45–72.
Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford: Oxford University
Press.
O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:
Oxford University Press.
Strawson, P. F. (1959). Individuals: An Essay in Descriptive Metaphysics. London:
Methuen.
van Inwagen, P. (2002). ‘Free Will Remains a Mystery’. In R. Kane (ed.), The
Oxford Handbook of Free Will. Oxford: Oxford University Press: 158–79.
172 E. J . LOWE
7
Agent Causation in aNeo-Aristotelian Metaphysics
JONATHAN D. JACOBS AND
TIMOTHY O’CONNOR
Freedom and moral responsibility have one foot in the practical realm
of human affairs and the other in the esoteric realm of fundamental
metaphysics—or so we believe. This has been denied, especially in the
metaphysics-bashing era occupying the first two-thirds or so of the twenti-
eth century, traces of which linger in the present day. But the reasons for
this denial seem to us quite implausible. Certainly, the argument for the
general bankruptcy of metaphysics has been soundly discredited. Arguments
from Strawson and others that our moral practices are too deeply embedded
in human life to rest on anything as tenuous as a metaphysical doctrine far
from the thoughts of ordinary people would seem to prove too much: we
can easily imagine fantastic scenarios far from the thoughts of ordinary
people—involving, say, alien manipulation or massive deception—that, if
true, would clearly undermine claims to freedom and responsibility. For still
other philosophers, the separation of the moral life from (some) metaphys-
ical issues is prescriptive, not descriptive: it is a recommendation that we
revise ordinary moral thought by severing its allegedly problematic links to
metaphysics. (Some philosophers appear to hover undecided between such
a prescriptive project and a Strawsonian descriptive claim.) We suspect that
the prospects of retaining the binding force of ordinary moral thought, were
such a reconceived moral practice widely embraced, are bleak. A transition
to something closer to moral nihilism seems at least as likely. In any case, our
interest here is in descriptive metaphysics, not revisionary.
To say as we do that freedom and moral responsibility have a partly
metaphysical character is not to suggest that they can be had only if some
highly specific version of a particular metaphysical framework is correct.
Instead, we suggest in what follows, it is a broadly neo-Humean metaphysics
that is not hospitable to freedom (for reasons distinctive to the metaphysics),
while a broadly neo-Aristotelian metaphysics is. But we also think (and it is
the main aim of our paper to show) that different versions of the neo-
Aristotelian metaphysics lead to rather different metaphysical accounts of
free and responsible action. Specifically, we will argue that (1) the most
satisfactory account of human freedom within the broadly neo-Aristotelian
metaphysics is agent-causal, but that (2) two different versions of the general
metaphysics will lead to important differences in the agent-causal account of
freedom. Adjust the details of your general metaphysics, and the details of
your account of freedom are transformed in significant ways. Action theory
cannot properly be pursued in isolation from general metaphysics.
1. Freedom and neo-Humeanism
David Lewis popularized a certain form of neo-Humean metaphysics,
according to which causal facts and the laws of nature are reducible to
facts concerning the global spatiotemporal arrangement of fundamental
natural properties (which we allegedly may conceive in non-dispositional
terms). Roughly, the laws are the best system of generalizations over such
natural facts, where bestness is determined by the optimal balance of
simplicity and “strength” (explanatory power). Causation in turn consists
in a restricted kind of counterfactual dependence of one event on another,
where the counterfactuals are grounded in cross-world similarities.1 There
are well-known problems with counterfactual accounts of causation, but we
will not render any pessimistic verdict here.2 Furthermore, the problem that
1 The locus classicus is Lewis’s article “Causation,” reprinted in 1986. (We note that Lewis allows for
temporally remote causation by defining causal chains in terms of stepwise counterfactual dependencies,
but it is unnecessary to fuss about such details here.)
2 For discussion, see the essays in Collins et al. (2004), which includes ‘Causation as Influence’, in
which Lewis proposed a revision of his theory. Hitchcock (2001) and Woodward (2003) have defendedrather different counterfactual accounts that employ the structural equations framework that was given a
major articulation and development by Pearl (2000). For discussion of these developments, see Menzies
(2008).
174 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
we do press against a neo-Humean account of free action is not dependent on
a counterfactual theory of causation. It is a problem for any reductive account
of causation, and we discuss Lewis’s picture simply for the sake of concreteness.
Within the neo-Humean framework, intentional agency is naturally under-
stood in terms of the counterfactual dependence of behavior or behavior-
guiding intentions on appropriate beliefs, desires, or intentions the agent had
immediately before and as the behavior occurs. That human beings act is
(nearly!) uncontroversial. That we act freely can more plausibly be questioned.
We assume here that both metaphysical freedom and moral responsibility are
incompatible with causal determinism. Necessary conditions on free actions
include plausible compatibilist constraints (e.g., the absence of strong internal
or external compulsion) and that they are not determined to occur over some
interval terminating in the initiation of the action.
The inclusion of a non-negligible degree of indeterminism in one’s
account of the proximate genesis of free actions is thought by many to give
rise to problems of explanation and control. But questions of explanation and
control are better posed within particular metaphysical frameworks. It seems
to us that if the neo-Humean framework is accepted, indeterminism need
not present a special problem of control. Causation is just counterfactual
co-variation of a certain kind, and the neo-Humean can readily describe a
form of co-variation of motivational factors and behavior that applies to the
indeterministic case.3 (Indeed, this fact has been insufficiently recognized by
compatibilists who have held that something approximating determinism is
necessary for freedom.) We should require only that the objective chance of
the behavior’s occurring would have been much less in the absence of those
factors. Furthermore, the counterfactual dependence of the chance of
behavior on psychological facts with which the agent identifies is all that it
could be for a person freely to form a choice. (Irreducible agent causation, for
example, makes no sense in this metaphysics, so its omission can hardly be
judged a deficiency.4) Hence, a suitably textured, causally indeterministic
theory of free action gives everything that a neo-Humean could sensibly
3 For an excellent discussion of this issue, see Clarke (1995).4 Obviously, we are further assuming, though less contentiously, that whatever broad metaphysical
account of contingent reality is correct for our world will hold for all worlds involving contingent concrete
particulars. It is not the case that some worlds are neo-Aristotelian while others are neo-Humean. Without
this assumption, the neo-Human account might well be deficient on grounds that there is a kind of direct
control of action had by some possible agents though by no agents in a neo-Humean world.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 175
want for an account of metaphysical freedom (a fact that is insufficiently
recognized by some agent causationists).5
We defer consideration of what is or is not explainable with respect to
undetermined action, treating it in the context of our preferred libertarian
account of freedom. It is true that, given a position held by some neo-
Humeans (and others as well) that there are special explanatory limits in
indeterministic worlds, there will indeed be a serious problem of explan-
ation facing any indeterministic account of agency. But that position is not
compulsory, and we will suggest below that it is implausible.
In our view, the above neo-Humean account of free agency is founded
on a deeply problematic general thesis of causal reductionism. By taking the
fact of A’s being a cause of B to be a reducible, massively extrinsic relation—
grounded in what occurs elsewhere and elsewhen—we empty the funda-
mental idea that causes “produce” or “bring about” their effects of any clear
content.6 Since agency is a causal notion, this problematic consequence
carries over: on a neo-Humean analysis, the sense in which my beliefs and
desires here and now bring about my present action is at best very weak tea.
A fortiori, extrinsic analyses, on which whether or not psychological factors
are causes of behavior is metaphysically determined in large measure by
what happens in the distant reaches of spacetime, provide a bizarre account
of a free action’s being, as we commonly say, “directly controlled by” the
agent, such that it was “up to her” what she would do in the particular
circumstances.7 Our ordinary sense of control with respect to freedom of
5 Indeed, the extrinsic grounding of particular causal facts in the neo-Humean framework might lead
one to doubt the necessity of indeterminism for freedom. See Beebee and Mele (2002). Unfortunately
for the neo-Humean, this same extrinsicality renders it doubtful as an account of causation generally and
of agency in particular, as we argue immediately below.
6 We should acknowledge that “causation” in folk usage probably cannot be neatly lined up with a
fundamental relation in the world, on any likely metaphysical account. The folk, for example, often
speak of causation by absences, as when one says that Susan’s failing to water her neighbor’s plant caused it
to die. On any plausible metaphysical account, there simply are no absences available to stand in a
fundamental relation. In our view, it is most plausible to suppose that ordinary causal talk only roughly
tracks an important fundamental relation in the world, which, to avoid contentious semantic disputes,
we may call “M-causation.” The folk speak truly (often enough) even when speaking of causation by
absences. But such truths are grounded in facts concerning “positive” circumstances that stand in the M-
causal relation (whose nature we sketch in the next section). Conversation with Gunnar Bjornsson has
helped clarify our own thinking here. There is also a nice discussion of this matter in Ted Sider’sWriting
the Book of the World (2011), 15–16 and 75–6.7 See O’Connor (2009) for a development of this point. Gunnar Bjornsson has pointed out in
discussion that it is open to the neo-Humean to modify her account as follows: our concept of natural
law require there to be some minimal score on the balance of simplicity and strength. In neo-Humean
176 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
action manifestly points to something that supervenes on the local circum-
stances in which we act—or, at any rate, circumstances much more local
than those thousands of years in the past or future.
2. A Neo-Aristotelian Metaphysics
and Event-Causal Libertarianism
There is more than one path away from the neo-Humean’s causal reduction-
ism. Here we will consider only the path that we favor: a neo-Aristotelian
metaphysics that assigns a central role to primitive causal powers. On this
view, natural properties are, or of necessity confer, causal powers on their
bearers.8While the neo-Humean’s properties are intrinsically inert, the neo-
Aristotelian’s are intrinsically powerful.9 If the neo-Humean world is
ungoverned, since laws are merely descriptions of contingent regularities,
the neo-Aristotelian world is self-governed, since laws are necessary descrip-
tions of the powerful natures of properties.10Neo-Humean causation is a sort
of counterfactual co-variance, but neo-Aristotelian causation is the exercise
of an irreducible causal power.11 The details of the broadly neo-Aristotelian
metaphysics need not concern us here. What are important are the general
ideas, first, that properties are (or confer) primitive causal powers and,
second, that causation is the exercise of such powers.
In order to understand the nature of indeterministic causation within the
neo-Aristotelian metaphysics, it is helpful to contrast it with another sort of
picture that some contemporary philosophers endorse. On the latter, causal
worlds where the patterns in one cosmic neighborhood or compact world-segment (such as the one we
currently occupy) sharply differ from those in others, we should say that the laws themselves vary from
one world segment to another. In this way, we needn’t say that what contingently occurs in very remote
regions of spacetime are needed to fix what actions I bring about (or whether I ever so much as act at all).
We grant that reducing the extent of extrinsicality serves to improve the view. But since it is the very
nature of the view to give an extrinsic account of causation, this move cannot make the implausibility go
away. There are ever so many neo-Humean worlds where memories and seeming historical traces are
radically misleading beyond a short threshold into the past and where the patterns will abruptly change or
simply cease in the very short future.Whether or not these things are in fact so just seems beside the point
when we ask whether a present bodily motion is something that I freely bring about. (Note that the point
concerns metaphysical determination, not epistemic justification.)
8 See, e.g., Shoemaker (1980, 1998), Heil (2003), Mumford (2004), Bird (2007), and Martin (2008).9 For full discussion, see Jacobs (2011).10 See, e.g., Mumford (2004) and Bird (2007).11 See, e.g., Martin (2008), Mumford (2009), and Bird (2010) for discussion.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 177
indeterminism is thought of as causation of probability. Indeterministic causes,
no less than deterministic ones, are always and uniformly efficacious. They
do not cause the undetermined outcome of an indeterministic process, but
instead cause the outcome’s objective chance of occurring (generally by
raising the probability, though in certain kinds of cases the probability may
actually be lowered). Beyond helping to fix the prior chances of an event,
there is nothing more that a cause does. Where the chance is 1, the cause
suffices for the effect, and so we naturally, if misleadingly, say that it brings
about the effect. But strictly speaking, nothing brings about the effect,
whether the chance is 1 or less than 1. Only the prior chance is brought
about.
This causation of probability view is perhaps assumed (consciously or not)
in objections to the intelligibility of indeterministic agency. If an outcome is
not brought about by anything, it’s hard to see how it can be something that
the agent controls and that we may fully explain in terms of her reasons for
acting. But we should reject the causation of probability interpretation in favor
of a probability of causation alternative precisely because the former makes the
occurrences of events in indeterministic worlds utterly mysterious. There is
no reason within a causal powers metaphysics to suppose that causes must
always produce their characteristic effects, so that in indeterministic scen-
arios we have to resort to the fiction of regular causings of objective
chances. We should suppose instead that indeterministic causes produce
their effects though they need not have done so: they are propensities towards
a plurality of possible effects. They are sufficient for each of them only in the
sense that they are all that is needed, not in the sense that they are a causally
sufficient condition.12 Every indeterministic event is produced, though
none is necessitated. Causation, whether deterministic or indeterministic,
is a singular relation—the very same relation. The prior probability of one
event’s causing another (with limit case of 1) is simply a measure of the
strength of its (single-case) propensity to do so, which helps to fix applicable
laws of nature.
Let us apply this understanding to an indeterministic account of human
free action that, like the neo-Humean account above, is rooted in a causal
theory of action generally. According to it, when an agent freely acts, her
web of motivational states is jointly disposed towards two or more choices,
12 Anscombe (1971) famously develops this point.
178 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
to varying degrees. Whichever choice is made, it will have been caused by
some relevant motivation of the agent, a motivation with which she
identifies.13 The exercise of agent-control consists in the causal efficacy of
one’s motivations, and freedom further requires the openness of the future
to (or consistency of the past and laws with) a plurality of specific outcomes.
Such is a plain vanilla version of event-causal libertarianism.14
While agents, on this account, do not have any less control over what
they do than agents in a corresponding deterministic scenario, they also do
not have more. Indeterminism in the causal link between motivations and
choice opens up a plurality of alternatives unavailable on determinism, but
the agent does not seem to settle which of the options is taken in a sense
robust enough for the agent to be morally responsible. Autonomous
control seems to require more than compatibilist control plus plural alter-
natives. Consider two event-causal libertarian universes, whose histories
have been precisely the same until a time at which two intrinsically
identical agents (including psychological propensities towards the same
possible choices with the same degrees of strength) make diverging
choices. It does not seem correct to say that it was up to the respective
agents, something that they were individually responsible for, that one
chose the path of insult and the other that of gracious forbearance. It’s not
that choices in these worlds would be “freakish,” the “result of pure
chance,” and so not something that the agents in any sense did. It’s merely
that the control that is exercised is of an insufficient variety to ground
robust freedom and responsibility.
But this objection presupposes the intelligibility of a stronger, more
robust variety of control. Unlike on the neo-Humean metaphysics, there
does seem to be space on the neo-Aristotelian account for such an alterna-
tive, as we will now show.
13 The condition that the causing motivation be one with which the agent “identifies” is intended to
handle possible cases where an agent might be subject to a powerful and perhaps momentary “alien”
desire. We needn’t concern ourselves here with different accounts of this notion of “identification.”
14 The foremost recent defender of this theory, Kane (1996), augments the account with further
conditions on the process by which reasons result in choices. Ekstrom (2000) locates the requisite
indeterminism in a special subset of actions—those in which an agent critically evaluates her own
conception of the good and comes thereby to have certain preferences that regulate ordinary actions.
These proposals are interesting and it is worth considering the issues that they raise in their own rights.
But they do not, in our judgment, suffice to answer the fundamental concern with event-causal
libertarianism that we raise immediately below.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 179
3. The Standard Agent-Causal Alternative
Agent causalists maintain that freedom requires a distinct, enhanced kind
of control from the causal efficacy of internal states with which one
identifies. Responsibility-grounding control resides in an indeterministic,
ontologically fundamental causation of a choice or action-guiding inten-
tion by the agent. Taking a feature as a metaphysical primitive is a reliable
way to ensure that one’s overall theory really does allow for the feature,
instead of offering a pale substitute in the manner of various implausible
reductionisms. And the longstanding difficulty of giving a plausible analy-
sis of our pre-theoretical notion of autonomous control suggests that the
gambit of primitive posit is not simply absurd.
We should be careful to distinguish the agent causationist’s position from
that of non-causalists (e.g., Ginet 1990, Goetz 1988, McCann 1998, and
Pink 2004). Both positions agree that autonomous control rests on a
primitive capacity to form intentions (or volitions, according to the theo-
rist’s preference). But the agent causationist insists that this capacity is—and
can only be—causal in nature. The non-causalist, by contrast, ascribes
‘active power’ or ‘the power of choice’ to the agent while insisting that
these terms are to be understood non-causally. However, it is unclear to us
what this means. It seems to us that the term “power” is being misappropri-
ated for rhetorical purposes. Better that these theorists simply say that
nothing causes free choices or volitions but that, notwithstanding, which
choice is made is controlled by the agent, in virtue of the fact that the choice
is his.15 Such a statement is clearer—though clearly false, in our estimation.
In assessing the agent-causal account, we need to consider the role of the
agent’s motivational states in the production of their undetermined choices.
Randolph Clarke (2003) proposes an “integrationist” account on which free
actions are caused both by the agent (qua substance cause) and by certain of
the agent’s motivational states (qua event cause). Clarke proposes that, in the
presence of a “live” agent-causal capacity, it is a law of nature that:
15 It is worth noting that defenders of libertarian accounts of freedom that are ostensibly event causal
sometimes respond to the problem of control by emphasizing not that the choice is caused by the agent’s
reasons but simply that it is “his”—it occurs within the agent. We think that the tendency of event
causalists when pressed to shift between a causal and a non-causal, “ownership” account of control is
revealing.
180 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
(a) whatever action is performed will be caused by the agent,
(b) a particular reason will cause an action only if the agent causes it, and
(c) the agent will cause an action only if some corresponding reason also
causes it.
It seems to us that, absent further explanation, a lawful and symmetrical
causal yoking of this sort is mysterious. Surely one or the other causal factor
will be in the driver’s seat (given, as Clarke says, that they are not each partial
causes). And we want there to be at least one sort of explanatory asymmetry:
it is because the agent had those reasons that he (qua agent cause) caused the
action that he did, not the other way around. Yet Clarke can’t say that the
state of having those reasons indeterministically brought about the agent-causal
event on pain of making indeterministic causation by reasons more funda-
mental than (because prior to) agent causation in the production of an action.
The resulting account would seem to offer no improvement over a simple
event-causal account that dispenses with primitive agent causation.16
O’Connor (2008) suggests a different account of the way that reasons
influence agent-causal actions. He suggests that while agent-causal events
are unproduced by other events, they are probabilistically structured by myriad
factors, especially the agent’s own motivational states. As a result, agents
have a continuously evolving, objective propensity to cause intentions to
act in ways they take to be suited to their ends. More carefully, the idea is
that motivational states act causally on the persisting capacity of an agent
freely to form an intention to act, altering the objective strength of (or
generating) the dispositions the agent has to form specific intentions within
certain intervals. The influence of reasons so conceived is not unlike how
things go according to the causation of probability interpretation of indeter-
ministic causation. However, O’Connor’s account of the influence of
reasons on agent-causal choices is not offered as an account of the nature
of causation itself, and it does not have the absurd consequence that nothing
brings about the specific outcome of an indeterministic process. This
account has the advantage (over Clarke’s integrationist account) of offering
a unified picture of the flow of causal influence, and it does so without
sacrificing the core agent-causal commitment to the exercise of a power that
is not itself in turn produced by previous events.
16 For further discussion, see O’Connor and Churchill (2006).
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 181
Note that on this neo-Aristotelian framework, having reasons (under-
stood as motivational states) is having certain kinds of causal powers.17 So, as
an agent first comes to have reasons for a course of action that was
previously (subjectively) unmotivated, she comes to have new powers of
choice and action. In a case where she comes to have additional reasons for
an already motivated action type, the power so to act is not altered.
However, insofar as the different reasons also motivate different actions of
however fine-grained a type, new powers are thereby acquired. And
distinct reasons must have the potential to motivate somewhat different
action types under at least some possible circumstances, for otherwise they
could not be individuated within the causal powers framework.
On O’Connor’s account of agent-causal power, there is one persistent
agent-causal power, a power to form an intention to act. And an agent with
that power can have differing specific propensities so to act, depending on
what reasons the agent has. The reasons are, in part, powers to act on the
persistent agent-causal power, to alter its strength. It is worth noting,
however, that the causal powers metaphysics does not, by itself, entail
O’Connor’s view of agent-causal power. Indeed, on some versions of the
causal powers metaphysics, specific propensities are essential to causal
powers.18 On this way of thinking about it, when a substance has a power
to do some action, A, the power is a power to do A with some specific
probability in specific circumstances. Were the probabilities different, even
slightly, it would be a different power. Hence, on this view of powers,
reasons are powers to bring about various agent-causal powers, each confer-
ring specific probabilistic tendencies towards specific outcomes in specific
circumstances. Speaking loosely, one has “agent-causal power.” In strict
truth, however, there are a family of related agent-causal powers; all such
powers are similar in being powers of the agent to bring about an intention
with some specific content. This issue, however, is orthogonal to our
central topic, as either view—O’Connor’s view of a persistent agent-causal
17 “Reasons” can refer to normative reasons, or the conditions (generally external to the agent’s
psychological states) that rationally or morally justify a particular course of action for an agent in a
given circumstance, whether or not the course of action is taken or the agent even acknowledges the
existence of the reason. “Reasons” can also refer to motivational reasons, the agent’s own reasons for doing
what he does, wise or foolish as may be. In this latter sense, having a reason is a psychological state or set
of states (such as beliefs, desires, and intentions) that motivates the agent towards and potentially explains
certain courses of action. It is this latter, motivational sense of “reasons” that is in view here.
18 See, e.g., Jacobs (2011).
182 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
power, or the alternative view of a family of related agent-causal powers—
can be accepted in both the above causal powers metaphysics and the
modified causal powers metaphysics to which we turn now.
4. A Modified Causal Powers Metaphysics
On the above causal powers metaphysics, the causes of events that do not
involve agents, and indeed of many events that do involve agents, are
events. The addition of agent causation to such a picture therefore involves
the addition of a new kind of causation. But there is an alternative account
of causation that fits well within a causal powers metaphysics, on which the
causes of all events are substances. All causation, on this view, is substance
causation. E. J. Lowe (2008) argues for this view roughly in the following
way. Causation is the exercise or manifestation of a power. The cause is the
thing that has the power. But only substances have powers. Therefore, only
substances are causes. Events are the having of a power by substances, and
those powers are exercised or manifested by substances.
On Lowe’s version of the substance causation view, substances cause
effects by manifesting a power. But whenever a substance causes an effect, it
does so ultimately by manifesting a non-causal power. The rock caused the
tree’s breaking, by rolling into it. And the rolling of the rock is a manifest-
ation of the rock’s non-causal power to roll. In other words, the rolling of
the rock does not consist in the rock’s causing anything. It consists in, say,
the rock’s changing position.
In many cases, such manifestations of non-causal powers may themselves
be effects. Not so with what Lowe calls spontaneous powers: when a sub-
stance manifests its spontaneous power, it is not caused to do so by anything,
and its manifestation of the spontaneous power does not consist in its causing
anything. Such is Lowe’s account of both radioactive decay and free action. In
both cases, the fundamental source of action is the exercise of a non-causal,
spontaneous power. That is to say, the fundamental action, by which the
agent or atom does cause something, when it does, is an event that is neither a
causing nor caused. In the case of free action, it is a willing. A willing is not a
causing—that is, it does not consist in the substance causing an effect. And it is
not caused by anything. Still, the agent does cause something, say, raising her
hand, by willing to raise her hand, when her willing is effective.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 183
In our judgment, Lowe’s analysis unhelpfully complicates the substance-
causal powers metaphysics. By introducing a fundamental distinction
between the manifesting of a power and that by which a substance manifests
the power, he invites a question concerning the nature of the “by” relation
that has no satisfactory answer that we can see. And it is unclear how this
posit offers any improvement over a simpler analysis on which a substance’s
causing an effect simply is its exercising a causal power. What’s more,
Lowe’s view leads to an even clearer problem in the account of freedom.
For central to the account is a non-causal sort of power, something that we
encountered above in discussing non-causalism. Lowe’s uncaused volitions
appear to be no different, intrinsically, from the non-causalist’s volitions or
choices. As there, so too here: it just seems misleading to call the spontan-
eous occurrence of such events “the exercise of non-causal powers,” given
that the events have no causes and no internal causal structure. We think it
doubtful that control can be understood in non-causal terms.
For these reasons, we find the following analysis to be preferable: a
substance’s having a property is its having a causal power of a specific sort.
A substance’s causing an effect is its manifesting such a power or its
co-manifesting a power with other substances. (Note that on this view,
causation is non-transitive, since causes are substances and effects are
events.) In some possible cases, given the totality of properties had by an
object and its situation, the effect is causally determined to occur. The
conjunction of interacting powers yields a probability of 1 that the substance
or substances will cause that very type of effect. In other possible cases, the
effect is causally undetermined. Here, the exercise of more than one power
is possible and presumably each is probable to some specific degree.
The modified neo-Aristotelian ontology, then, is one on which substances
have powers, and all and only substances are causes of effects.When a substance
causes an effect, it exercises its power to do so—that is, a substance’s causing of
an effect is identical to its exercising its power to bring the effect about.
5. Agent-Causal Libertarianism and the NewOntology
Some of the basic claims of the agent causationist concerning freedom carry
over from the event-causal to the substance-causal powers metaphysics.
First, agents can be literal causes of actions. (Better: actions are agent-causings
184 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
of intentions.) Second, nothing produces an agent’s causing of an intention.
(Since on this metaphysics, in general, nothing produces any causing of an
event by a substance, whether its activity is determined or not.) Finally, the
motivation for incompatibilism about freedom, such as it is, remains.19
However, the substance-causal powers metaphysics forces some changes
to an agent-causal account of freedom. On the event-causal powers meta-
physics, reasons either structure the agent-causal power, in O’Connor’s
sense, or they cause the agent to have the specific agent-causal power she
has. What do reasons do, according to the modified framework now under
discussion? Strictly speaking, nothing. The agent (and the particles that
compose her—see below) do things. Some of those doings are the exercise
of a fundamental agent-causal power to form intentions to act which the
agent has in part because she has the reasons she does. So, reasons are
causally relevant. She wouldn’t have done what she did, and wouldn’t have
been capable of doing what she did, were she not to have those (or other)
reasons so to act. But, strictly speaking, reasons are not causes, since reasons
are not substances, and only substances are causes.
As noted in the previous section, O’Connor (2000, 2008) proposes from
within an event-causal powers framework that what reasons do is to confer
“carried propensities” or “tendencies” on a generic and persisting agent-
causal capacity. This causation of probability has to be taken on board as a
kind of influence that differs from the probability of causation that charac-
terizes non-intentional indeterministic causes. An advantage of the frame-
work now under consideration is thus ideological simplification on this point.
We also have the further ideological simplification resulting from the fact
that agent causation is not a fundamentally distinct kind of causation. What
is distinctive about agent causation among other varieties of substance
causation on this view is merely that the cause is conscious, intentional,
and freely chooses the ends for which it will act.20 As we see it, each of these
19 We note the interesting fact that the substance-causal metaphysics enables one to make good sense
of Markosian’s (1999) claim that agent causation is consistent with compatibilism. (Whereas it is not clear
that it is compatible with the more common event-causal powers metaphysics.) We disagree with
Markosian, however, that compatibilism becomes more attractive once one endorses an agent-causal
theory of free action. Once conceptual space is opened up for determined agent causation, the question
of whether there is a substantive distinction between being produced by me and being freely produced by
me is on the table. As we see it, a suitably formulated version of the Consequence Argument for
incompatibilism is compelling (see O’Connor 2000, Ch. 1).20 Ruth Groff suggested a similar view in conversation.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 185
distinctive sorts of capacity is ontologically fundamental. (They are intercon-
nected in certain ways, with the last, in particular, presupposing the first two.)
“Fundamental” here means that having specific instances of these types of
capacity does not consist in (and is not “constituted” or “realized” by) the
agent’s having some set of other capacities, or in the agent’s parts’ having
certain capacities and standing in certain relations. Given (what we take to be)
the fact that human agents are composed systems, we must think of human
persons as ontologically emergent substances. Among the powers of our
fundamental parts are powers collectively to cause system-level properties/
powers, powers that are sustained as long as the system (the person) retains the
requisite form of organized complexity. As the bearers of fundamental powers,
the person is an ontologically fundamental, albeit composed substance.21
Agents freely act because they (literally) cause their effects with the conscious
aim of attaining certain ends, and their doing so is not settled by features of
their situation (whether external or internal) up to the time of the action.
We have, then, two versions of agent causalism, one embedded with an
event-causal, neo-Aristotelian metaphysics, the other within a substance-
causal, neo-Aristotelian metaphysics. On the first view, agent-causal power
is a power to cause things in a fundamentally distinct sort of way, involving
agent causation rather than the typical event causation. The reasons that an
agent has are powers to structure her agent-causal power, either by causally
influencing the strength of a persisting agent-causal power or by bringing
about a new, slightly different agent-causal power. On the second view,
when an agent causes an event, it does not involve a unique sort of
causation. All causation is substance causation. Rather, the uniqueness
comes from the sort of substance, a conscious substance influenced by
reasons, and the sort of effect, an intention to act for a certain reason.
6. Revisiting the Alleged Problems of Controland Explanation
Some philosophers contend that agent causation, even if coherent, cannot
solve the causal indeterminist’s problems of explanation and control. We
21 See O’Connor and Jacobs (2003) for a detailed account of this picture of ontologically emergent
substances within an event-causal powers metaphysics.
186 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
consider a typical way that each of these problems is presented against agent
causation and argue that they fail.
Agent Causation and the Problem of Explanation
Consider first the problem of explanation.Where an event is undetermined,
there was some objective chance that an alternative type of event might
have occurred—the alternative’s obtaining had a non-zero causal probabil-
ity in the total set of circumstances. This implies that any cause or causes that
one might cite in a putative explanation of the actual event was consistent
with the occurrence of the alternative. It appears to follow that the cause
cannot explain why the actual event obtained rather than the possible
alternative. From this it might seem to follow that the cause cannot, after
all, ‘fully’ explain the actually occurring event itself, since to fully explain
why an event occurred is inter alia to explain why it occurred rather than any
alternative.
Against this argument, we note that not all causal explanations of events
must be contrastive or imply the availability of contrastive explanations, for
every possible contrast. As Peter Lipton (1990) made clear, a request for a
contrastive explanation (“Why P rather than Q?”) presumes that there is an
explanatory relationship between fact (P) and “foil” (not-Q); it presumes
that the occurrence of P and the non-occurrence of Q can be given a
unifying explanation. But this assumption plainly will not hold for every
such pairing even in a deterministic world—as when the occurrence of
P and the absence of Q are completely unrelated matters. In an indetermin-
istic world, contrastive explanation will also fail (plausibly) wherever P and
Q are mutually exclusive; each had a substantial chance of occurring, and
P was not significantly more probable than Q. But it does not follow that there
can be no explanation of P, or that whatever non-contrastive explanation there may
be of P will be somehow deficient—of a lesser variety of explanation than contrastive
explanation. We explain—really explain—an indeterministic outcome P by
citing and describing the causal factor or factors that brought it about,
including the cause’s objective set of objective propensities and the most
salient, proximate causes of its having such propensities.
The point is a familiar one in scientific explanations of indeterministic
phenomena unrelated to free action. If there are a plurality of possible
outcomes of the interaction of a pair of particles, the particular outcome
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 187
that obtains has an explanation in terms of propensities of the two particles
which actually were manifested, bringing about that particular result. Once
one understands the indeterministic nature of those propensities and others
that were not, but might have been, manifested on that occasion, one
realizes that there is nothing further to explain about the situation.
We can even explain why there can be no true contrastive explanation of
the fact that P occurred rather than Q by underscoring the indeterministic
nature of the causal source in question. Philosophers in the grip of the
Principle of Sufficient Reason profess mystification at this scenario, but
don’t give any kind of argument. If we grant that there can be indetermin-
istic causal mechanisms (or agents), then deterministic causes are just the
limit case of a continuum of probabilistic causes, and which sorts of explan-
ation it is appropriate to seek depends on which sort of world we occupy.
The application of this general point to our account of human free action
is as follows: as we come through various causes to have motivations to act
in various ways, the interplay of these motivations and other influences
result in an array of propensities to choose and act of varying strengths.
Suppose that while deliberating on what to do on a Saturday afternoon, I am
disposed with a strength of 0.3 to help a friend repair her deck, owing to my
awareness that she wants to get it done soon, could do so more easily if she is
helped, has helped me in similar ways in the past, together with my desires
to be and to be perceived as helpful to her. And suppose that I am disposed
with a strength of 0.7 to watch a football game instead, for the obvious
reasons. I choose to help my friend. Question: Why did I so choose?
Answer: I so chose because I was motivated by my awareness that my friend
wanted to get the job done, etc. That is to say, those beliefs and desires were
the predominant factors determining my propensity so to choose. And that
is a perfectly good explanation of why I chose as I did, even though there
will not be an explanation of why I chose to help rather than to watch the
football game.22
22 It is worth making explicit here how we would respond to Davidson’s (1963) famous challenge to
theories of (free) action according to which reasons are not causes. (As we noted earlier in the text, on
our view, motivational reasons are not causes, strictly speaking.) Davidson asks how, if reasons are not
causes, we will distinguish cases in which an agent has distinct reasons A and B for choice C but does so
for reason A from similar cases in which she does so for B, and from still other cases in which she does so
for both A and B. But our view supplies a ready answer: the agent acts for all and only those reasons she
had which made a non-negligible contribution to her propensity to choose as she did.
188 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
Agent Causation and the Problem of Control
Recall the argument directed against the causal indeterminist that was based
on a comparison of indeterministic worlds identical up to a certain time at
which intrinsically identical agents make diverging choices. As the causal
indeterminist conceives things, agents don’t cause events—only events do.
Where choices are causally undetermined, some motivational state of the
agent brings about the choice though it was possible that another state have
brought about a different choice. Hence, one set of motivations cause the
choice to insult in world W1, while a distinct set of motivations cause
the choice to refrain in W2. We judged that in this scenario, it is not up to
the respective agents, something that they were individually responsible for,
to make the choices they do. Indeterminism of this sort confers opportunity
but not an enhanced form of control that enables the agent responsibly to
exploit the opportunity.
Some argue that the agent causationist can do no better (Haji 2004; see
also the related, intra-world “rollback” argument in van Inwagen 2000). For
here, too, there is nothing whatsoever about the one agent, right up to the
moment of the choice, that distinguished her from the other, and so nothing
about her that made the difference to what she did. Each had the same
propensity to insult or refrain. Once again, it seems that neither agent
controlled the way their respective cases unfolded in such a way that it
was up to her that she spoke the insult or refrained.
How one replies to this objection depends, in part, on whether one
accepts the event-causal or substance-causal powers metaphysics. On the
former, there is available the very simple reply that the objection fails to take
seriously the concept of agent causation, which is conceived on this meta-
physics as a primitive form of control over undetermined, single-case
outcomes. The agent’s control is exercised not through the efficacy of
prior states of the agent (as on causal theories of action), but in the action
itself. Susan’s causing her intention to publicly insult her opponent is itself
an exercise of control. And since, ex hypothesi, it is quite literally the agent
herself generating the outcome, it is hard to see how the posited form of
control could possibly be improved upon.
But on the substance-causal powers metaphysics, agent causation is not a
special sort of causation, and so cannot be a special, agential form of control
simply by dint of the basic sort of causation manifested. Return, then, to the
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 189
two indeterministic worlds, W1 and W2, identical up to a certain time at
which intrinsically identical agents make diverging choices. And let us
consider three such pairings: a neo-Humean pair of ‘worlds,’ an event-
causal libertarian pair of ‘worlds’, and a purely substance-causal pair of
‘worlds.’ (We put ‘worlds’ in single quotes to mark the fact that it will be
disputed whether any of these descriptions match genuinely possible
worlds.) As on the event-causal powers metaphysics, the proponent of
the substance-causal powers metaphysics can insist on the significance of
the fact that only in the latter pair of worlds are agents literally the causes of
their choices. This fact, construed narrowly, is not sufficient to ground the
claim that agents in these substance-causal worlds but not the others are
responsible for their choices, since it is possible (and indeed common) for
substances to cause their effects without purpose, consciously or otherwise.
But it does mark a relevant and important difference that, when combined
with the fact that the causation is consciously goal-directed, grounds a more
robust variety of control than is possible in either neo-Humean or purely
event-causal neo-Aristotelian worlds. The agent, herself, consciously and
intentionally brought about the effect in both worlds. That, we submit, is a
sufficiently robust form of control to ground freedom and responsibility.
7. Conclusion
We have not here endorsed, let alone argued for, the substance causation
metaphysics. Our aim was only to explore how embracing it would alter the
way that we conceive of metaphysical freedom. There is a more general
moral that we hope will become more widely embraced in action theory
and philosophy of mind: metaphysics matters.23 Debates over reductionist,
epiphenomenalist, or emergentist alternatives concerning both intentional-
ity and consciousness, and over freedom and determinism, turn more on
general metaphysical positions than is commonly acknowledged.24
23 We are not alone in drawing that conclusion. See Beebee and Mele (2002) and the evolving debatebetween Jaegwon Kim and his recent critics over non-reductive physicalist accounts of mental causation.
24 Jonathan Jacobs worked on this project during an NEH summer seminar, “Metaphysics and
Mind,” led by John Heil in 2009. He would like to thank the NEH for its support, and John Heil,
Jonathan Lowe, and the participants in the seminar for their feedback. He would also like to thank the
John Templeton Foundation, for a grant supporting work on this project.
190 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
References
Anscombe, E. (1971). Causality and Determination: An Inaugural Lecture. Cambridge:
Cambridge University Press.
Beebee,H. andA.R.Mele (2002). ‘HumeanCompatibilism’.Mind, 111(442): 201–23.
Bird, A. (2007). Nature’s Metaphysics: Laws and Properties. Oxford: Oxford Univer-
sity Press.
—— (2010). ‘Causation and the Manifestation of Powers’. In A. Marmodoro (ed.),
Powers: Their Grounding and their Manifestations. Abingdon: Routledge: 160–8.
Clarke, R. (1995). ‘Indeterminism and Control’. American Philosophical Quarterly,
32: 125–38.
—— (2003). Libertarian Accounts of Free Will. New York: Oxford University Press.
Collins, J., N. Hall, and L. Paul (eds.) (2004). Causation and Counterfactuals. Cam-
bridge, MA: MIT Press.
Davidson, D. (1963). ‘Actions, Reasons, andCauses’. Journal of Philosophy,60: 685–700.
Ekstrom, L. W. (2000). Free Will: A Philosophical Study. Boulder, CO: Westview
Press.
Ginet, C. (1990). On Action. Cambridge: Cambridge University Press.
Goetz, S. (1988). ‘A Noncausal Theory of Agency’. Philosophy and Phenomenological
Research, 49: 303–16.
Haji, I. (2004). ‘Active Control, Agent-Causation, and Free Action’. Philosophical
Explorations, 7(2): 131–48.
Heil, J. (2003). From an Ontological Point of View. Oxford: Oxford University Press.
Hitchcock, C. (2001). ‘The Intransitivity of Causation Revealed in Equations and
Graphs’. Journal of Philosophy, 98: 273–99.
Jacobs, J. D. (2011). ‘PowerfulQualities, Not Pure Powers’.TheMonist, 94(1),81–102.
Kane, R. (1996). The Significance of Free Will. New York: Oxford University Press.
Lewis, D. (1986). ‘Causation’. In Philosophical Papers: Volume II. Oxford: Oxford
University Press: 159–213.
Lipton, P. 1990. ‘Contrastive Explanation’. In D. Knowles (ed.), Explanation and its
Limits. Cambridge: Cambridge University Press: 247–66.
Lowe, E. J. (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford:
Oxford University Press.
McCann, H. (1998). The Works of Agency. Ithaca: Cornell University Press.
Markosian, N. (1999). ‘A Compatibilist Version of the Theory of Agent Causation’.
Pacific Philosophical Quarterly, 80(3): 257–77.
Martin, C. B. (2008). The Mind in Nature. Oxford: Oxford University Press.
AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 191
Menzies, P. (2008). ‘Counterfactual Theories of Causation’. The Stanford Encyclo-
pedia of Philosophy. Available at http://plato.stanford.edu/entries/causation-
counterfactual/.
Mumford, S. (2004). Laws in Nature. London: Routledge.
—— (2009). ‘Passing Powers Around’. The Monist, 92(1): 94–111.
O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:
Oxford University Press.
—— (2008). ‘Agent-Causal Power’. In T. Handfield (ed.), Dispositions and Causes.
Oxford: Clarendon Press: 189–214.
—— (2009). ‘The Significance of Persons’. In J. Aguilar and A. A. Buckareff (eds.),
Philosophy of Action: 5 Questions. Birker�d, Denmark: Automatic Press/VIP.
O’Connor, T. and J. R. Churchill (2006). ‘Reasons Explanation and Agent
Control: In Search of an Integrated Account’. Philosophical Topics, 32: 241–54.
O’Connor, T. and J. D. Jacobs (2003). ‘Emergent Individuals’. Philosophical Quar-
terly, 53: 540–55.
Pearl, J. (2000). Causality. Cambridge: Cambridge University Press.
Pink, T. (2004). Free Will: A Very Short Introduction. Oxford: Oxford University
Press.
Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and
Cause: Essays Presented to Richard Taylor. Dordrecht: D. Reidel: 109–36.
—— (1998). ‘Causal and Metaphysical Necessity’. Pacific Philosophical Quarterly,
79(1): 59–77.
Sider, T. (2011). Writing the Book of the World. Oxford: Clarendon Press.
van Inwagen, P. (2000). ‘Free Will Remains a Mystery’. Philosophical Perspectives,
14: 1–20.
Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation.
Oxford: Oxford University Press.
192 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR
8
Mental Causation and DoublePrevention
S. C. GIBB
Each of the following four claims seems individually plausible and yet they
appear to form an inconsistent set:
(1) Relevance: Mental events are causally relevant in the physical domain.
(2) Closure: Every physical event contains only other physical events in its
transitive causal closure.
(3) Exclusion: As a general rule, events are not causally overdetermined.
(4) Distinctness: Mental events are not physical events.
The apparent inconsistency of these claims gives rise to the problem of
mental causation. Responses to this problem typically provide reasons to
reject one of the claims. Indeed, the various positions in the mental caus-
ation debate can, to a large extent, be distinguished by the claim that they
reject. Hence, eliminativism and epiphenomenalism both reject Relevance.
Interactive substance dualism and anti-physicalist forms of property dualism
typically deny Closure. Most forms of non-reductive physicalism deny (or
disambiguate) Exclusion. Psychophysical reductionism rejects Distinctness.
An alternative way of responding to the problem of mental causation is to
argue that, contrary to appearances, the four claims are in fact consistent.
Indeed, given that the rejection of any one of the claims encounters serious
difficulties, if there were a way of reconciling them, this would seem to be
the most desirable option. E. J. Lowe has presented one way of doing this
which rests upon the idea that the causal role of mental events in the physical
domain is that of making the fact that a causal tree of neural events converge
upon a particular bodily movement non-coincidental (Lowe 1993, 2000,
2008). In this paper I want to sketch an alternative way of reconciling the
claims. The account that I offer is, in many respects, sympathetic to Lowe’s.
Like Lowe’s account, it resolves the problem of mental causation by
providing mental events with a specific causal role in the physical domain.
Moreover, like Lowe, I deny that this causal role is that of initiating any
single physical event or set of physical events in the chains of neurophysio-
logical causation that terminate in bodily movement. Finally, like Lowe,
I consider the resulting account to be one that is anti-physicalist in nature—
according to it, the mental must be something over and above the physical.1
Despite these similarities, there are two crucial differences in our
accounts. First, the causal role that I shall suggest that mental events play
in the physical domain is very different from the one that Lowe proposes.
According to the account that I shall propose, the causal role of mental
events in the physical domain is to serve as ‘double preventers’. Secondly,
unlike Lowe’s account, which is to a large extent neutral between various
accounts of the causal relation, the account of psychophysical causation that
I wish to advance is not. It is an account of psychophysical causation which
emerges from the acceptance of a powers theory of causation—a theory of
causation that is gaining increasing popularity in recent metaphysics.2
}1 outlines the problem of mental causation. }2 provides an account of thepowers theory of causation and examines how it deals with cases of double
prevention. Assuming this theory of causation, }3 argues that an understand-ing of mental events as double preventers in the physical domain provides
a possible way of reconciling Relevance, Closure, Exclusion, and Distinctness.
Finally, }4 considers why the resulting account of mental events as double
preventers is also attractive from a phenomenological point of view.
1. The Problem of Mental Causation
When we consider our relation to the physical world, little seems more
obvious than the claim that mental events are causally relevant in the
1 For the purpose of this paper, one can take a neutral stance between the various forms of anti-
physicalism. That is, one need not choose between substance dualism and full-blooded property dualism.
2 For recent formulations of the powers theory of causation, see Heil (2003), Martin (2008), Mumford
(2009), and Bird (2010). The idea that powers can provide the basis for a theory of causation is also foundin Harre and Madden (1975), Cartwright (1989), Ellis (2001), and Molnar (2003).
194 S . C. GIBB
physical domain. (Note, mental events are here to be taken as things such as
beliefs and desires, that is, as things that have intentional content.) For
example, my desire to catch someone’s attention and my belief that I can
do this by raising my arm seem to be causally responsible for my arm’s raising.
It is precisely because I had this desire and this belief that my arm raised.
Moreover, in normal circumstances, had I not wanted to raise my arm,
my arm would not have raised. The acceptance of Relevance is, however,
problematic if one also accepts Closure, Exclusion, and Distinctness—the
combination of Closure and Exclusion appears to rule out the causal relevance
of mental events in the physical domain, unless, contrary to Distinctness,
mental events are physical events. Before explaining how I think that the
four claims can be reconciled, these additional premises call for a few explana-
tory comments.
I take events to be the causal relata, where an event is the instantiation of a
property by a substance at a time. (For the purpose of this paper, I am not
assuming any particular metaphysical account of substance.) A mental event
is the instantiation of a mental property by a substance at a time. A physical
event is the instantiation of a physical property by a substance at a time.
The identity of events requires the identity of their properties (as well as their
substances and times). Hence, for mental events to be identical with physical
events, mental properties must be identical with physical properties. As
the identity of events requires the identity of their properties, Distinctness—
the claim that mental events are not identical with physical events—
straightforwardly follows from the argument from multiple realizability.
According to this well-known argument, mental properties are multiply
realized by and, hence, cannot be identical with, physical properties.
While my formulation of Exclusion is quite standard, the formulation of
Closure requires some explanation. The formulation of Closure is Lowe’s
(Lowe 2000, 581). By the ‘transitive causal closure’ of event P, Lowe means
the set of events ‘which includes every event which stands in the ancestral of
the “immediate cause” relation to P’. That is, the set of events which
includes ‘the immediate causes of P, the immediate causes of those causes,
the immediate causes of those causes . . . and so on’. GivenClosure, where P is
a physical event, every event in this set must be physical. Now this is a very
strong version of the causal closure principle—indeed, given its strength,
one may wonder whether the mere combination of Closure with Relevance
and Distinctness is inconsistent. However, I have chosen this formulation of
MENTAL CAUSATION AND DOUBLE PREVENTION 195
Closure precisely because of its strength. In what follows, I shall demonstrate
that, even if one allows the physicalist a causal closure principle as strong as
this, Closure, Exclusion, Relevance, and Distinctness can in fact be reconciled.
As noted earlier, my solution to the problem of mental causation has
emerged from my acceptance of a powers theory of causation. It is therefore
this theory of causation to which I now turn.
2. The Powers Theory of Causationand Double Prevention
I shall not attempt to defend the powers theory of causation here for this
would take me too far from the central aims of this paper. Instead, I shall
simply provide one way of outlining this account of causation and consider
its analysis of cases of double prevention.
According to the powers theory of causation, an account of causation can
be provided in terms of powers or dispositions. (I use these terms inter-
changeably.) The powers theory of causation therefore requires a realist
stance towards dispositions. Given this realist stance, all intrinsic properties
are dispositional, where a property is dispositional if, solely in virtue of
instantiating it, a substance possesses a certain power. Because of its fragility
a porcelain vase is disposed to break when dropped on a hard surface.
Because of its sphericity a ball is disposed to roll when placed on a slope.
The realist stance takes this talk ontologically seriously. The power to break
is built in to some property (or set of properties) of the vase, and it is in
virtue of instantiating this property that the vase is disposed to break when
dropped. Similarly, the power to roll is built into some property of the ball,
and it is because the ball instantiates this property that the ball is disposed to
role when placed on a slope.
The realist stance is in opposition to Lewis’ conditional account of
dispositions which claims that the ascription of fragility to the vase is wholly
analysable in terms of one or more statements about the vase’s behaviour in
a set of counterfactual circumstances (Lewis 1997). It is also in opposition to
those that maintain that dispositional properties reduce to the qualitative
properties of their bearers, and thus, for example, those that consider the
vase to be fragile because of its microstructural properties where these
196 S . C. GIBB
are wholly non-dispositional in nature (Armstrong 1997, ch. 5). However,
to maintain a realist stance towards dispositions is not necessarily to maintain
that properties are exhausted by their dispositionality—properties need not
be clusters of powers. It is also consistent with those accounts which claim
that every property is both dispositional and qualitative.3 Here I shall main-
tain a neutral stance between these different forms of realism.
If one is to take dispositions ontologically seriously then it is essential that
one distinguishes between a disposition and its manifestations. (Although
note that the manifestation of a disposition is itself dispositional in nature—
no intrinsic property, according to this approach, is in pure act.) A specific
disposition is either actual or it is not. To be actual a disposition need not be
manifesting any manifestation. Indeed, it need never manifest any manifest-
ation. Thus a porcelain vase that is never dropped is still fragile despite the
fact that it never manifests this fragility, and so is a porcelain vase perman-
ently encased in bubble wrap. Unmanifesting dispositions are not, therefore,
unactualized possibilia—a description which, as Martin has commented, is
more fitting of unmanifested manifestations (Martin 2008, 12).
A particular manifestation of a disposition nearly always depends on the
presence of other dispositions. (This is precisely one of the reasons why a
disposition might never manifest a particular manifestation, as the other
dispositions upon which this manifestation depends might be permanently
absent.) If a vase is fragile, then the breaking of the vase when it is dropped
on a surface depends not only on the fragility of the vase but also the
hardness of the surface. It follows that the breaking of the vase when it is
dropped on a hard surface is a manifestation not only of the vase’s fragility
but also the surface’s hardness. The vase’s fragility and the surface’s hardness
are, in Martin’s words, ‘reciprocal disposition partners’ and the breaking of
the vase is their mutual manifestation (Martin 2008; see also Heil 2003).
I shall assume, along with Martin and others, that a disposition might—and,
indeed, often will—manifest itself differently with different disposition
partners. Thus, for example, while the surface’s hardness and the vase’s
fragility are reciprocal disposition partners for the mutual manifestation
which is the vase breaking, the surface’s hardness and a rubber ball’s
3 For the first approach towards powers, see Shoemaker (1980). For the second, see Martin (2008),Heil (1998) and Heil (2003).
MENTAL CAUSATION AND DOUBLE PREVENTION 197
bounciness are instead reciprocal disposition partners for the mutual mani-
festation which is the ball bouncing.
These considerations about dispositions and their manifestations lend
themselves to an account of causation in terms of the mutual manifestation
of reciprocal disposition partners.4 The porcelain vase being dropped on a
hard surface causes it to break. In such a case, the vase’s breaking is the
mutual manifestation of the vase’s fragility and the surface’s hardness. My
standing on the broken pieces of the vase causes my foot to bleed.
According to the powers theory of causation, my foot’s bleeding is the
mutual manifestation of the porcelain’s sharpness and my foot’s softness.
The rubber ball being dropped on a hard surface causes it to bounce.
Here, the ball’s bouncing is the mutual manifestation of the ball’s bounci-
ness and the surface’s hardness.
Observe that, given this account, the questionable distinction between
‘the cause’ and its ‘background conditions’ should be dispensed with. Given
the distinction between causes and background conditions, the porcelain
vase’s fragility is the cause of the vase’s breaking when it hits the hard
surface, while the surface’s hardness is a mere background condition that
is necessary for this causal relation to take place. The powers theory of
causation rejects such talk. The breaking of the vase is a mutualmanifestation
of the vase’s fragility and the surface’s hardness. Both play an equal role in
contributing to the effect. Neither should be relegated to the background.
This is merely a sketch of one way of formulating the powers theory of
causation—a sketch which obviously needs to be further developed for it to
provide a satisfactory account of causation. However, it is sufficient for the
purpose of this paper, as its primary interest is with the powers theory’s
analysis of double preventers.
Let us therefore turn to the topic of double prevention. Double preven-
tion occurs when an event that would prevent another event from having a
certain effect is itself prevented from doing so. To give an example, let us say
that one of the attractions at a fair is a simple game in which to win the prize
the player must break a glass bottle by hitting it with a ball. But the game is
rigged. Directly in front of the bottle, and blocking the bottle from any ball
that is thrown at it, is a small barrier that is invisible to the player. The barrier
prevents a ball from ever breaking the bottle. Fred is about to take aim at the
4 See Heil (2003) and Martin (2008).
198 S . C. GIBB
bottle and Sally who is running the fairground attraction suddenly takes pity
on him. She presses a button and this destroys the barrier at the moment that
he releases the ball. (For argument’s sake, let’s say that the destruction of the
barrier is also invisible to the player.) Fred’s ball smashes the bottle and he
wins the prize. Sally’s pressing of the button is a double preventer. It
prevents the barrier from preventing the ball breaking the bottle. This can
be represented diagrammatically as follows:5
In Figure 8.1, a is Fred’s throwing of the ball, b is the breaking of the
bottle, y is the barrier being in front of the bottle and x is Sally’s pressing of
the button. An arrow with a solid line depicts a causal relation. A solid line
ending in a dot depicts an inhibitory connection. A broken line ending in a
dot depicts an inhibitory connection that failed to occur. (And, for later
diagrams, a broken line ending in an arrow shall be used to depict a causal
connection that failed to occur.) A circle around a letter signifies the non-
existence of the relevant event.
How should cases of double prevention be analysed given the powers
theory of causation? Well, to answer this, let us start with the question of
how cases of prevention should be analysed given this theory. Just as a
particular manifestation of a disposition typically depends on the presence of
certain reciprocal disposition partners, it also typically depends on the
absence of others. This is because one disposition may be disposed to
prevent the manifestation of another. It may do so in one of two ways.
A disposition may prevent the manifestation of another disposition because
the manifestation of the first disposition leads to the actual loss of the second
disposition. Alternatively, the disposition might be retained but its mani-
festation blocked by the manifestation of the other disposition. The above
example happens to involve both types of prevention. (Although this is
yx
a b
Fig. 8.1
5 To represent this as simply as possible, I use the kind of neuron diagrams that Lewis (1986) does.
MENTAL CAUSATION AND DOUBLE PREVENTION 199
certainly not true for all cases of double prevention.) The glass bottle is
disposed to break if a ball is thrown at it. However, if a barrier is placed in
front of the bottle, this disposition is not manifested. The solidity of the
barrier prevents the mutual manifestation that is the bottle’s fragileness and
the ball’s momentum and hardness. In this particular case the disposition is
retained—the bottle is still fragile—but as a result of another disposition the
bottle is prevented from manifesting its fragility. The other preventer—the
button being pressed—prevents the barrier’s solidity from standing in a
reciprocal disposition partnership with the ball’s hardness by destroying
the barrier and hence its powers.
In cases of double prevention a disposition that is disposed to prevent the
manifestation of another disposition, is itself prevented from doing so by the
presence of a third disposition. Hence, in the above example, the solidity of
the barrier is disposed to prevent the breaking of the bottle, but is itself
prevented from doing so by the pressing of the button. The consequent
breaking of the bottle is a mutual manifestation of a complex set of dispos-
ition partners which include the ball’s momentum, the bottle’s fragility, and
so on.
What is particularly noteworthy about the powers theory of causation’s
analysis of double prevention is that, according to it, double preventer
events are not causes of the event that they have prevented from being
prevented.6 Hence, taking the above example, according to the powers
theory, the button being pressed is not a cause of the breaking of the bottle.
Why is this?
According to the powers theory, the pressing of the button obviously
cannot be an immediate cause of the breaking of the bottle. If causation is the
manifestation of powers, then a cause and its immediate effect must be
spatially and temporally simultaneous or overlapping.7 And, yet, the pressing
of the button and the breaking of the bottle might be both spatially and
temporally remote from one another. Nor, according to the powers theory,
could the pressing of the button be an indirect cause of the breaking of the
bottle. That is, it cannot be the case that the pressing of the button causes the
breaking of the bottle, by causing the destruction of the barrier. To allow
6 For further defence of this claim, see Mumford and Anjum (2009). Mumford and Anjum do not
formulate the powers theory of causation in exactly the way that I do. However, the central points that
they make regarding the powers theory and double prevention still stand.
7 See, for example, Heil (1998), 187, Martin (2008), 46, and Mumford and Anjum (2009), 287.
200 S . C. GIBB
this causal chain of events, one must accept that the destruction of the
barrier is a cause of the bottle’s breaking—more specifically, one must
accept that the absence of the barrier is a cause of the bottle’s breaking.
Absences cannot be causes according to the powers theory—an absence
cannot bear powers and hence cannot be disposed to act in any way.8
Similar considerations apply in the case where disposition D prevents dis-
position D1 from preventing the manifestation of disposition D2 by merely
blocking D1’s manifestation (as opposed to bringing about D1’s non-exist-
ence). Given the power’s theory, D1’s not manifesting itself—an absence of
a manifestation—cannot be a cause of the manifestation of D2. Hence, there
can be no chain of continuous causation from the event involving D to the
event involving D2.
The powers theory of causation’s conclusion that double preventers are
not causes of the event that they prevent from being prevented differs from
the one drawn by many of the other theories of causation that are dominant
in the philosophy of causation. Hence, to give but one example, according
to the counterfactual theory of causation, Sally’s pressing of the button is a
cause of the breaking of the bottle because there is a chain of counterfactual
dependence linking the two events. If Sally hadn’t pressed the button, then
the barrier wouldn’t have been destroyed and if the barrier hadn’t been
destroyed then the bottle wouldn’t have broken.9
The fact that the powers theory of causation does not count double
prevention as causation is arguably one of its advantages. If double preven-
tion is causation then this has various worrying consequences. Mumford
and Anjum observe that it requires one to accept, not only that there is wide
scale macro-causation at a distance and that absences are causes, but also that
causation is not an intrinsic matter (Mumford and Anjum 2009, 280). That
said, to leave double preventers out of the causal story would be to miss
something crucial out. Isn’t it precisely because Sally pressed the button that
Fred’s ball broke the bottle? An explanation of why Fred’s ball broke the
8 See, for example, Martin (1996), 64.9 My suggestion is not that the powers theory of causation is the only theory of causation that
concludes that double preventers are not causes of the event that they prevent from being prevented.
Hence, for example, the energy transference theory of causation also shares this conclusion because there
is no transfer of energy between the relevant events. (Elsewhere I argue that the energy transference
theory of causation should not provide the basis for a theory of mental causation, as it provides an
account of causation that is physically biased. See Gibb (2010).)
MENTAL CAUSATION AND DOUBLE PREVENTION 201
bottle which focused only on the ball, the bottle, and the throwing of the
ball would be incomplete. Why, according to the powers theory of caus-
ation, is this? For this theory of causation to be plausible, a positive account
of the role of double preventers needs to be provided. Here, I shall offer my
account which hinges upon the distinction between causing an event and
permitting an event to be caused.
Although, according to the powers theory of causation, a double pre-
venter event does not cause the event that it prevents from being prevented,
it permits (or, in other words, allows) the event to be caused. This role is
an objective one, not a merely explanatory one—the fact that a further
event is required to permit the relevant causal relation to take place is quite
independent of our attitudes and interests. Hence, for example, although
Sally’s pressing of the button is not a cause of the breaking of the
bottle, Sally’s pressing of the button permits the breaking of the bottle. It
permits the breaking of the bottle by permitting a causal relation involving
Fred’s ball and the bottle to take place. It permits the causal relation to take
place by causing something else—the destruction of the barrier.
This grounds the counterfactual dependence relation between Sally’s
pressing of the button and the breaking of the bottle—if Sally had not
pressed the button, and hence not permitted Fred’s ball to hit the bottle,
then the bottle would not have broken.10
It also explains why Sally’s pressing of the button should appear in an
explanation of why Fred’s ball broke the bottle. Sally’s pressing of the
button plays an essential role in its occurrence. Indeed the role of an
event that permits a cause to bring about an effect is no less important
than the role of the cause.11
10 Note, therefore, that given my account, causes are not always sufficient for their effects. Take all of
the events that are, in this particular instance, causes of the breaking of the bottle. These are not
collectively sufficient for the breaking of the bottle. In other words, given that all of these causes exist,
it does not follow that the breaking of the bottle occurs. The double preventer event—Sally’s pressing of
the button—must also exist for the breaking of the bottle to occur. But the double preventer event is not
a cause of the effect. As a consequence, given this account, the formulation of the causal closure principle
according to which every physical effect has a sufficient physical cause is misconceived because of its
underlying assumption that every physical effect has a sufficient cause. Note that the acceptance of
Closure does not involve any such assumption.
11 I should emphasize that I am not suggesting that this gives rise to two concepts of causation, which
is a conclusion that Hall (2004) comes to in his discussion of double prevention, distinguishing between
causation as production and causation as counterfactual dependence. According to my account there is
only one concept of causation and it is that described by the powers theory of causation. To permit an
event to be caused is not to cause it.
202 S . C. GIBB
Other theories of causation, such as the counterfactual theory of caus-
ation, do not allow one to recognize the crucial distinction between an
event that causes another event and an event that permits an event to cause
another event—the distinction between causing and permitting a causing.
Both count as causation under the counterfactual theory of causation
because both involve counterfactual dependence. The fact that the counter-
factual theory does not recognize this distinction is arguably an error on its
part, for the events appear to play very different roles.
Assuming a powers theory of causation, let us now return to the mental
causation debate.
3. The Double Prevention Solution to the Problemof Mental Causation
Let us refer to the event that is the firing of neuron N1 as ‘n1’, the event that
is the firing of neuron N2 as ‘n2’ and the event that is Fred’s hand’s moving
as ‘b1’. Consider a possible world in which N1’s firing is disposed to make
N2 fire and this, in turn, is disposed to make certain muscles in Fred’s body
contract and hence make his hand move. For the sake of simplicity let us
assume that no other dispositions are required for these manifestations.
Hence, n1 causes n2 and n2 causes b1. Now let us add that n2’s causing
b1 is prevented by the presence of mental event m2, where m2 is Fred’s
desire to keep his body still. More specifically, that neuron N2 retains its
disposition to make Fred’s hand move, but its manifestation is blocked by
the presence of Fred’s desire. Let us say that Fred has this desire and his hand
is therefore still. (Call this last physical event b2.) In such a world, Closure is
clearly false. This can be represented diagrammatically as follows:
Now consider a second possible world. It is identical with the first in
many respects. Neuron N1 fires in the brain of Fred’s counterpart. n1 causes
b2
n1
m2
n2 b1
Fig. 8.2
MENTAL CAUSATION AND DOUBLE PREVENTION 203
n2. n2 causes b1 but its doing so would be prevented by the presence of m2.
In this possible world Fred has m2, but he also has the conflicting, stronger
desire to move his hand (m1)—let us say due to a bad case of pins and
needles. Fred therefore moves his hand. It might be the case that Fred retains
the desire to keep his body still, but its manifestation is blocked by the
presence of Fred’s overriding desire to move his hand to get rid of the pins
and needles. Alternatively, it might be the case that gaining the desire to move
his hand causes Fred to lose the desire to keep his body still. Either way, the
presence of Fred’s desire to move his hand prevents the manifestation of his
desire to keep his body still. m1 is therefore a double preventer. Fred’s desire
to keep his body still (m2) is disposed to prevent his hand moving (b1), but is
itself prevented from doing so by his desire to move his hand (m1). Taking
the second case, it can be represented diagrammatically as follows:
In this second possible world there is no violation of Closure. By prevent-
ing m2 from preventing n2 from causing b1, m1 is not thereby a cause of
b1—according to the powers theory of causation, double preventers are not
causes of the event that they prevent from being prevented. The cause of b1
is n2 whose cause is in turn n1. Both b1 and n2 therefore contain only other
physical events in their transitive causal closure, as Closure demands.
There is, however, an obvious problem with the proposal as it stands.
Although in the second possible world there is, as a matter of fact, no
violation of Closure, there seems to be great potential for such a violation.
What if n1 causes n2, but m1 isn’t there to prevent m2 from preventing n2
from causing b1 and, hence, we have a similar situation to the one described
in the first possible world? For Closure to be true, in any case where there is
n2 and m2, m1 must be there to prevent m2 from preventing n2 from
causing b1.
This would be the case if the existence of some event in the chain of
neurological events leading to n2 entailed the existence of m1. Hence, for
m1 m2 b2
b1n1 n2
Fig. 8.3
204 S . C. GIBB
example, if the existence of n1 entailed the existence of m1. But what would
explain the entailment relation between n1 and m1? The existence of n1
would obviously entail the existence of m1 if n1 was identical with m1. But
this would be to reject Distinctness which is one of the claims upon which
this discussion is premised. Alternatively, the existence of n1 would entail
the existence of m1 if m1 was distinct from but ‘realized’ by n1. Again,
I think that this suggestion must be rejected. The notion of realization is a
notoriously elusive one, but common to most accounts of realization is the
idea that every causal power of the realized event is also a causal power of
the realizer event (but not vice versa).12 However, this is not the case with
m1 and n1. Given the proposal under consideration, m1 has a causal power
that n1 does not have, namely, the power to affect m2. m1 cannot,
therefore, be realized by n1.
Of course, if these were the only two ways of explaining the entailment
relation between n1 and m1, then one might be forced to accept that any
causal power that m1 has must in fact also be a causal power that n1 has. But
fortunately one is not forced to accept either of these ways of explaining the
proposed entailment relation. A third proposal, which is entirely consistent
with both the distinctness of m1 and n1 and their causal independence, is
that the existence of n1 entails the existence of m1 because whatever causes
n1 also causes m1. This proposal is represented by the following diagram,
where n0 is some further neurological event.
Let me make a number of clarificatory points in defence of this sugges-
tion. First, although whatever event causes n1 must also cause m1, it is not
the case that whatever event causes m1must also cause n1. (Quite clearly for
Closure to be true, it is not necessary that m1’s existence entails n1’s.)
m1 m2 b2
b1n1 n2
n0
Fig. 8.4
12 An important exception to this is Peter Menzies’ account of realization (Menzies 2003). For thepurpose of this paper, I do not have the space to explore this interesting account.
MENTAL CAUSATION AND DOUBLE PREVENTION 205
Second, one might object that even though m2 does not actually have
any physical effect, the claim that m2 would prevent b1 and that it would
bring about b2 if it weren’t for the presence of m1 is still in conflict with the
causal closure of the physical domain. More specifically, there must be
irreducibly psychophysical laws relating m2 and b1 and relating m2 and
b2, even though these laws are never implemented because of the presence
of m1. The claim that there are irreducibly psychophysical laws violates the
causal closure principle, for according to it all of the basic laws must be
purely physical ones.13
To reply, the claim that there are no irreducibly psychophysical laws is
certainly not a consequence of the causal closure principle as it has here been
formulated. Closure states that every physical event contains only other
physical events in its transitive causal closure. This is quite consistent with
the existence of the kind of irreducibly psychophysical laws that I would
have to allow. To rule out such laws, an even stronger closure principle
would be required. However, I fail to see what plausible support could be
mustered—either metaphysical or empirical—for a closure principle of the
required strength. Given the powers theory of causation, laws are nothing
more than generalized claims about causal relations and causal relations are
to be accounted for in terms of powers. Therefore to rule out the kind of
irreducible psychophysical law that I would have to allow, one must
basically provide a reason to rule out the claim that a mental entity could
be disposed to bring about or to prevent a physical entity. But I fail to see
what plausible reason one could give. It is certainly not a consequence of the
powers theory of causation—the theory of causation upon which this
discussion is premised. Nor is it suggested by any fact of physics. Yes, if a
mental event actually did, for example, prevent a physical event, one would
have good grounds for saying that this was, for example, a violation of the
laws of conservation of energy and momentum. But the point is that the
mental never actually does prevent the physical even though it is disposed to
do so. Hence, the conservation laws are never actually violated.
Thirdly, in the causal system that has been presented, one may be uneasy
about the fact that it is not naturally possible for Fred’s desire to keep still to
manifest all that it is disposed for—that is, it is not naturally possible for it to
prevent n2 from causing b1 and for it to cause b2.
13 I’m grateful to E. J. Lowe for raising this possible objection.
206 S . C. GIBB
In response, first note that the claim is not that it is naturally impossible
for Fred’s desire to keep still to ever be satisfied. Rather, the suggestion is
that it is naturally impossible for Fred’s desire to keep still to be satisfied if
neuron n0 fires and hence if Fred has the stronger and conflicting desire to
move his hand. Prior to having the desire to move his hand, Fred’s desire to
keep still might, for example, itself have been playing the role of double
preventer in which case Fred’s desire to keep still would have resulted in
him keeping still.
The point still stands, however, that it is naturally impossible for Fred’s
desire to manifest all that it is disposed for. It is naturally impossible, for
example, for Fred’s desire to keep still to prevent n2 from causing Fred’s
hand’s moving. However, this is arguably not a worry if one takes dispos-
itions seriously, as indeed one must if one is to accept the powers theory of
causation. Dispositions are not made real by their manifestations—they exist
regardless of whether they are manifesting any manifestation. Whether a
disposition can manifest a manifestation depends on both the presence and
absence of other dispositions. Nature rules out certain combinations of
dispositions and allows others. As a consequence, it permits certain mani-
festations of certain dispositions but prevents others. Martin comments that:
‘Salt in a world lacking H2Owould have many of its readinesses unfulfilled’
(Martin 2008, 6). If we accept Martin’s point then, just as in a world lacking
H2O, salt could never manifest all that it is disposed for, in a world in which
n1 is the only cause of n2 and n1 is always accompanied by m1, m2 could
never manifest all that it is disposed for.14
In the proposed system, Closure is therefore true. Indeed it is precisely
because of a mental event, namely m1, that it is true. However, Distinctness,
Exclusion, and Relevance are also true. The proposed system is obviously
compatible with Distinctness—m1 and m2 are not physical events. It is also
clearly compatible with Exclusion—neither m1 nor m2 cause any physical
event and hence do not threaten to causally overdetermine any physical
14 Indeed, Lowe has pointed out to me that the laws of physics provide a good reason to think that
there are unmanifestable dispositions. Quark confinement is the hypothesis that quarks cannot be
isolated singularly. Quarks only exist in combination with other quarks—a total of three in the case of
baryons and two in the case of mesons. Up quarks have a positive charge of two-thirds. Down quarks
have a negative charge of one-third. But, given quark confinement, only charges that are multiples of
these unit charges can be manifested in the form of attraction or repulsion as dictated by Coulomb’s law.
I’m very grateful to Lowe for raising this point.
MENTAL CAUSATION AND DOUBLE PREVENTION 207
event. Finally, regarding Relevance, Fred’s desire to move his hand is no less
important in an account of why Fred’s hand moved, than Sally’s pressing of
the button is in an account of why Fred’s ball broke the bottle. Just as Sally’s
pressing of the button permits the breaking of the bottle, Fred’s desire to
move his hand permits his hand to move. In more detail, m1 (Fred’s desire
to move his hand) permits b1 (the moving of his hand) by permitting the
causal relation between n2 and b1 to take place. It permits the causal relation
between n2 and b1 by preventing m2 from preventing the causal relation
between n2 and b1. If it had not done this, m2 would have prevented n2
from causing b1 and b1 would not have occurred. Fred’s desire therefore
plays an essential role in the occurrence of the hand moving. This role is just
as important as that played by n2. Obviously, it also follows from this
account that it is true that if Fred had not had the desire to move his hand
then his hand would not have moved. That is, b1 counterfactually depends
on m1. This is because if m1 had not occurred, and hence not prevented m2
from preventing n2 causing b1, then b1 would not have occurred.
Given this account, an explanation of Fred’s hand’s moving which simply
focused on the chain of neurological events leading to the hand’s moving
would be incomplete. It is true that one would have specified the complete
set of causes of the hand’s moving. But something crucial would have been
left out of one’s account, because it is in virtue of Fred’s desire to move his
hand that the causal relation between n2 and b1was allowed to take place in
the first place. In this respect, my account of psychophysical causation shares
a very important feature with that of Lowe’s. As Lowe explains, the role that
he gives mental events in the physical domain—that of making the fact that
a causal tree of neural events converge upon a particular bodily movement
non-coincidental—is completely invisible to the scientist who studies only
the physical events and their causal relationships (Lowe 2008, ch. 3). This is
also true of my account of psychophysical causation. An empirical investi-
gation of the chain of neurological events that gives rise to b1will not reveal
any role for non-physical events in the explanation of why b1 occurred.
This is because there are no gaps in this chain of neurological events—every
event in it has an immediate physical cause. Because of this it would be quite
reasonable for the neuroscientist to conclude that a complete account of
why b1 occurred can be given in terms of these neurological events. But, if
my account is correct, this conclusion would be mistaken for it is a non-
physical event (m1) that permits the causal chain of neurological events to
208 S . C. GIBB
give rise to b1 in the first place. This role will be invisible to the neuroscien-
tist. To give an analogy, going back to the example of Sally and Fred, a
passerby focusing solely on the movement of the ball from Fred’s hand to
the bottle will be completely unaware of Sally’s role in the breaking of
bottle, no matter how carefully he looks. Similarly, focusing on the chain of
neurological events leading to b1 will not allow one to detect m1’s role in
bringing about b1, no matter how carefully one studies this chain.
This is not to suggest that my proposal has no empirical consequences. If
empirical evidence established that our decision to perform a bodily move-
ment coincided with (or, indeed, came before) the initiation of a causal
chain of neurological events leading to the bodily movement, this would
prove my account false. In the model that I have presented, neurological
event n0 occurs before Fred has the desire to move his hand and it is
presumably n0 (or, perhaps, some prior neurological event that causes n0)
that initiates the chain of physical events that leads to his hand’s moving. But
empirical evidence establishes no such thing. Indeed, according to Libet’s
experimental findings, it seems to establish quite the opposite (Libet 1985).
Empirical evidence suggests that a ‘readiness potential’ begins in the brain
up to a second or more before a voluntary bodily movement. This raises the
question of whether the conscious experience of deciding to perform the
bodily movement also appears so far in advance. Libet’s famous neurological
experiments seem to demonstrate that it does not. In these experiments
subjects were asked to spontaneously flick their wrist whenever they wanted
to and were asked to note the position of a moving dot when they first
became aware of their decision to move their wrist. Libet found that the
onset of the readiness potential commonly began several hundred millisec-
onds before the subject’s conscious experience of deciding to move. If these
findings are correct then, as Libet observes, the role of mental events cannot
be to initiate chains of causation that give rise to bodily movement. If this
were their role, then one would expect the conscious experience of decid-
ing to move one’s wrist to occur before the onset of the readiness potential
that mediates the bodily movement. Libet’s experiments suggest that it is
instead an unconscious cerebral process that initiates the bodily movement.
In cases of voluntary bodily movement, rather than initiating the chain of
physical events that gives rise to the bodily movement, Libet considers that
the role of a mental event must be to somehow ‘select and control it, either
by permitting or triggering the final motor outcome of the unconsciously
MENTAL CAUSATION AND DOUBLE PREVENTION 209
initiated process or by vetoing the progression to actual motor activation’
(Libet 1985, 529).
Both Libet’s experimental results and the conclusions that he draws from
them clearly fit with the model of psychophysical causation that I have
presented. For the purpose of argument, let us say that n0 is the neurological
event that marks the onset of the readiness potential that initiates the chain of
physical events that give rise to Fred’s hand’s moving. As is consistent with
Libet’s experimental results, Fred’s decision to move his hand occurs after n0,
and hence after the onset of the readiness potential. Moreover, my account
provides a way of explaining how mental events permit the final motor
outcome. A mental event permits the motor outcome by permitting a
neurological event to cause the motor outcome. It permits this causal relation
by preventing a further mental event from preventing this causal relation.
4. Phenomenology and the DoublePrevention Account
I have argued that, within the framework of a powers theory of causation,
understanding mental events as double preventers provides a possible way
of making the combination of Closure, Relevance, Exclusion, and Distinctness
consistent. It is, however, one thing to present a solution to the problem
of mental causation that is plausible from a metaphysical point of view,
and quite another to show that it provides the correct description of
the actual world. And, indeed, from a phenomenological point of view,
one may at first baulk at the idea of mental events playing the role of
double preventers. But in fact the more that one considers it, the more
attractive it becomes. I shall finish this discussion by briefly considering
some of the reasons why.
In a famous passage from The Principles of Psychology, William James
describes the experience of getting out of bed on a cold morning:
We know what it is to get out of bed on a freezing morning in a room without a
fire, and how the very vital principle within us protests against the ordeal [ . . . ] We
think how late we shall be, how the duties of the day will suffer; we say, ‘I must get
up, this is ignominious’, etc.; but still the warm couch feels too delicious [ . . . ]
Now how do we ever get up under such circumstances? If I may generalize from my
210 S . C. GIBB
own experience, we more often than not get up without any struggle or decision at
all. We suddenly find that we have got up [ . . . ] It was our acute consciousness of
both the warmth and the cold during the period of struggle, which paralyzed our
activity then and kept our idea of rising in the condition of wish and not of will. The
moment these inhibitory ideas ceased, the original idea exerted its effects. ( James
1981, 1133)
In James’ example, the desire to get up and start the day and the desire to
stay in the warm bed are in conflict. It is presumably, according to James,
the overriding desire to get up which causes the temporary suspension of
our desire to stay in the warm bed. When we cease to have the desire to stay
in bed, we ‘suddenly find that we have got up’. Here James is not suggesting
that the bodily behaviour is automatic—we get up because of our desire to
get up. But nor is it the case that the desire to get up has re-exerted itself all
over again. Rather, according to James, we find ourselves acting on our
original desire to get out of bed when the conflicting desire to stay in bed
momentarily ceases. As James goes on to comment; ‘the immediate point of
application of the volitional effort lies exclusively within the mental world.
The whole drama is a mental drama’ (James 1981, 1168).
This fits in nicely with an account of mental events as double preventers.
My desire to stay in the warm bed would have prevented me from getting
up. However, my desire to get up causes me to temporarily cease to have
this desire. My desire to get up therefore does not exert itself on some
neurological event but rather on some antagonistic mental event—my
desire to stay in the warm bed. The ‘drama’ is, therefore, as James puts it,
a wholly mental one. Once my desire to stay in the warm bed is suspended
as a result of my overriding desire to get up, this permits the relevant causal
chain of physical events to proceed and I find myself getting up.
Now as James makes clear, not all examples of voluntary action are like
this. According to him, decisions with effort merge into those without it.
These in turn merge into ideo-motor actions—that is those movements that
follow unhesitatingly and immediately from our notion of them in the
mind—which in turn merge into involuntary reflex acts (James 1981,
1178). In the case of ideo-motor acts, James explains that when we perform
the act we are ‘aware of nothing between the conception and the execution.
All sorts of neuro-muscular processes come between, of course, but we
know absolutely nothing of them. We think the act, and it is done [ . . . ]’
MENTAL CAUSATION AND DOUBLE PREVENTION 211
( James 1981, 1131). Thus, for example, to give one of James’ examples,
whilst talking I become aware of some dust on my sleeve, and without
interrupting the conversation, brush the dust from my sleeve (James 1981,
1131). As James explains ‘[i]n all this the determining condition of the
unhesitating and resistless sequence of the act seems to be the absence of any
conflicting notion in the mind. Either there is nothing else at all in the mind, or
what is there does not conflict’ (James 1981, 1132).
If James’ account is correct, then the existence of ideo-motor actions
might be thought to present a problem for an account of mental causation in
terms of double prevention—after all, such cases, according to James,
involve the absence of any conflicting notion in the mind. However,
these kinds of cases do in fact fit with the account that I have offered—
the difference being that in such cases the relevant mental event has no
mental event to prevent to permit the causal relation between the relevant
physical events to take place. If I desire to remove some dust from my sleeve
and I have no conflicting desires, then there is nothing that my desire must
do to permit the causal relation between the relevant physical events to take
place. It is precisely for this reason that the sequence of acts is, as James puts
it, ‘resistless’.
For reasons that I have only had the space to touch upon here, from a
phenomenological point of view, understanding mental events as double
preventers provides the basis for a rich and potentially diverse account of our
actions.
To conclude, I have argued that the powers theory of causation reveals
an important distinction between the role of causing an event and the role
of permitting an event to be caused. Given the powers theory, double
preventers are not causes of the event that they prevent from being
prevented. Rather, they permit the event to be caused, and they do so
by preventing the event that would have prevented it. This role is a
crucial one and allows us to see why double preventers should play a
central role in our explanation of the occurrence of certain events and also
in our counterfactual judgements regarding them. I went on to suggest
that if mental events are double preventers then this provides a way of
reconciling Relevance, Closure, Exclusion, and Distinctness. According to this
claim, a mental event permits a certain bodily movement to take place by
permitting a neurological event to cause the bodily movement. It permits
212 S . C. GIBB
this causal relation by preventing a mental event that would have pre-
vented the causal relation between the neurological event and the bodily
movement. The resulting account is not only consistent with the empir-
ical evidence, but also coincides with some of James’ crucial insights into
the phenomenology of will.15
References
Armstrong, D. M. (1997). A World of States of Affairs. Cambridge: Cambridge
University Press.
Bird, A. (2010). ‘Causation and the Manifestation of Powers’. In A. Marmodoro
(ed.), Powers: Their Grounding and their Manifestations. Abingdon: Routledge:
160–8.
Cartwright, N. (1989). Nature’s Capacities and their Measurement. Oxford: Oxford
University Press.
Ellis, B. (2001). Scientific Essentialism. Cambridge: Cambridge University Press.
Gibb, S. C. (2010). ‘Closure Principles and the Laws of Conservation of Energy and
Momentum’. Dialectica, 64: 363–84.
Hall, N. (2004). ‘Two Concepts of Causation’. In J. Collins, N. Hall, and L. Paul
(eds,), Causation and Counterfactuals. Cambridge, MA: MIT Press: 225–76.
Harre, R. and E. H. Madden (1975). Causal Powers: A Theory of Natural Necessity.
Oxford: Blackwell.
Heil, J. (1998). Philosophy of Mind: A Contemporary Introduction. New York:
Routledge.
——(2003). From an Ontological Point of View. Oxford: Clarendon Press.
James, W. (1981). The Principles of Psychology, ed. F. H. Burkhardt. Cambridge, MA:
Harvard University Press.
Lewis, D. (1986). ‘Postscript C to Causation’. In Philosophical Papers, Vol. 2.
Oxford: Oxford University Press: 184–8.
——(1997). ‘Finkish Dispositions’. Philosophical Quarterly, 47: 143–58.
Libet, B. (1985). ‘Unconscious Cerebral Initiative and the Role of Conscious Will
in Voluntary Action’. Behavioural and Brain Sciences, 8: 529–66.
Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.
15 I’m very grateful to Jonathan Lowe for helpful comments on earlier drafts of this paper. I would also
like to thank James Clarke and audience members at the Durham Philosophy Department’s Research
Seminar. This paper was completed with support from the Arts and Humanities Research Council’s
Research Grant AH/F009615/1 ‘The New Ontology of the Mental Causation Debate’.
MENTAL CAUSATION AND DOUBLE PREVENTION 213
——(2000). ‘Causal Closure Principles and Emergentism’. Philosophy, 75: 571–86.
——(2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford
University Press.
Martin, C. B. (1996). ‘How It Is: Entities, Absences and Voids’. Australasian Journal
of Philosophy, 74: 57–65.
——(2008). The Mind in Nature. Oxford: Oxford University Press.
Menzies, P. (2003). ‘The Causal Efficacy of Mental States’. In S. Walter and
H. D. Heckmann (eds.), Physicalism and Mental Causation. Exeter: Imprint
Academic: 195–224.
Molnar, G. (2003). Powers: A Study in Metaphysics, ed. S. Mumford. Oxford:
Oxford University Press.
Mumford, S. (2009). ‘Passing Powers Around’. The Monist, 92: 94–111.
Mumford, S. and R. L. Anjum (2009). ‘Double Prevention and Powers’. Journal of
Critical Realism, 8: 277–93.
Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and
Cause. Dordrecht: D. Reidel: 109–35.
214 S . C. GIBB
9
The Identity Theory as a Solutionto the Exclusion Problem
DAVID ROBB
This chapter is about a proposed solution to the exclusion problem, one I’ve
defended elsewhere (Robb 1997, 2001; Heil and Robb 2003). Details aside,
it’s just the identity theory: mental properties face no threat of exclusion
from, or preemption by, physical properties, because every mental property
is a physical property. Here I elaborate on this solution and defend it from
some objections. One of my goals is to place it in the context of a more
general ontology of properties, in particular, a trope ontology.
The exclusion problem takes several forms. The version I confront here is
generated by three principles:
Efficacy: Mental properties can produce physical effects.
Closure: Only physical properties can produce physical effects.
Dualism: Mental properties are not physical.
The problem is that each seems true, but the triad is apparently inconsistent.
These principles depart in some ways from Robb (1997). For example,
I here explicitly frame Efficacy (what I used to call Relevance) and Closure in
terms of causal production (compare Kim 2007). Properties here are powers
to produce characteristic physical manifestations in the appropriate circum-
stances (Molnar 2003; Heil 2003, ch. 8). It seems to me that it’s here the
exclusion problem is most clearly a problem of metaphysics, and even, as
I hope to show, ontology. There are, granted, less metaphysically loaded
versions of the problem, along with corresponding solutions. These appeal
not to powers or production, but to, say, explanation (Burge 1993), coun-
terfactual dependence (Yablo 1992; Loewer 2007), or causal intervention
(Woodward 2008). I hope that much of what I have to say below sheds light
on these alternative conceptions of the problem, but I will not explore these
connections here.
I rejectDualism, at least on the most straightforward reading on which the
principle is incompatible with the other two (more on this below). Every
mental property is physical. In particular, every mental property is the
physical property that would otherwise threaten to preempt it with respect
to physical (and especially behavioral) effects. More simply, every mental
property is what some would call its physical “base” or “realizer.” So this is
an identity solution to the problem. This proposal, however, needs consid-
erable refinement and defense, my aims in the rest of this chapter. I’ll
proceed by responding to a number of objections.
O1: Psychophysical property identity was undermined decades ago by the multiple
realizability argument (Fodor 1974; Boyd 1980; Putnam 1980). Indeed, the contem-
porary debate is really just about how to save mental causation in the face of the non-
reductive physicalism established by this argument. Proposing an identity theory at
this point ignores these results and disengages from the contemporary debate.
R1: According to the multiple realizability argument, mental properties
are not physical because they are multiply realizable in the physical. Put
another way, when we ascribe mental properties, we abstract away from
details of physical implementation. On the functionalist version of this
argument, to instantiate a given mental property is to be in some state or
other that plays the defining causal role of that mental property. Since this
realizer or role-filler state can be any of a variety of physical properties, the
mental property cannot be identified with any one of them. Mental prop-
erties are thus second-order properties, at best realized in, but distinct from,
any physical property.
The objector is right that this argument is almost universally endorsed
among those contributing to the mental causation literature. This may be
what leads Yablo (1997, 255), for example, to say that accepting Dualism is
included in “the price of admission” to the mental causation debate. But
while I grant that the argument is sound, its relevance to the exclusion
problem is not so clear.
Start with a few roles properties are thought to play (compare Campbell
1990, 29; Oliver 1996). For example, properties are features; they are, that is,
the truthmakers for (some) predications. The truthmaker for “This apple is
216 DAVID ROBB
round,” for example, is a property, roundness. Properties also sometimes
appear as types. Here properties play the role of a one over many, something
shared by objects of the same type or kind: various round things all have the
same property, roundness. And properties also appear as powers, where this
role can be indicated by the qua locution: an apple causes a certain kind of
impression in soft clay qua round thing—in virtue of its roundness—but not
qua red thing, in virtue of its color.
Now it’s at least not obvious that the same sort of entity answers to
“property” or “roundness” in each case. The three roles—feature, type,
power—may not be filled by the same thing. An ontology of properties
would work this out. But for now, it’s enough to point out that the multiple
realizability argument is aimed (and, I grant, succeeds) at distinguishing
mental and physical types. What the argument shows is that there is no
one-to-one match between mental and physical types. So the psychophys-
ical “property dualism” warranted by this argument is type dualism. It
remains open whether there is any such dualism of features or powers.
But this question is central to the exclusion problem, for it’s clearly proper-
ties as powers that appear in Closure and Efficacy. For all that’s been said so
far, type dualism is compatible with these principles.
Now it could turn out that features, types, and powers coincide, that the
same sort entity fills all three roles. But this is a heavyweight ontological
claim, one that goes far beyond O1’s seemingly innocent appeal to the
multiple realizability argument. And this is where a trope ontology becomes
directly relevant, for according to this ontology, tropes are features and
powers, but something else is a type. Typical among trope theorists is to say
a type is a resemblance class of tropes (Williams 1966; Campbell 1990; Bacon
1995) or, what may amount to the same thing, a collection or plurality of
resembling tropes. The mental type elation, for example, will be a class
of tropes, all of which resemble one another closely enough to count as
tropes of that mental type. But it’s compatible with this that (1) all of the
tropes making up elation are physical, (2) not all of these tropes resemble
one another closely enough to count as a physical type, and (3) some of
them do resemble one another closely enough to count as a physical type:
each cluster of closely resembling tropes will make up a physical type that
“realizes” elation. (For more on this, see R6 below and Robb 1997.) What
results is type dualism but trope monism: a psychophysical identity theory of
features and powers.
THE IDENTITY THEORY AS A SOLUTION 217
Return then to the exclusion problem and the principles that drive it. Let
the “properties” in these principles appear consistently as powers, so that the
principles are incompatible. The proposed solution is to reject Dualism in
favor of an identity theory: mental powers are physical. If O1 protests that
the multiple realizability argument undermines this identity theory, my
reply is that O1 conflates properties as types with properties as powers.
(Or if the objector insists on stipulating that the properties appearing in
Dualism are types, then my reply is that the three principles are consistent
after all, as the other two explicitly concern powers: the principles no longer
generate a problem.) These remarks are not intended to establish a trope
ontology, but just to show that O1’s appeal to multiple realizability is not
ontologically innocent: the nature of properties must be confronted directly
if one is to claim, as so many do, that the multiple realizability argument
blocks an identity solution to the exclusion problem.
O2: Even if the classic multiple realizability argument concerns types, a similar argument
can be deployed concerning tropes, and thus powers. A mental trope is more compos-
itionally plastic than any physical trope. Consider an elation trope E and the complex
physical trope P with which E is allegedly identical. E could survive the change of a
single neuron or particle, while P could not. The tropes are thus distinct, resulting again
in Dualism, but this time it’s explicitly a dualism of tropes, and so powers (compare
Boyd 1980; Pereboom 2002).
R2: Unlike the original multiple realizability argument, this newer ver-
sion threatens an identity solution to the exclusion problem, for it claims to
directly establish a dualism of powers. But the intuitions driving this newer
argument are far more controversial than those behind the original version.
While it’s no doubt true that an elation trope could exist in the absence of P,
by what right does O2 claim that E itself, that very trope, could exist
without P? If this is supposed to be just a brute intuition, it is one that
I do not share.
Still, perhaps I can advance a bit beyond a clash of basic intuitions (though
it’s admittedly hard to avoid question-begging on this front). The judg-
ments of type identity and distinctness in the classical multiple realizability
argument are fundamentally driven by judgments of similarity, and espe-
cially imperfect similarity: physically diverse creatures, while not exactly
resembling, are similar enough to fall under the same mental type. We can
rely here on the fact that higher-level types by their nature abstract away
218 DAVID ROBB
from micro-differences, permitting less than perfect resemblance. But O2
cannot help itself to an analogous claim about tropes, for it’s not nearly as
clear that what’s at the ontological ground level will, like types, permit
higher, more abstract layers. Being ground-level and maximally determinate
would seem to go hand in hand. Conflicting intuitions aside, then, there
seem to be general reasons to be suspicious of the claimed plasticity of
mental tropes.
O3:Multiple realizability is not the only reason to be a property dualist. There are several
other arguments against the identity theory. For example: Mental properties are
irreducibly subjective or private, while no physical property is (compare Jackson 1982).
When ascribing mental properties, we are subject to normative or holistic constraints
that do not bind us when we ascribe physical properties (Malcolm 1970; Davidson
1980). A being—my ‘zombie twin’—could duplicate all of my physical properties yet
lack my mental properties (Chalmers 1996). I might have existed disembodied, with
all of my mental properties but with no physical properties (Yablo 1990).
A psychophysical property identity would have to be necessary, but it appears
contingent, and there is no plausible way to explain away this appearance (Kripke
1980). And there are others. These are directly aimed at distinguishing mental
and physical features, which, at least for the trope theorist, are powers. One
cannot, then, shrug off such arguments as irrelevant to Dualism and the
exclusion problem.
R3: These do indeed threaten an identity theory of mental and physical
features, and thus of powers.1 If any one of these arguments is sound, then it
looks as if Dualism is unavoidable, so that either Efficacy or Closure must be
abandoned. I have no Master Reply to these arguments: they must be
confronted individually, something I won’t attempt here. (I’ve taken on
the zombie in Robb 2008.) However, I will make this more limited reply:
much of the mental causation literature is conducted in the context of non-
reductive physicalism, the pairing of Dualism with the thesis that the mental
is always realized in the physical. The driving argument for Dualism in this
context is the multiple realizability argument, which I hope to have shown is
not nearly as favorable to a dualism of powers as some imagine. But
the arguments in O3 are importantly different, for if they are sound, non-
reductive physicalism is false, and a more robust form of property dualism—
1 I’ll assume in what follows that features and powers coincide: the same sort of “property” (namely, a
trope) fills both rolls.
THE IDENTITY THEORY AS A SOLUTION 219
feature and power dualism—takes its place. In that case the dialectic changes
considerably, and the aim of an identity theorist such as myself is not to
accommodate the dualist argument—as I try to accommodate (classical)
multiple realizability—but to confront it directly.
O4: The identity solution merely relocates the exclusion problem. While psychophysical
trope identity may rescue the causal efficacy of mental powers (tropes), it still leaves
open whether they are causally efficacious qua mental. Put another way: if mental
powers are both mental and physical, why not think they are causally efficacious only
in virtue of being physical? Closure would motivate this problem, and we’re back
with, if not the same exclusion problem, at least very a similar one (Noordhof 1998;
Shoemaker 2003, 434; Macdonald and Macdonald 2006, 552–3).
R4: The general thought behind this objection seems to be that we will
have solved the exclusion problem only if we finally arrive at something—
of whatever ontological category—that’s only mental, that is, mental but
not physical. Anything that’s both mental and physical, the thought goes,
invites exclusion worries all over again. So, for example, Davidson’s (1980,
1993) desired stopping point is at mental events, which, he says, are physical.
But since mental events are both mental and physical, exclusion worries
arise, so that we must show that mental events are causes in virtue of their
mental features (powers). Similarly, the line goes, if our desired stopping
place is with these mental powers, it had better turn out that they’re not also
physical, for then we get the same problem all over again, so that we must
show that mental powers are causally efficacious in virtue of their mental
features (higher-order powers). And so on.
But there’s something suspect in this general line of objection. If psycho-
physical identity at the desired stopping place continues to invite exclusion
worries, why shouldn’t other psychophysical relations at the desired stop-
ping place invite similar worries? Suppose, for example, that mental powers
are in fact only mental—that is, mental but not physical—yet are immanent
in (Yablo 1997, 275) or nothing over and above (Wilson 2005) the physical,
where this may be spelled out in terms of, say, realization (Boyd 1980), the
determinable–determinate relation (Yablo 1992), constitution (Pereboom
2002), metaphysical necessitation (Bennett 2008), de re, a priori determin-
ation ( Jackson 2006), or something else. And suppose that immanence, in
whatever form it takes, does in fact secure the efficacy of mental powers
with respect to physical effects. Is there still a lingering worry that these
220 DAVID ROBB
powers are efficacious, not in virtue of being mental, but merely in virtue of
being immanent in the physical (Lowe 1993, 632–3)? Maybe mental powers
are efficacious, but only because they piggyback on their physical base
powers. This appears to be, if not the original exclusion problem, one
that’s very similar to it.
Now the immanence theorists cited above may suspect that these worries
are somehow ill-conceived. But then why should it be that when we move
to the most intimate form of immanence—namely, identity—these worries
are legitimate? It seems to me that whatever one’s view of immanence, the
way to stop these recurring qua questions is not to ban immanence in the
physical at the desired stopping place. It’s to show that qua questions—and
the exclusion worries that threaten—at that place are somehow illegitimate.
And this is where I think an ontology of properties will again be relevant.
In the work cited earlier, I’ve argued that at the level of powers, qua
questions are illegitimate because powers do not themselves have higher-
order features or powers. More simply: there are no tropes of tropes. Such
higher-order tropes, I’ve argued, are explanatorily idle, threaten to start a
vicious regress, and are just plain odd. Here I’ll make just the oddity point.
Start again with how the qua questions arise for an event identity theorist
such as Davidson: even if mental events cause behavior, they should do so in
virtue of their mental features. For example, the feel of a pain or the content of
a belief must be efficacious. For this purpose, mental features (powers) are
recognized. But now suppose these mental powers are physical, and one
wants to raise qua questions again. Is the motive still that we want the mental
features of a mental power to do some causal work? But this is what strikes
me as odd. A phenomenal trope, for example, doesn’t have a qualitative
feel—at least not in the sense of having a qualitative feature—it is a
qualitative feel. A mental power to cause a bit of behavior doesn’t have a
power to cause such behavior: it is such a power. To raise qua questions at
this level, invoking features of features (powers of powers), looks strange,
and appears to commit a category mistake.
A worry closely related to O4 is that while psychophysical property
identity may secure the causal efficacy of mental powers, it doesn’t secure
‘distinctively’ mental causation, or causal efficacy for mental powers ‘in their
own right’ (e.g., Lowe 1993, 632; Wilson 2009, 150). But again, this seems
to assume that mental powers on the identity theory have a dual nature, so
that (in accordance with some suitably modified version of Closure) only the
THE IDENTITY THEORY AS A SOLUTION 221
physical nature is engaged when mental properties are causally efficacious.
But on the trope ontology, there is no such division within a trope: a mental
power, its mental nature, and its physical nature are all one and the same. In
this sense, a mental trope is both fully mental and fully physical.
O5: The identity solution advanced here combines psychophysical trope identity with type
dualism: one and the same trope can be a trope of two types. But this is impossible,
since tropes are individuated by their constituent types. (For more detailed discussion
of this argument, see Ehring 1996; Whittle 2007.)
R5: The objector here is apparently thinking of tropes on analogy with
Kimian events (Kim 1993), which are individuated, in part, by their con-
stituent types. On such an analogy, if mental and physical types are distinct
(and if a trope can have only one constituent type), no mental trope can be a
physical trope.
But tropes here are not complex entities with types as constituents.
Distinguish two conceptions of a trope (Daly 1997): as a complex entity (a
substance’s instantiating a universal), or as a fundamental entity. On the
latter conception, which I endorse, tropes are of types, but they don’t have
types as constituents. Such a view is required by a trope ontology, on which
tropes are the basic building blocks, the “alphabet of being.” I take it no
building block can have, as a constituent, the derived entities it grounds.
And if tropes don’t have types as constituents, there’s no general barrier to a
trope’s being of more than one type.2
However, while there’s nothing in general to prevent a trope’s being of
distinct types, there would be if the types in question were incompatible.
For example, no trope can be a red trope and a green trope. Returning to
the issue at hand, O5 might insist that mental and physical types are
incompatible, again resulting in psychophysical trope dualism. The dualist
arguments from O3, if sound, would deliver this result: if to be mental is,
say, to be irreducibly subjective, and to be physical is to be irreducibly
objective, then these types would exclude one another. But what about the
anti-reductionist argument in play here, the multiple realizability argument?
This, I claim, shows only that mental types are not physical; it doesn’t show
that they are anti-physical, that is, that they exclude the physical.
2 An analogy to Davidsonian events is suggestive, though it’s just an analogy: I would not want to say
that tropes are events, Davidsonian or otherwise.
222 DAVID ROBB
As a way of fleshing out this point, consider a version of O5 from
Shoemaker (2003, 434, note omitted):
If we think of the instantiation of a property as the conferring on something of
the conditional powers associated with that property, then when properties
confer different sets of conditional powers, the instantiation of one of them is
not identical with the instantiation of the other.
If “property” here means type and “instantiation” means trope (power) this
looks like an argument that mental types are anti-physical in the relevant
sense: no trope can be a trope of both a mental and physical type. In reply,
I grant that no mental type has the same set of causal powers associated with
it as any physical type. (Indeed, one might take this to be precisely the lesson
of multiple realizability.) But this does not show that mental types are anti-
physical. Suppose the causal powers of one type may be, as Shoemaker
himself believes, a proper subset of the causal powers of another (see also
Wilson 1999; Whittle 2007). Types standing in this intimate relation won’t
be incompatible, and this in fact entails that their corresponding tropes will
be identical (assuming the background of a trope ontology). Returning to
the case at hand, if the causal powers associated with mental type M are a
subset of the causal powers associated with one of its physical realizers P,
then trope identity follows: every P-trope will be an M-trope. And if M has
only physical realizers, every M-trope will be some physical trope or other.
O6: Trope identity and type dualism are nevertheless incompatible. We can show this by
assuming trope identity and deriving type identity. Take types to be resemblance
classes of tropes, and assume trope identity, so that a given mental type, say elation, is
a class of physical tropes. Now if this is to be a genuine type, its tropes must exactly
resemble, since (Gibb 2004, 471):
It is only those classes of tropes with the greatest possible degree of unity, that is,
sets of exactly resembling tropes, which can be substituted for universals. This
can be seen by the formal properties of the relation of resemblance. Whilst all
resemblance relations are reflexive and symmetrical, it is only in the case of exact
resemblance that the relation of resemblance is transitive.
From here, the route to type identity is quick. The physical tropes making up elation
must exactly resemble. But then elation is itself a physical type, for it’s no doubt
true—perhaps analytic—that exactly resembling physical tropes belong to the same
physical type. So if elation collects all and only those tropes belonging to a given
physical type, elation is that physical type. Moreover, if all of the tropes making up
THE IDENTITY THEORY AS A SOLUTION 223
elation exactly resemble, we’ve lost any sense in which this mental type is multiply
realizable in physically diverse creatures. The upshot is that the trope identity solution
is forced into type identity: elation, and every mental type, is a physical type.
R6: It seems to me that this objection goes wrong at the beginning:
physical tropes making up mental types needn’t exactly resemble. In fact, for
the trope theorist, this is the lesson of multiple realizability. Color provides a
standard analogy: while any determinate shade of red (shade type, that is)
consists of exactly resembling tropes of that shade, red itself, the determin-
able, is a class of tropes with more relaxed resemblance standards: some red
tropes exactly resemble one another—these exact-resemblance classes form
the maximally determinate shades that realize red—but exact resemblance
isn’t required. For example, the type red contains, say, the scarlet trope of
my blanket and the crimson trope of my chair, and these two inexactly
resemble. Mental types work in much the same way: some of the tropes in
elation exactly resemble. These classes of exactly resembling physical tropes
form all of the physical types that realize elation. But exact resemblance isn’t
required for membership in elation. For example, a trope of human elation
won’t exactly resemble a trope of Martian or dolphin elation.
By appealing to inexact resemblance, I run up against the passage from
Gibb quoted earlier. Following Armstrong (1989, 122–3), Gibb says that it is
“only those classes of tropes with the greatest possible degree of unity, that
is, sets of exactly resembling tropes, which can be substituted for universals.”
But granting this, I reply that mental types—that is, classes of mental
tropes—don’t substitute for universals (compare Whittle 2007, 71). Let
them instead substitute for what Armstrong calls second-class properties,
properties (types) that, while not universals, need to be recognized in any
ontology as part of the manifest image. (Armstrong sometimes suggests
colors as an example.) The idea is not foreign to trope theory. For example,
it seems to have been Williams’ (1966, 81) point when he says that classes of
inexactly resembling tropes “provide a less definite universal.” And Camp-
bell (1981, 484) notes that “The closeness of resemblance between the tropes
in a set can vary. These variations correspond to the different degrees to
which different properties [types] are specific.”3 A second-class property, or
a less definite universal, or a less specific property, is still a natural—not
3 Both this and the Williams passage are quoted in Bacon (1995, 17), though Bacon himself insists on
exact resemblance, as does Macdonald (1998, 334).
224 DAVID ROBB
conventional, not gerrymandered—class of tropes, though it may be, to
borrow a term from Lewis (1983), less than perfectly natural. It’s just that it’s
a class of tropes whose requirements for membership are more relaxed than
those more determinate classes that, I grant, are uniquely qualified to
substitute for universals.
This may be the best place to address the following question: If all of the
tropes in elation (or in any alleged mental type) are physical, what makes this
a mental type? Put another way, if all of the tropes in this class are physical,
what makes them mental as well, and in particular, elation tropes? Here I’m
neutral and say: deploy your favored theory of mentality. Functionalists, for
example, will say that what makes all of these elation tropes is that they all
have the causal profile definitive of elation. Those inclined toward qualia
can say that what makes them elation tropes is their qualitative feel. Those
who think intentionality is the mark of the mental can say that what makes
them elation tropes is their representational content. I take no stand here
on which is these is correct, only that each is compatible with inexact
resemblance between tropes of the same mental type.
O7: But why tropes? If the exclusion problem calls for an identity solution, type identity is
available, for there are versions of the type identity theory, such as Kim’s (1998, 2005),
that accommodate multiple realizability. Since an ontology of properties already requires
types to fill the role of “one over many,” types might as well be pressed into service as
powers as well.
R7: There are some close similarities between the trope identity solution
here and the type identity solution favored by Kim,4 which itself has some
affinities with type identity theories from Lewis (1994) and Armstrong
(1968). A full comparison between the views would take its own paper,
but here I sketch what I take to be the main advantages of a trope identity
solution over a Kim-style type identity solution.
First (Robb and Heil 2008, }6.5), the trope identity solution does not
require, as Kim’s type identity does, that mental types fragment into many
structure-restricted types.On the trope identity solution, there is a single type
elation, and many tropes of that type. What unifies them into a single mental
type is their (inexact) resemblance. I consider this a slim advantage at best,
4 Kim presents this brand of type identity as the best way to save mental causation, but he stops short
of endorsing it: see Kim (2005, 161).
THE IDENTITY THEORY AS A SOLUTION 225
however, for it seems that the plurality of structure-restricted “elations”
recognized by Kim might also be united into a single type by inexact
similarity, even if Kim himself (1998, 111) is not inclined to recognize such
unity.
Second, Kim concedes that his own reductivist account of mental caus-
ation is second-best:
The best, or the most satisfying, outcome would have been a vindication of
mental causation along the lines of nonreductive physicalism; that would have
allowed us to retain mentality as something that is causally efficacious and yet
autonomous vis-a-vis the physical domain.
But the best outcome, as we saw, is not to be had. The next best outcome, in
fact our only hope at this point if mental causation is to be saved, is physical
reductionism. Physical reduction would save causal efficacy for mentality, at the
cost of its autonomy. Reductionism allows only one domain, the physical
domain, but the mental may find a home in that domain. (Kim 2005, 159)
The suggestion here is that something valuable is lost in the move to a
reductionist (or identity) solution to the exclusion problem. Kim says that
what’s lost is autonomy from the physical, but if autonomy here is just taken
to be Dualism, then it’s not clear why losing that should be mourned.
Perhaps Kim has in mind here the distinctively mental contribution of mental
powers, the efficacy of the mental as such mentioned above in R4. But if
that’s what is lost in Kim’s account, then this is a reason to favor the trope
identity solution, for there is no such loss on a trope ontology: for reasons
given in R4, nothing distinctively mental is missing in either mental tropes
or their causal efficacy.
A third reason to favor trope identity solution over its type identity
sibling is a familiar ontological worry: it’s not clear that types are the right
sorts of the things to be powers, as they would have to be if a Kim-style
solution is to be tenable. Types play the “one over many” role for properties,
and so whatever exactly types are, they seem to be “spread out” over their
various instances. (This is the case, I take it, even for Kim’s structure-
restricted types.) Powers, however, are local, in re, here-and-now. The
point is clearest if types are taken to be resemblance classes, for a class, in
addition to being abstract, is not local in a way a power must be (for a bit
more on this, see Heil and Robb 2003, 175–6). Types may be useful in
explanation, but this epistemic role should not be confused with the meta-
physical role of powers, a role for which types seem to be ill-suited.
226 DAVID ROBB
O8: Still, why tropes? What’s essential to this identity solution is that mental powers are
local, in re, and above all, physical. But for all that tells us, mental powers are not
tropes but Armstrong-style universals (Armstrong 1989, 1997). Like tropes, Arm-
strong-style universals are both features and powers. The main difference is that
Armstrong-style universals are, well, universals: they are wholly present in multiple
instances, and moreover (in some cases) are types. But this difference doesn’t seem to
make a difference as far as an identity solution to the exclusion problem goes. Tropes
or Armstrong-style universals would do just as well (Whittle 2007, 70; Heil 2008;
Maurin 2008).
R8: There are indeed multiple and substantive similarities between a
trope ontology and Armstrong’s ontology of universals. For example, in
their respective ontologies, both tropes and Armstrong-style universals are:
(1) features; (2) powers;5 (3) spatiotemporal, at least when we restrict our
attention to physical properties; (4) instantiated in re; (5) dependent beings,
depending, in particular, on the objects that instantiate them; (6) maximally
determinate: neither ontology permits the sort of “layering” one finds among
types; (7) responsible for sameness of type; that is, they are the grounds or
truthmakers when two objects are of the same type or kind.
The similarities between the two ontologies are so striking that it’s
tempting to deny there is a deep difference here. Indeed, at one point
Armstrong (1989, 139) wonders, following a suggestion from H. H. Price,
whether the difference between these apparent rivals is merely one of
alternative languages: what the trope theorist describes as exactly resembling
features, the Armstrong describes as one and the same. Both describe, in
their own ways, the same underlying facts. There are a number of way this
suggestion could play out. Let S1 and S2 be two exactly resembling features.
The suggestion could be the relatively innocuous claim that S1 and S2,
while strictly distinct, are nevertheless identical in a looser sense. This seems
to be the lesson of Williams’ (1986) later view on universals, what Campbell
(1990) calls ‘painless realism’. Alternatively, one might appeal to the doc-
trine of relative identity (Geach 1980), so that S1 and S2 are distinct tropes,
but the same universal. Or the suggestion could be a conventionalism of
sorts, so that there’s no objective fact of the matter of whether S1 is S2: the
5 There is a difference here: tropes in themselves bestow causal powers, while Armstrong-style
universals do so only given the laws of nature. But both are powers in the sense that they are those
entities that are causally efficacious, whether or not such efficacy depends on laws of nature.
THE IDENTITY THEORY AS A SOLUTION 227
apparently distinct ontologies result from the neutral facts plus either of two
conceptual or linguistic overlays chosen for, say, pragmatic reasons (Carnap
1956; Sidelle 2002). I can’t follow through on these lines here, or maybe
anywhere. But supposing it turns out that one of them is correct, what
becomes of O8? If the worry is that Armstrong’s ontology would do just as
well as the trope ontology in response to the exclusion problem, my reply
would be that it would do just as well, but only because it’s the same
ontology, differently described.6
This is as far as I want to pursue this option. Suppose there is a genuine
ontological difference, as there appears to be, between Armstrong-uni-
versalist and trope-theoretic versions of the identity solution. Is there any
reason to favor one over the other? I doubt an advantage will be found
within the confines of the exclusion problem. The ontologies appear
isomorphic in any respect relevant to this problem. If either has advantage,
it would have to come from more general considerations.
One minor edge the trope ontology has is that it can give a uniform
account of types. As before, call the ground-level “properties,” whether
tropes or universals, features. For the universalist, all types are resemblance
classes of features except for maximally determinate types. Every maximally
determinate type is itself a feature, a single universal. But for the trope theorist,
there’s no such abrupt change at the most basic level: all types, even the most
determinate, are classes of resembling features. What’s distinctive of the basic
level of types is that the features of those types exactly resemble. But there’s no
categorial difference at this level for the trope theorist. Here ontological
continuity strikes me as a virtue, though this is admittedly a small advantage.7
Another potential advantage for the trope ontology may be found in the
metaphysics of causality. It’s typical for a trope theorist to insist that tropes
are better suited than universals—even Armstrong-style universals—to play
6 I take causal efficacy to be, like causation itself, extensional: a power to produce a certain effect is so
no matter how described. Matters would be different if, say, explanation rather than causal efficacy were
in play. Explanation is intensional, so that even if the two ontologies are equivalent, there may be
explanatory reasons to favor the conceptual apparatus of one over that of the other.
7 The advantage for the trope theorist would be greater if the universalist were forced to say that the
higher-level types are not genuine (because not universals). In that case, the trope theorist but not the
universalist could admit elation, e.g., as a legitimate type: compare Ehring’s (2003, 384) response to O8.But it seems to me that a universalist can be a realist about higher-level types (whether or not Armstrong
himself is). They just will not count as genuine universals, but something else, such as the ‘second-class
properties’ mentioned in R6.
228 DAVID ROBB
the role of powers: Campbell (1981, 480), for example, makes the point, as
does Honderich (1992, 246–7). Representative is this passage from Camp-
bell (1990, 23):
It is not the stove, the whole stove, that burns you; not even the whole stove
here now. For its solidity, iron structure, enamel surface and smoothness have
nothing to do with it. It is the temperature that does the damage. Moreover, it is
not any temperature, or temperature in general, but this particular case of
temperature, among the myriads in the world, and even among the many the
stove has during its life. Yesterday’s stove temperature is quite innocuous. It is
today’s that burnt you.
Again, accommodation of the ontology of causes into the trope scheme is so
smooth because what is required is an element that combines particularity with a
very restricted qualitative character, since causes are always features (almost
always a small selection from the host of features present) and every particular
cause is a particular feature or constellation of features.
Campbell’s point here is complicated by his taking tropes to be the causal
relata, while I’m neutral on this matter. But the argument applies equally as
well if a causally efficacious property of a cause—a power—must, like causes
themselves, combine elements of particularity and qualitativity.8 Moving
beyond Campbell’s argument, one might also motivate the trope ontology
in the context of a more general metaphysics of causation. Here I have in
mind, for example, Ehring’s (1997) recent defense of the transference
theory, one which makes essential use of tropes. In any case, however one
evaluates Ehring’s trope-based metaphysics of causation, its relevance to O8
helps to confirm a primary theme of this chapter: the ontology of properties
is not optional when confronting the exclusion problem.9
References
Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge &
Kegan Paul.
—— (1989). Universals: An Opinionated Introduction. Boulder, CO: Westview Press.
8 Combination, however, is only a metaphor here: for reasons given earlier, a trope is its particularity
and qualitativity.
9 A version of this chapter was read in September 2009 at Davidson College and at “The New
Ontology of the Mental Causation Debate,” a conference held at Durham University. I thank the
audiences on those occasions for helpful comments, questions, and objections.
THE IDENTITY THEORY AS A SOLUTION 229
Armstrong, D. M. (1997). A World of States of Affairs. Cambridge: Cambridge
University Press.
Bacon, J. (1995). Universals and Property Instances: The Alphabet of Being. Oxford:
Blackwell.
Bennett, K. (2008). ‘Exclusion Again’. In J. Hohwy and J. Kallestrup (eds.), Being
Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford
University Press: 280–306.
Boyd, R. (1980). ‘Materialism with Reductionism: What Physicalism Does Not
Entail’. In N. Block (ed.),Readings in Philosophy of Psychology, Vol. 1. Cambridge,
MA: Harvard University Press: 67–106.
Burge, T. (1993). ‘Mind–Body Causation and Explanatory Practice’. In J. Heil and
A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 97–120.
Campbell, K. (1981). ‘The Metaphysic of Abstract Particulars’. Midwest Studies in
Philosophy, 6: 477–88.
—— (1990). Abstract Particulars. Oxford: Basil Blackwell.
Carnap, R. (1956). ‘Empiricism, Semantics, and Ontology’. In Meaning and Neces-
sity, 2nd edn. Chicago: University of Chicago Press: 205–21.
Chalmers, D. A. (1996). The Conscious Mind: In Search of a Fundamental Theory.
Oxford: Oxford University Press.
Daly, C. (1997). ‘Tropes’. In D. H. Mellor and A. Oliver (eds.), Properties. Oxford:
Oxford University Press: 140–59.
Davidson, D. (1980). ‘Mental Events’. In Essays on Actions and Events. Oxford:
Clarendon Press: 207–28.
—— (1993). ‘Thinking Causes’. In J. Heil and A. Mele (eds.), Mental Causation.
Oxford: Clarendon Press: 3–18.
Ehring, D. (1996). ‘Mental Causation, Determinables and Property Instances’.
Nous, 30: 461–80.
—— (1997). Causation and Persistence: A Theory of Causation. Oxford: Oxford
University Press.
—— (2003). ‘Part–Whole Physicalism andMental Causation’. Synthese, 136: 359–88.
Fodor, J. A. (1974). ‘Special Sciences (Or: the Disunity of Science as a Working
Hypothesis)’. Synthese, 28: 97–115.
Geach, P. T. (1980). Reference and Generality, 3rd edn. Ithaca: Cornell University
Press.
Gibb, S. C. (2004). ‘The Problem of Mental Causation and the Nature of Proper-
ties’. Australasian Journal of Philosophy, 82: 464–76.
Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.
—— (2008). ‘Modes and Mind’. In S. Gozzano and F. Orilia (eds.), Tropes,
Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 13–30.
230 DAVID ROBB
Heil, J. and D. Robb (2003). ‘Mental Properties’. American Philosophical Quarterly,
40: 175–96.
Honderich, T. (1992). ‘Causation: One Thing Just Happens After Another’. In
L. E. Hahn (ed.), The Philosophy of A. J. Ayer. La Salle: Open Court: 243–70.
Jackson, F. (1982). ‘Epiphenomenal Qualia’. Philosophical Quarterly, 32: 127–36.
—— (2006). ‘On Ensuring that Physicalism is Not a Dual Attribute Theory in
Sheep’s Clothing’. Philosophical Studies, 131: 227–49.
Kim, J. (1993). ‘Events as Property Exemplifications’. In Supervenience and Mind.
Cambridge: Cambridge University Press: 33–52.
—— (1998). Mind in a Physical World. Cambridge, MA: MIT Press.
—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-
sity Press.
—— (2007). ‘Causation and Mental Causation’. In B. P. McLaughlin and J. Cohen
(eds.), Contemporary Debates in Philosophy of Mind. Oxford: Blackwell: 227–42.
Kripke, S. A. (1980). Naming and Necessity. Cambridge, MA: Harvard University
Press.
Lewis, D. (1983). ‘New Work for a Theory of Universals’. Australasian Journal of
Philosophy, 61: 343–77.
—— (1994). ‘Lewis, David: Reduction of Mind’. In S. Guttenplan (ed.),
A Companion to the Philosophy of Mind. Oxford: Blackwell: 412–30.
Loewer, B. (2007). ‘Mental Causation, or Something Near Enough’. In
B. P. McLaughlin and J. Cohen (eds.), Contemporary Debates in Philosophy of
Mind. Oxford: Blackwell: 243–64.
Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.
Macdonald, C. (1998). ‘Tropes and Other Things’. In S. Laurence and
C. Macdonald (eds.), Contemporary Readings in the Foundations of Metaphysics.
Oxford: Blackwell: 329–50.
—— and G. Macdonald (2006). ‘The Metaphysics of Mental Causation’. Journal of
Philosophy, 103: 539–76.
Malcolm, N. (1970). ‘Scientific Materialism and the Identity Theory’. In
C. V. Borst (ed.), The Mind/Brain Identity Theory. London: Macmillan: 171–80.
Maurin, A. (2008). ‘Does Ontology Matter?’ In S. Gozzano and F. Orilia (eds.),
Tropes, Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 31–56.
Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford: Oxford University
Press.
Noordhof, P. (1998). ‘Do Tropes Resolve the Problem of Mental Causation?’
Philosophical Quarterly, 48: 221–6.
Oliver, A. (1996). ‘The Metaphysics of Properties’. Mind, 417: 1–80.
THE IDENTITY THEORY AS A SOLUTION 231
Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,
99: 499–531.
Putnam, H. (1980). ‘The Nature of Mental States’. In N. Block (ed.), Readings in
Philosophy of Psychology, Vol. 1. Cambridge, MA: Harvard University Press:
223–31.
Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:
178–94.
—— (2001). ‘Reply to Noordhof on Mental Causation’. Philosophical Quarterly, 51:
90–4.
—— (2008). ‘Zombies from Below’. In S. Gozzano and F. Orilia (eds.), Tropes,
Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 155–74.
—— and J. Heil (2008). ‘Mental Causation’. In E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy, Fall 2008 edition, available at http://plato.stanford.
edu/entries/mental–causation/.
Shoemaker, S. (2003). ‘Realization and Mental Causation’. In Identity, Cause, and
Mind, Expanded Edition. Oxford: Clarendon Press: 427–51.
Sidelle, A. (2002). ‘Is There a True Metaphysics of Material Objects?’ Philosophical
Issues, 12: 118–45.
Whittle, A. (2007). ‘The Co-instantiation Thesis’. Australasian Journal of Philosophy,
85: 61–79.
Williams, D. C. (1966). ‘The Elements of Being’. In Principles of Empirical Realism.
Springfield, IL: Charles Thomas: 74–109.
—— (1986). ‘Universals and Existents’. Australasian Journal of Philosophy, 64: 1–14.
Wilson, J. (1999). ‘How Superduper Does a Physicalist Supervenience Need to
Be?’ Philosophical Quarterly, 49: 33–52.
—— (2005). ‘Supervenience–based Formulations of Physicalism’.Nous, 39: 426–59.
—— (2009). ‘Determination, Realization and Mental Causation’. Philosophical
Studies, 145: 149–69.
Woodward, J. (2008). ‘Mental Causation and Neural Mechanisms’. In J. Hohwy
and J. Kallestrup (eds.), Being Reduced: New Essays on Reduction, Explanation, and
Causation. Oxford: Oxford University Press: 218–62.
Yablo, S. (1990). ‘The Real Distinction Between Mind and Body’. Canadian
Journal of Philosophy, Supplement 16: 149–201.
—— (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.
—— (1997). ‘Wide Causation’. Philosophical Perspectives, 11: 251–81.
232 DAVID ROBB
10
Continuant Causation,Fundamentality, and Freedom
PETER SIMONS
the main element in the notion of a thing or continuant is the permanency of
functioning that can be discerned in a series of characterised manifestations,
presented in the course of time, as they may be observed in a temporally
continuous, or discrete, series of acts. Thus the notion of a continuant is
constructed in terms of temporal connection and causal determination
W. E. Johnson1
Continuants and Occurrents
Objects in time divide into continuants and occurrents.2 Occurrents are
events, processes, and states. They have temporal as well as spatial extension.
The temporal extension allows us to speak of their temporal parts, that some
temporal parts are earlier or later than others, that longer temporal parts
include shorter, and so on. Continuants by contrast have no temporal
extension, only spatial. A continuant exists at different times in its own
right, and not because it has a temporal part that exists at that time. So when
a continuant changes it has first one property and then another property
incompatible with the first. By contrast when one temporal part of an
occurrent is one way and a later temporal part is another, the property in
question inheres in different items, the temporal parts in question. For
1 Johnson (1924), 98. 2 Johnson (1924), 78–101.
example, the noise of a car engine gets louder and higher in pitch when the
accelerator is depressed. So while occurrents may vary over time in how
they are, because their different temporal parts vary, only continuants
genuinely change. The modes of existing in time of continuants and
occurrents are called ‘enduring’ and ‘perduring’ respectively by David
Lewis.3 Though I shall treat continuants and occurrents as equally accept-
able denizens of the world it will be important to my account of continuant
causation that occurrents have metaphysical priority over continuants.
I shall say that a continuant participates in an event or other occurrent
when this event is part of the life of the continuant in question.
Further Assumptions
Additional assumptions I shall be making but not arguing for in this paper are
the following. Firstly, that determinism is false, so that at any given time
certain aspects of the future with respect to that time have not at that time
been fixed or determined. I call this the open future assumption. Of course the
future is not totally open: some things are bound to happen, at least some
things of a certain kind. For example at some time in the future, we know not
when, current estimates say in about 5 billion years, the sun will become a red
giant and burn up the earth. The precise details of that process have yet to be
determined, but some process of that sort appears inevitable. The second
assumption is that we have some freedom of action: we are able ourselves to
determine in some respects how things turn out. We are familiar with this
phenomenologically from everyday experience, and while some philoso-
phers have denied or doubted it, I do not and I shall not here be arguing for
freedom, which is another enterprise. Rather the task is to see how freedom
may be accommodated in the account of causation. Much of the point of
stressing causation by continuants among the adherents of continuant caus-
ation, in particular agent causation, lies in its role in securing the acceptance of
a coherent account of the phenomenon of freedom. So a denier of freedom is
not the addressee of this paper, although the account of causation we offer is
compatible with the denial of freedom: I am simply trying to establish that it
does not entail lack of freedom. Finally, and more speculatively, I shall pursue
3 Lewis (1986), 202. The terminology is due to Mark Johnston.
234 PETER SIMONS
mydiscussionwithin the framework of amonistic naturalism, the view that all
entities in the universe are within spatiotemporal nature. This rules out
abstract entities, an eternal deity, and any other entity, such as an immortal
soul, which is not by its nature spatiotemporal. One may call this position
‘physicalism’ provided one does not mean by this that everything about the
world can be expressed in the impoverished idiom of physics. Since I regard
all entities in nature to be generated from an ontological basis that includes
fundamental physical kinds and a suite of basic ontological factors4 I prefer the
term ‘generative naturalism’. Finally, and this is consonant with generative
naturalism, that mental states and events are physical: mind/body monism.
Occurrent Priority and Continuant Participation
One could treat participation of a continuant in an occurrent as a meta-
physically basic and unanalysable relationship. This would be the natural
position to adopt if continuants and occurrents are equally primitive and
mutually irreducible categories of entity. The problem is that participation
looks too intimate to be anything like an external relation such as being next
to, but nor is it so intimate as to be identity or part–whole. While not
denying that continuants exist, I claim they are metaphysically posterior to
occurrents, and their nature as persistent enduring individuals has a straight-
forward explanation.5
A continuant like a ball, a rifle, a spacecraft, a human being, exists for a
period of time, changing in various ways throughout its life. For it to
continue as one and the same individual across time certain things about it
have to remain in existence. A ball cannot be cut up or flattened, a rifle
cannot be sawn into pieces, a spacecraft cannot break apart, and a human
being cannot long stop respiring and having electrical activity in the brain.
When we look more closely at what keeps the continuant in existence,
however, we come across processes, more saliently in the case of organisms,
but present also in the more passive cases. The continuous exchange of
photons among the atoms of a body hold it together against forces which
might tend to pull it apart. The constituent atoms are held together by the
exchange of gluons mediating the strong nuclear force. And so on. At the
4 Simons (2009). 5 Simons (2000).
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 235
most physically basic level of apparently partless objects like quarks and
leptons, it might be thought that these simply passively endure, rather than
persist via a constituting process. But quantum theory tells us that at the
most basic level there are no particles, only field values at spatiotemporal
locations. Whether the ultimate ontological assay of this situation will
replace the continuant/occurrent duality with something more basic is
not yet clear, so the claim of occurrent priority partout is presently a
speculative conjecture, and is here acknowledged as such.
The importance of processes to the existence and persistence of continu-
ants also comes out via a different route. When we say truly that a certain
object exists, we may raise the question as to what makes this statement true.
A truth-maker is an entity which makes a statement true because it exists.
Therefore a singular existential statement such as ‘Socrates exists’ would
appear to be made true by the entity its subject term designates, in this
case, Socrates. But there is a snag. The truth-maker for a statement is
supposed to necessitate its truth: that is to say, necessarily, the statement is
true if the truth-maker exists. And indeed Socrates quite happily fulfils this
role with respect to his own existential statement. But qua continuant,
Socrates exists if and only if he exists at some time. So consider any statement
of the form ‘Socrates exists at t’, where t is an arbitrary time. At those times at
which Socrates does not exist, e.g., 600 bce or 600 ce, the statement in
question is false. It is only true for those times at which he exists. So take one,
e.g., 400 bce, and take one particular point in that year. Call this time ‘T’.
What makes it true that Socrates exists at T (or, taking account of the fact that
it is in our past, that Socrates existed at T)? It cannot be Socrates, since it is not
essential to Socrates that he existed at T. He might have died earlier. So
Socrates himself does not necessitate that he exists at T. What then might do
that job? It must be something that, unlike Socrates, has to exist at T if it exists
at all, and has to be intimately related to Socrates. This can only be something
that it has its locations in time essentially, and somehow by existing makes
Socrates exist. This is a process vital to Socrates: its occurring at T is sufficient
for Socrates to exist at T. In fact it is something like a temporal section of
those processes which are necessary and sufficient for Socrates to be alive at
T.Whatever such a process is, it is by its occurring that Socrates exists at T, so
it is a truth-maker for that temporally relativized existence statement.
Continuants exist because they exist at some time. They exist at some
time by virtue of something whose existence at that time is necessary to it
236 PETER SIMONS
and which hand on the sustaining role from one time to succeeding times.
These it seems must be vital processes, whatever these are in detail. The
continuant exists from one time to another by virtue of the continuation of
processes vital to its existence. The relation between phases of such pro-
cesses may be called genidentity. It is in general an equivalence relation, and
applying abstraction to it we arrive at the continuant itself as the invariant
which is identical among the genidentical processes. Although there is more
to genidentity than causal continuity, causal continuity from one phase of a
process to later phases is an important aspect of genidentity. Other things
going on within and to a continuant that are not parts of its vital processes are
nevertheless parts of the total process of what is going on with the continuant
at the time. These include in the case of a person their thoughts and actions,
and within the total process there are smaller strands with causal connections
going back and forth, including the strands leading from thoughts and voli-
tions to actions. These being all processes or other occurrents and their parts,
this is all O-causation. That is what participation consists in: processes that are
part of the total process for the continuant are alongside and rendered possible
by the vital processes in virtue of which the continuant exists.
Again it is an open question whether this analysis applies at the basic
physical level of fields and field values, but for ourselves and the things that
concern us with regard to our lives and actions, the analysis appears tenable.
Continuant and Occurrent Causation
Arguably the most important phenomenon of the natural world is causation,
which I take to be the determining of what happens by something else in
time. This only makes real sense under the open future assumption. The
assumption of naturalism rules out the possibility of any supernatural inter-
vention or determination. Within naturalism there are two potential ways
in which this determination might happen: determination by continuants
and determination by occurrents. It is the thesis of this paper that determin-
ation by occurrents, what I call occurrent causation, is the metaphysically prior
form, and that continuant causation is a derivative and secondary form
of causation, which cannot operate except via occurrent causation. The
ultimate reason for this is the ontological priority of occurrents over con-
tinuants, to which we come now before returning to causation. To have
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 237
short expressions available I shall say continuant causation is C-causation and
occurrent causation is O-causation.
Let’s look at some common or garden examples of occurrent and con-
tinuant causation. First, a bat hitting a ball and the ball’s momentum
changing as a result. This can happen whether the ball was previously
moving or not, and it can happen in a dramatic way, the ball drastically
changing its direction and velocity of motion, or in a mere glancing
deflection. The point is that the event of the bat and the ball colliding
causes the ball’s velocity to change. The collision is an event, brief but not
instantaneous, since the change in momentum is caused by internal distor-
tion and restitution of the ball, which takes a short interval. The subsequent
flight of the ball usually lasts much longer than the collision. This is a typical
case of occurrent-to-occurrent causation. There are of course participant
continuants, namely the bat and the ball, but the changes to them are events
or occurrents.
A more complex case concerns a gas explosion. We imagine a gas pipe or
joint begins to leak gas into the surroundings. This may go on for hours.
The concentration of gas in the surrounding air builds up. Then for some
reason there is an electric spark, and this ignites the gas–air mixture, which
combusts and expands rapidly: an explosion. The explosion, a short-lived
event but with clear phases, is caused by the spark, a short-lived event, but
the spark is only a triggering cause because of the leak. So both the long-
lived leaking and the brief spark are parts of the total cause. Both are
occurrents, and so is the resulting explosion. This is typical for occurrent
causation: the contributory factors are several and it is their conjunction or
coming together which determines that the effect occurs.
We now consider continuant causation. Imagine a snooker player who is
aiming to pot the black. He lines up his cue with the cue ball and the black,
hits the cue ball towards the black and as a result the black ball rolls across the
surface of the table and into the pocket. We say ‘He potted the black’, and
‘pot’ is here an agentive achievement verb. Of course what he did in the first
instance was to move the cue in a certain way so that it hit the cue ball at a
certain angle and in a certain position and with a certain velocity, and the
rest was beyond his control, but it was his skill in hitting the cue ball which
ensured the black was sunk, since the ensuing process of rolling, colliding,
and rolling took place without the player’s further interference, and barring
an earthquake or other unexpected disruptive intervention, the outcome
238 PETER SIMONS
was determined from the moment the cue hit the cue ball, indeed a little
before then. We can quite congruously and correctly say the snooker player
caused the black to go into the pocket.
Take another example with a longer timeline. We imagine a river flowing
swiftly downhill. In the middle of the river is a rock which protrudes from the
water surface and disturbs the water flow. Behind and near the rock, the flow
is turbulent, whereas upstream of the rock and well to the sides and down-
stream, the flow is laminar. We quite happily and correctly say the rock causes
the turbulence. Were it not to be there, and nothing else in its place, the river
would flow smoothly over that place. The way in which the obstacle causes
turbulent flow, while familiar, is, like all turbulence, poorly understood. The
rock and the turbulence may coexist happily for a long period.
Priority of Occurrent Causation
If the fundamental metaphysical entities in space and time are occurrents, then
the fundamental form of causation must be O-causation. C-causation exists
only because continuants do, continuants exist only becauseO-causation does,
so O-causation is prior to C-causation. The solution to the problem of
freedom therefore cannot rest with C-causation, since that is a half-way
house to what is metaphysically fundamental. The problem of freedom is for
that reason not as easily resolved as the adherents of continuant causation
would like. That does not mean it is insoluble, but it does help to explain why
the question of determinismof events in and around the person is so difficult to
reconcile with the idea of freedom, since we cannot simply hive off the free
cause to something outside the usual (O-) causal order.
The derivation from C-causation from O-causation explains also some-
thing that the proponents of agent causation cannot. When an agent acts
freely, that action takes place at a certain time. The time at which the agent
causes the action is therefore either at or just before the action. What makes
it true that the agent causes the action at this time? Since it is contingent to
or inessential of the agent that he or she does so, at any time, let alone this
particular time, then the agent cannot be the truth-maker, for the existence
of the agent is not sufficient for the agent to cause this action at this time, the
agent being a continuant, and the action inessential to the agent. If there is a
truth-maker, it would have to be something whose existence necessitated
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 239
that the action occurs. That points back again to occurrents, but unlike in
the more speculative theory that the continuant is itself constituted by
occurrents, here we have unproblematic occurrents to call upon, namely
those immediately preceding and O-causing the action. The causation can
be timed to the moment the action becomes physically unavoidable, and
that will be because the last O-cause comes about, and its time of occurring
is essential to it.
Of course the agent causation theorist may wish to insist that the state-
ment that the free action occurred at such and such a time has no truth-
maker. But if we can offer a theory in which it does, and the truth-maker is
unproblematic, then that is another reason to accept O-causation as more
basic than C-causation.
But frequently when we use the verb ‘cause’ we are talking about
C-causation, a relation between a continuant and some occurrent it causes.
How is this talk to be squared with the metaphysical prority of O-causation?
Agentive and Causative Verbs
Languages contain many verbs whose meanings include the notion of
causation. Usually this comes from the idea of the agent, typically but not
invariably the subject of the sentence, doing something, describable either
by the verb itself or by the verb in conjunction with noun objects and
complements. We give illustrative examples from English but other lan-
guages provide similar examples:
Sean cleaned the car
Sometimes the subject is not a continuant such as a human being but an
event or other occurrent:
The avalanche destroyed the houses
Some languages have causative cases or special causative endings which
allow us to say that an agent caused (usually in an unspecified way) some-
thing to happen or be the case, specified by a clausal complement.6 In
English this is effected by certain verbs such as ‘make’, ‘have’, ‘get’, and
‘cause’, which are accordingly called causative verbs:
6 For a taxonomy of causatives see Dixon (2000).
240 PETER SIMONS
Jane had the car serviced
Sam made Mary get up
Henry got Sam to leave the room
Alfonse caused the table to tip over
As can be guessed from these samples, agentive and causative verbs more
often take continuant subjects than occurrent subjects. But this linguistic
fact cuts no deep metaphysical ice. It is important at the descriptive,
everyday level but only a datum at the level of fundamental metaphysics.
It does not prove that continuant causation is more basic than occurrent,
since it is almost certainly due at least in good part to the fact that we
ourselves are continuants and have a wide range of verbs to talk about
ourselves and our actions. So now we can formulate our
Main Thesis
Continuant Causation as a Way of Explaining
Freedom
Since agent causation is invoked in good part in order to give a metaphysically
defensible account of freedom, let’s take an example of a free action, one well
known because of its historical importance: the assassination of President John
F. Kennedy by Lee Harvey Oswald. The crucial event in the sequence
bringing about the President’s death is the third and fatal head shot fired by
Oswald. The relation betweenOswald and his act is one of agent-causation: it
is a form, arguably the most important form, of continuant causation.
The argument for freedom invoking this analysis is that since Oswald’s
act is not caused by any prior states or events or processes, it is immune to
criticisms to the effect that these and not Oswald are causally responsible for
Kennedy’s death, or that Oswald was merely the passive vessel or conduit
for an ultimately fatal sequence of events. This argument for agent causation
Continuant causation consists in a continuant’s participation in occurrents which are
occurrent causes of further occurrents, which are said by ‘cause’ and by agentive and
other causative verbs to be caused by the continuant in question.
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 241
has suasive power because it offers a way to escape the supposed incompati-
bility of determinism and freedom.
The picture of freedom it portrays is this: the agent, by bearing or being
host to some (typically) mental event C-causes some (typically) bodily event
which has O-effects which (partly) determine the future: i.e., determine
that some things happen that might not have happened and others do not
happen that might have happened. Note that this analysis involves both
agent-causation and mental causation, both C- and O-causation.
This account of free action raises a budget of metaphysical problems,
which must be reasonably resolved before we can be confident we have a
decent account of freedom (more still may be required). They are:
The old problem of free will and determination
The problem of the relation between the mental and the physical
The problem of the relation between continuants and occurrents
The problem of the relation between C-causation and O-causation.
Our position rejects C-causation as primitive and attempts to explicate it via
O-causation. So our take on these problems is: compatibilism with indeter-
minism with regard to freewill, monism with regard to the mental and the
physical, occurrent priority and an abstractionist account of continuants,
and on the main topic of this paper, O-causation priority. While nothing in
this account is original, it is offered as a coherent and naturalistically
acceptable package.
Three Examples
To illustrate our main thesis we consider three examples in greater detail: a
simple Humean billiard ball example, the Challenger accident of 1986, and
again Oswald’s assassination of Kennedy in 1963.
Example 1: Ball hits ball
A white ball, having been hit by a player’s cue, rolls across the baize towards
a stationary black ball. It hits this, causing the black ball to move away.
There are three phases of this scenario. In the first phase, the black ball is
stationary and the white ball is approaching it. The second phase is the
collision, during which the two balls are in contact. We often treat collisions
242 PETER SIMONS
as momentary or instantaneous events, but they cannot be, because it is only
by changes within the two colliding balls that momentum is transferred
from one to the other. When the balls first come into contact, the inertia of
both as the electromagnetic forces holding each together come into oppos-
ition cause both balls to distort internally, becoming flattened along the
plane of their contact. This sets up internal stresses within the balls which act
to restore the balls to their undisturbed spherical shape. These stresses which
operate while the two balls remain in contact slow the incident white ball
down and speed the target black ball up, in a direction and with a speed
determined by various factors. The third phase consists in the black ball
moving away and the white ball either moving differently than before the
collision or now remaining stationary. So when we say:
The white ball propels the black ball away
this is to be analysed metaphysically according to our Main Thesis as:
for some event E, the white ball is the active participant in E and E O-causes the
resulting motion of the black ball
The white ball is the active participant and the black ball the passive
participant in this particular exchange since the white ball is initially moving
but the black ball is not. If both balls were initially moving both would be
active as well as passive. The event E is in our example the collision of the
two balls, but this information, while a reasonable inference, is not analyt-
ically contained in the sentence (e.g. the balls might be both magnetized and
the change caused by magnetic repulsion). We may suppose neither ball is a
more important or major participant than the other.
Here it seems to me we have a perfectly acceptable O-causation account
of the interaction. The balls are participants in events and processes which
stand in O-causal relations.
Example 2: The Challenger Accident
This event was minutely investigated after it occurred. On 28 January 1986 a
space shuttle was launched fromKennedy Space Center at Cape Canaveral on
mission STS-51-L.About73 seconds into the flight a rapid burning of escaping
fuel occurred, causing the shuttle, comprising two solid rocket boosters
(SRBs), an external fuel tank and the space orbiter Challenger to break up.
All seven astronauts on board were killed. Subsequent disaster investigations
by a committee of inquiry set up by President RonaldReagan established that:
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 243
A combustion gas leak through the right Solid Rocket Motor aft field joint
initiated at or shortly after ignition eventually weakened and/or penetrated the
External Tank initiating vehicle structural breakup and loss of the Space Shuttle
Challenger during STS Mission 51-L.7
Investigation of the debris and footage of the launch established the causes
beyond doubt. They were:
Conditions
Low ambient launch temperature of 2ºC (–2ºC at the crucial place)
Low resiliency of SRBO-ring seals at low temperatures (factor 5 compared to 24ºC)
Out-of-round sections due to prior use (wider gap between tang and clevis at the
seal joint).
These standing conditions, coupled with the build-up of forces and tempera-
tures occurring during the launch led via a sequence of events, detailed in the
investigation, to the explosion and disintegration of the vehicle. The narrative
of the Report, whichmakes sobering reading, is simply a string of events, one
after the other. Continuants are active or passive participants in the events.
They figure chiefly as bearers of causal powers (roundness, resiliency, com-
bustibility, exerting pressure, being under pressure . . . ). The report is under-
standably coy about mentioning the crew, whose deaths constituted the
tragedy of the disaster.
Once again, the causal story is one of event after event, with the nature of
the causation made manifest by the natures of the successive events. Most
disaster reports by their nature have this form.
Example 3: The Kennedy Assassination
We now come to a case of human action, so one closer to the issues of
freedom and continuant causation. It is one we mentioned before. Again
this is an event which has been anatomized minutely many times since it
occurred on 22 November 1963. The Warren Commission Report con-
cluded that the causal story is like this:
Conditions
Oswald hates Kennedy
Oswald is a trained marksman
7 Rogers et al. (1986).
244 PETER SIMONS
Oswald owns a rifle
Oswald works in Dallas at the Texas Book Depository.
Events and Processes
Oswald learns of Kennedy’s visit to Dallas
Decides to (forms intent to) attempt to kill Kennedy
Travels to Irving Texas to get rifle
Returns with rifle
Smuggles rifle into ‘Sniper’s Nest’ in Texas Book Depository next to motor-
cade route
Presidential car drives through Dealey Plaza
On seeing Kennedy, Oswald fires twice: one hit (back of neck), one miss
Reloads breech by operating rifle bolt
Aims again
Squeezes trigger
Rifle fires
Bullet traverses intervening space
Bullet hits Kennedy in the head
Impact causes massive brain trauma, as a result of which
Kennedy dies.
The crucial event in this sequence is clearly Oswald’s firing the third shot.
Assuming that this was a free action on his behalf, we may ask what caused
it. Proponents of the agent causation theory claim it wasOswaldwho caused
it, he did that (maybe an act of volition, maybe a basic voluntary movement)
which caused the gun to fire and thereby caused Kennedy to die. Oppon-
ents of C-causation as ontologically basic deny this and claim that the
cause(s) of the crucial event lie in events, processes and states in and around
Oswald leading up to the firing of the fatal shot. That we say Oswald caused
Kennedy to die is because there are events in which Oswald is the active
participant which caused Kennedy to die, as per our main thesis.
There being these causes does not take away fromOswald’s responsibility
for his action, since the occurrent causes do not pre-empt Oswald from
acting: rather they are part of what his so acting consists in. His action is
free because he is in control: that means he consciously initiates, monitors,
and steers the action. We know from our own case what this is like so there
is no point in belabouring it. If someone claims not to knowwhat it is like to
act freely they are either confused, mistaken, or lying. Since Oswald’s act of
pulling the trigger is causally determined by prior events and conditions, and
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 245
yet is free, I am subscribing here to a form of compatibilism. Since the world
is in general indeterministic there is no need to worry about global deter-
minism, but I agree with those determinists and compatibilists who consider
that if there were sufficiently influential random factors at work in Oswald’s
action then he would not be in control, so for these local purposes we may
consider the act as determined, unproblematically O-caused, with the
crucial O-causing events being within Oswald.
The analysis of Oswald’s action makes him the agent as well as the host of
the crucial initiating events. I claim that there is no competition between
Oswald and the causing events for responsibility, since the term ‘responsi-
bility’ is being used in two different ways when we say the causing events
and Oswald were responsible for the action.
The Package
So to account for freedom within our assumptions, we consider that a
person P freely does an action A when:
P considers whether to do A or not,8 P decides to (forms an intent to) do A,
which either directly or in conjunction with later events O-causes A, where
P does not act under external duress or internal processes over which they
exercise no control.
We have not explicated what is involved in detail when P consciously
monitors and has control of A and some of the processes leading up to it.
Nor will we: the empirical detail is not for philosophy, though the philo-
sophical minimal requirements are that P be aware of the intention to do
something of the type of A and that P at some stage or other shortly before
doing A have been capable of not doing A. Freedom of action is an outcome
of considerable complexity in mentally endowed continuants, not a meta-
physical primitive. This view is consonant with monistic naturalism, and it
accepts continuant causation, but treats it as ontologically secondary to
occurrent causation. It is further open to confirmation or refutation in the
way any empirical account should be: by advances in science, in this case
8 Typically: in routine cases or cases where the action is part of an intended train, individual deliber-
ation about A may be lacking, but it is part of an intended sequence.
246 PETER SIMONS
brain science. It is also a hostage to various possible metaphysical (mis)
fortunes: determinism, dualism, idealism, panpsychism, and primitive
agent causation. I think it is a better framework for explaining freedom
than these other theories, but whether it or something recognizably similar
is right, time and research alone will tell.
References
Dixon, R. M. W. (2000). ‘A Typology of Causatives: Form, Syntax and Meaning’.
In R. M. W. Dixon and A. Y. Aikhenvald (eds.), Changing Valency: Case Studies
in Transitivity. New York: Cambridge University Press: 30–83.
Johnson, W. E. (1924). Logic, Vol. 3. Cambridge: Cambridge University Press.
Lewis, D. K. (1986). On the Plurality of Worlds. Oxford: Blackwell.
Rogers, W. P. et al. (1986). Report of the Presidential Commission on the Space Shuttle
Challenger Accident. Available at http://history.nasa.gov/rogersrep/genindex.htm
Simons, P. M. (2000). ‘Continuants and Occurrents’. Proceedings of the Aristotelian
Society, Supp. Vol. 74: 78–101.
—— (2009). ‘OnticGeneration:Getting Everything from theBasics’. InA.Hieke and
H. Leitgeb (eds.), Reduction—Abstraction—Analysis. Frankfurt/M: Ontos: 137–52.
CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 247
11
There is No Exclusion Problem
STEINVOR THOLL ARNADOTTIR
AND TIM CRANE
Introduction
Many philosophers want to say both that everything is determined by the
physical and subject to physical laws and principles, and that certain mental
entities cannot be identified with any physical entities. The problem of
mental causation is to make these two assumptions compatible with the
causal efficacy of the mental. The concern is that this physicalist picture of
the world leaves no space for the causal efficacy of anything non-physical.
The physical, as it is sometimes said, excludes anything non-physical from
doing causal work.
The general shape of the problem is not new. Leibniz famously argued
that Descartes’s conception of the relationship between mind and body had
no place for mental causation. On Descartes’s view, according to Leibniz,
the mind can only affect the body by changing the ‘direction of motion’ of
the body’s ‘animal spirits’. Descartes had held that in this way the total
‘quantity of motion’ was conserved in psychophysical interaction. But
Leibniz claimed that what should be conserved in these interactions is not
quantity of motion but (as we would now put it) quantity of momentum
(mass times velocity). So the mind cannot alter the direction of motion of the
animal spirits without altering the quantity of momentum in the physical
world. The physical law that Leibniz took himself to have discovered
excludes the mental from making a causal difference.1
1 For Leibniz’s views, see Leibniz (1695, 1696). For contemporary discussion, including of the
question of whether Leibniz had correctly interpreted Descartes, see Garber (1983) and Woolhouse
(1985).
Leibniz’s objection to Descartes was based on his view of the nature
of the physical world. Contemporary philosophers also see the problem of
mental causation as arising from assumptions about the physical world (see
Papineau 1990). Partly because of the need to accommodate mental caus-
ation given these assumptions, many philosophers have sought to find a
more intimate connection between the mental and the physical, holding
that mental entities are determined by or constituted by physical entities. But
the problem of mental causation has not gone away. The dominant worry
about mental causation in the last few decades is that the physicalist prin-
ciples to which most contemporary philosophers subscribe still leave no
space for mental causation.
First, it seems that anyone who holds that everything is determined by the
physical will have to say that the mental has physical effects if it is to have any
effects at all. Whatever else the determination thesis might involve, it
involves at least the thesis of global supervenience; the thesis that any
minimal physical duplicate of the actual world is a duplicate in every
respect—a duplicate ‘simpliciter’ ( Jackson 1998; Lewis 1986, 1994). In
other words: fix all the physical facts, and you fix all the facts there are.
But if no changes are possible without the appropriate changes in the
physical, then it is very reasonable to think that the mental must bring
about physical changes if it is to bring about any changes at all.2
Second, there is another widely accepted principle known as the causal
closure of the physical or completeness of physics, which claims that any
physical effect has sufficient physical causes.3 What this means is that
physical entities alone fix the occurrence of any physical effect (either
deterministically, or by fixing the chance of the effect). As we might put
it, once all the physical entities are in place, you don’t need to add anything
in order to get the effect to occur. ‘[P]hysics is causally and explanatorily self-
sufficient: there is no need to go outside the physical domain to find a cause,
or a causal explanation, of a physical event’ (Kim 2005, 16). But if that’s true,
according to this line of thought, then there simply is no room for their
2 This doesn’t strictly follow. All that follows is that in order for the mental to bring about any changes
in the world somethingmust cause changes in the physical. But we will ignore this here. We find it highly
plausible, in any case, that if your desire is to cause you to walk to the fridge it must do so by affecting
some physical change in your body.
3 One might reasonably question the truth of the closure principle (see, e.g., Sturgeon 1998;Cartwright 1999) but we will grant it for the purposes of this discussion.
THERE IS NO EXCLUSION PROBLEM 249
having any mental causes as well. Not, at least, if mental causes are distinct
from physical causes.
The exclusion problem, then, results from the alleged fact that the three
following theses cannot be jointly held:
Mental–Physical Efficacy: There are mental causes of physical effects.
Physical Causal Closure: All physical effects have sufficient physical causes.
Non-Reductivism: Mental causes are distinct from physical causes.
This has been particularly forcefully argued by Jaegwon Kim in a body of
work spanning more than two decades (see, e.g., Kim 1989, 1998, 2005).
And the intended moral is that as nobody ought to reject efficacy or closure,
the incompatibility shows non-reductivism to be untenable. Our aim here
will be to defend non-reductivism against this charge.
Philosophers have responded to the exclusion problem in a number of
different ways, some of which involve significant revisions of ontological
and metaphysical assumptions. Some believe that the problem can only be
solved by returning to reductive physicalism (e.g., Papineau 1990; Kim
1998); others that it should be solved by some kind of dualism (Lowe
2008); or that it needs a different account of the relata of causation (Robb
1997; Ehring 1999; Macdonald and Macdonald 1986; Gibb 2004); or that it
needs a different account of causation (Menzies 2008; Raatikainen 2010).
This is the form the debate has taken in the last decade or two. In this paper,
we shall return to the general form of the mental causation problem and
question one of the assumptions on which it rests.
In our view, we do not need to adopt a new metaphysics of causation, or
of the causal relata, in order to defend non-reductivism against the exclusion
problem. What we need is a proper demonstration of the assumptions
underlying the problem. Moreover, in order to resist the problem in the
way we suggest, we do not need to adopt any specific ontological views
about the relata or metaphysics of causation, or on fundamental ontology in
general. The correct metaphysics of causation is, of course, a huge and
important topic, and a full understanding of mental causation obviously
needs a full understanding of causation. But a response to the exclusion
problem does not need this. To the extent, then, that the exclusion problem
is the mental causation problem, the mental causation problem does not
await an answer to the question of what the correct fundamental ontology
should be.
250 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
1. Whose Problem is It?
Standardly, the exclusion problem is thought to be a problem for all and
only non-reductive physicalists. But the problem arises neither for all nor for
only non-reductive physicalists. In its current form, the exclusion problem
originates from a misguided criticism of Davidson’s anomalous monism in
the 1980s. Davidson’s critics complained that although his theory made
room for mental causes by identifying them with physical causes, it did not
give an adequate causal role to the mental ‘as mental’ (see, e.g., Stoutland
1980; Honderich 1982; Johnston 1985). In essence, the complaint was that
on Davidson’s account all the work is done by the physical properties of
events and this leaves no room for a causal contribution by mental proper-
ties.4 This criticism of Davidson is based on a misunderstanding of his theory
of events and causation, as a number of writers pointed out (e.g. Smith 1984;
Crane 1992; McLaughlin 1993). For Davidson himself denies that properties
play any role in causation. Causation, he holds, relates events as such, and
does not hold ‘in virtue of ’ the properties of those events, or in virtue of
how they are described.We are here using the term ‘non-reductivism’ to be
the thesis that mental causes cannot be identified with physical causes. But
sometimes the term is used more narrowly, for the thesis that mental
properties cannot be identified with physical properties. And if the term is
used in this latter sense, then the exclusion problem does not arise for all
non-reductivists. For Davidson is a non-reductivist in this sense and does
not face the problem.5
The exclusion problem, as it is standardly put forward, is a problem for
those who deny the identification of mental causes with physical causes, while
accepting supervenience and closure. This includes, but is not restricted to,
non-reductive physicalists. For non-reductivists may accept supervenience
4 More specifically, this is because causation implies laws, but mental events can be seen as instances of
laws only under their physical descriptions. But Davidson (1970) had argued that all mental events are
physical events, based on a particular theory of causation—that causation is a relation between events and
that events can be seen to instantiate laws only under some descriptions.
5 It should be noted, perhaps, that although the exclusion problem misses its mark as a criticism of
Davidson, it might nonetheless be thought to be a serious worry for anyone who holds that the (primary)
causal relata are Davidsonian events, while wanting to give a causal role to mental properties. As the
problem is standardly formulated, however, in terms of the sufficiency of one cause and the redundancy
of another, it misses the mark against such views. There may be some alternative formulation of the
problem, appealing to in virtue of locutions, on which a tension can be forced. But we shall not explore
this here. It is our belief that if the standard formulation fails, any such weaker formulation will fail also.
THERE IS NO EXCLUSION PROBLEM 251
and closure without subscribing to physicalism. What makes a non-reductive
view physicalist are two things: its commitment to the core physicalist thesis of
global supervenience, and its commitment to the claim that any fundamental
or ‘brute’ facts must be facts within physics (Horgan 1993). A view that accepts
the former claim but rejects the latter deserves the name ‘emergentist’ rather
than physicalist (see Crane 2001; cf. Kim 2010). Emergentists of this kind take
the mental to be intimately dependent on the physical, but they take this to be
a matter of brute fact, rather than a fact explicable by the facts of physics.
Although emergentists need not accept the closure principle, they may well
do so. And if they do, they too need a response to the exclusion problem.
2. What Exactly is the Problem?
But what exactly is the exclusion problem? Supposedly, the truth of the
closure principle precludes non-reductivists from saying that some physical
effects have mental causes. But why should that be?
The closure principle does not entail that physical effects have no non-
physical causes. For although the physical causes by themselves suffice to fix the
physical effects, this does not rule out that certain physical effects (certain bodily
movements, for instance) might have both physical and non-physical causes.
Indeed, it is perfectly compatible with closure that certain physical effects
(actions, for instance) always have both physical and non-physical causes. So
non-reductivism, it would seem, is not in fact incompatible with efficacy and
closure. To get from the claim that all physical effects have sufficient physical
causes to the conclusion that physical effects have only physical causes, Kim
relies crucially on two further assumptions:
Denial of Overdetermination. Mental causes do not overdetermine their effects.
The Exclusion Principle: ‘No single event can have more than one sufficient cause
occurring at any given time—unless it is a genuine case of overdetermination.’
(Kim 2005, 42)
It is against the backdrop of these assumptions that non-reductivism can be
seen to be incompatible with closure and efficacy. Indeed, if we combine
these assumptions with closure and efficacy, we get a valid argument for the
falsity of non-reductivism:
252 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
1. Mental–Physical Efficacy: There are mental causes of physical effects.
2. Physical Causal Closure: All physical effects have sufficient physical causes.
3. Denial of Overdetermination: Mental causes do not overdetermine their effects.
4. The Exclusion Principle: ‘No single event can have more than one sufficient
cause occurring at any given time—unless it is a genuine case of overdeter-
mination.’ (Kim 2005, 42)
Therefore,
C. Identity of Causes: Mental causes are identical to some physical causes.
The argument so formulated is neutral on what types of entities causes are. If
you think causes are events, then you will take this as an argument for the
identity of mental and physical events. If you think causes and effects are
properties, you will take it as an argument for the identity of mental and
physical properties. Kim’s own view is that instantiations of properties (or
what he calls ‘events’) are causes, and he argues via the identity of such
property instantiations to the identification of mental and physical proper-
ties (see e.g., Kim 1998). We will not assume any particular account of the
causal relata in our discussion. We think the argument fails to establish
the identity of causes, regardless.
A number of philosophers have recently rejected Kim’s exclusion argu-
ment (see especially Bennett 2003; Loewer 2007; Raatikainen 2010) by
questioning some of the assumptions behind it. While these responses have
their merits, they depend on adopting some more or less controversial theses
in the philosophy of language or metaphysics (e.g. Lewis’s semantics
for counterfactuals, causation as counterfactual dependence, or causation as
difference-making). Our approach, by contrast, is to undermine the argument
by making the weakest possible assumptions about causation and ontology. We
will argue that the exclusion principle is contrary to our ordinary judgements
about causation, has no strong independent defence, and ought to be rejected.
3. Overdetermination
Before getting on to our criticism of the exclusion principle, we need to
explain first what is meant by overdetermination and why mental causes and
physical causes do not overdetermine their effects.
THERE IS NO EXCLUSION PROBLEM 253
The cases most naturally thought of as cases of genuine causal overdeter-
mination are cases where there are two or more causes, each of which
would have been sufficient to bring about the effect in the absence of the other.
A classic morbid example is death by two assassins working independently,
where a man is killed by two bullets hitting him at once. Another is where a
window is shattered by two rocks making impact with it at the very same
time. What such cases have in common is that in whatever way the causes
are sufficient, they are sufficient independently of each other. This is shown by
the fact that had either of the causes been deleted, without anything being
added in its place, the other cause would still have caused the effect. What
makes the death of the assassinated man overdetermined, for instance, is that
had one of the assassins not shot him, he would still have been shot dead by
the other assassin. In his recent work, Kim has been explicit that in
formulating the exclusion principle and denying overdetermination he has
in mind overdetermination of this standard type.6
We should agree with Kim that mental and physical causes do not
overdetermine their effects in this way. This is because mental and physical
causes are not independent of one another. This leaves open, however, that
mental and physical causes are nonetheless numerically distinct. For distinct
sufficient causes need not be independent sufficient causes. Indeed, that there
might be distinct but dependently sufficient causes is what we should expect
if physicalism is true; and also if emergentism is true. Physicalists and
emergentists hold that the mental is very intimately dependent on the physical,
but identity is only one way in which this need be so. One popular way of
spelling out this dependence is to say that mental properties are realized by
physical properties, where realization may be understood as follows:
To realize is to ‘make real’ in a sense of ‘makes’ that is constitutive rather than
causal. So a property-realizer of a property is a property whose instantiation
constitutively makes real an instantiation of the realized property. (Shoemaker
2007, 10)
How exactly a theory of realization should be developed is a question that
we need not address here. The important thing in the present context is just
6 ‘The usual notion of overdetermination involves two or more separate and independent causal
chains intersecting at a common effect. Because of Supervenience, however, that is not the kind of
situation we have here. In this sense, this is not a case of genuine causal overdetermination’ (Kim
2005, 48).
254 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
that in claiming that the relationship between mental properties and their
physical realizers is constitutive, the physicalist requires a tighter relationship
than mere correlation (even if this correlation is nomological). The rela-
tionship that holds between mental property instantiations and their phys-
ical realizers is not akin to that which holds between smoke and fire, for
instance. It is more closely analogous to the sort of relationship that holds
between a statue and the lump of clay that makes it up. (That, of course, is a
relationship between particulars; but realization in the relevant sense is
intended to be an analogous relationship between properties.)
This analogy between realization for properties and constitution for
particulars is very helpful in evaluating both the exclusion principle and
the denial of overdetermination. Given this understanding of the relation-
ship between the mental and the physical, mental causes and their physical
realizers could not overdetermine their effects in the ‘independent assassins’
way. For holding everything else fixed, you could not delete one of them
from a given context without thereby deleting the other. If such a relation
were to hold between properties, then this is an excellent reason for denying
that mental causes and their realizers overdetermine their effects.
The important thing to stress here is that even when combined with
mental–physical efficacy and the closure principle, the denial of overdeter-
mination gives us no obvious reason to reject non-reductivism. For an effect
may have distinct causes without being overdetermined, granted that the
causes in question are suitably dependent (see Mellor 1995, 103–4 for a
similar point).
4. The Exclusion Principle
It is the fourth premise of the exclusion argument, the exclusion principle,
that challenges the compatibility of non-identity with premises (1)–(3).
According to this principle ‘no single event can have more than one
sufficient cause occurring at any given time—unless it is a genuine case of
overdetermination’. And if that’s true, then it is clear that non-reductivism
cannot make room for mental causation, given the truth of closure and
the denial of overdetermination. It is on this principle that the case against
non-reductivism rests and the remainder of our discussion will be devoted
to examining it.
THERE IS NO EXCLUSION PROBLEM 255
4.1. Causal Exclusion and Explanatory Exclusion
How did the exclusion principle enter this debate? In his first significant
intervention in this debate, ‘Mechanism, Purpose and Explanatory Exclu-
sion’ (1989), Kim talked in terms of what he called the principle of explana-
tory exclusion: ‘no event can be given more than one complete and
independent explanation’ (1989, 79). But over the years, Kim has changed
from talking about explanatory exclusion to talking about causation. His
causal exclusion principle states that ‘[n]o single event can have more than
one sufficient cause occurring at any given time—unless it is a genuine case
of overdetermination’ (Kim, 2005, 42).
Kim does not say why he has moved from talk of explanation to talk of
causation. But this move is certainly a move in the right direction, since the
problem of mental causation is a problem about causation and not explan-
ation.7 There is not even a prima facie difficulty of mental explanations being
incompatible with, or ‘crowded out by’ physical explanations (see Burge
1993). Any occurrence can be explained in countless ways, and there is no
incompatibility between any physical explanation of an event and a mental
explanation of the same event. So there seems to be little plausibility to the
idea that one explanation ‘excludes’ another.
However, Kim’s ‘explanatory exclusion’ principle does talk about
complete explanations. What is a complete explanation? Suppose it is the
conjunction of all the many different true explanations of the event—if we
can make sense of this idea. Then of course, no event can have more than
one complete explanation in this sense. So it seems that the explanatory
exclusion principle is either obviously false or trivially true. The same is not
true of the causal exclusion principle. Though perhaps it is the connection
with the (possibly illusory) idea of a ‘complete’ explanation of an event that
lies behind Kim’s thought that the exclusion principle is an obviously true
principle that requires no defence.
7 It should be said that how closely related explanatory and causal exclusion are depends very much on
your account of causation and of causal explanation. On Kim’s account the same things can serve as the
relata of causation and causal explanation (i.e., property instantiations or facts) while for others, e.g.,
Davidson, causation and explanation relate very different things. It is only on the latter sort of view that
causal explanatory exclusion is much less plausible than causal exclusion. It is fair to say, however, that if
one wants to press a perfectly general problem of mental causation, one ought to avoid building in
assumptions about the relation between causation and explanation. And on its face the principle of causal
explanatory exclusion seems a lot less plausible than the principle of causal exclusion.
256 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
4.2. Why the Causal Exclusion Principle is Substantive
Kim remarks that the causal exclusion principle is ‘virtually an analytic truth
with not much content’ (2005, 51). But given what he means by ‘genuine
overdetermination’, and given what he has to mean by ‘more than one’, this
claim is clearly mistaken. There is a principle in the vicinity of the exclusion
principle that is a better candidate for being ‘virtually an analytic truth’;
namely, that no single event can have more than one independently sufficient
cause occurring at any given time—unless it is a genuine case of overdeter-
mination. That much just falls out of the notion of overdetermination under
discussion. However, that principle entails no stronger conclusion than that
mental causes are not sufficient independently of physical causes, and as we
have seen that is not a claim that is in tension with non-reductive physical-
ism or emergence.
Kim’s causal exclusion principle is not specified in terms of independently
sufficient causes, but only in terms of distinct sufficient causes. It says that
there cannot be distinct sufficient causes of an event occurring at any given
time except in cases of genuine overdetermination. This claim, we argue, is
false. We have two main objections. First, it seems to us that far from being
an analytic truth, the exclusion principle is not even plausible on its face. It
conflicts with our causal judgements even before any physicalist commit-
ments enter the picture and is subject to a number of counter-examples.
Given this, we ought to demand some very good arguments to persuade us
that the principle is true. Our second objection is that we lack such
arguments.
4.3. On the Implausibility of the Exclusion Principle
Notice, to begin with, that the principle as stated involves the important
qualification ‘at the same time’. Without this qualification it would be
refuted by the simple fact that every effect has many causes, stretching
back across time. Think of a causal ‘chain’ where A causes B and B causes
C. Even without assuming the transitivity of causation, A and B can both be
causes of C. And if all causes are sufficient causes, then an effect can
therefore have many sufficient causes across time: everyone should accept
this. This is presumably why Kim adds the qualification, ‘at the same time’.
But why should it be more plausible with this qualification added? Even
at a time, events have many causes, and not just in cases of genuine
THERE IS NO EXCLUSION PROBLEM 257
overdetermination. J. L. Mackie’s (1965) famous discussion of a short-circuit
causing a fire illustrates this nicely. Putting to one side the distinction
between cause and background conditions, it is natural to say that the
presence of oxygen and the presence of flammable material are causes of
the fire, just as much as the short-circuit is. But these are states of affairs, or
property-instances, that exist at the same time as the fire. So the short-
circuit can have many causes occurring at the same time.
But it may be said (as Mackie himself did) that none of these are sufficient
causes. If ‘sufficient’ means absolutely sufficient on its own, then they are not
sufficient causes. But is there anything which is absolutely sufficient on its
own for the occurrence of an event? Those who believe in sufficiency in
this sense might appeal to the fact that there is an entire state of the universe
before the occurrence of the event which is sufficient for that event’s
occurrence. Now if determinism is true, then there must be such a state.
Whether or not this gives any plausibility to the claim that no event can
have more than one sufficient cause at the same time depends on the
relationship between this idea of the entire state of the universe, and the idea
of something’s being a cause, or a sufficient cause. Of course, there have been
theories which make a close connection between these ideas—Mill’s notion
of the ‘whole cause’ is the most famous—but these are specific accounts of
causation, and are not uncontroversial.
By contrast, the exclusion argument is not supposed to rely on any
particular account of causation. The exclusion principle is intended as a
general principle that one ought to accept whatever one’s account of
causation and the causal relata: so we should expect the principle to accord
with ordinary causal claims. But it does not appear to do so. Here is an
example. We are in general happy to attribute causal powers to ordinary
objects (see Lowe 2008). We say things like ‘the furniture scratched the
floor’, ‘my shoe gave me blisters’, or ‘the hammer made an indentation
in the clay’, for instance. But where we are happy to say that objects
caused things we are often happy to say at the same time that their parts
caused things. Suppose, for instance, that the indentation that my hammer
makes in the soft clay on top of which I place it, is made by the hammer’s
head and not by its shaft. There seems no tension in saying both that the
hammer caused the indentation in the clay and that its head did. But the
hammer and its head are numerically distinct things, so this would violate
the exclusion principle—so long as they are sufficient causes.
258 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
Are they sufficient causes? Not, of course, in the sense of ‘absolutely
sufficient’—the sense in which Mill’s entire state of the universe is suffi-
cient. But they are sufficient in the sense in which any cause we can know
about is sufficient. The causes that we cite as sufficient or necessary are only
sufficient or necessary given other factors, including other causes and maybe
the laws of nature. This is what people mean when they sometimes say that
causes are sufficient ‘in the circumstances’. Our everyday commonsense
way of thinking about causes—part of the data which the metaphysics of
causation is arguably supposed to explain—only requires causes to be suffi-
cient in this sense.
4.4. On the Lack of Argument
Given the above considerations, if we are going to defend the exclusion
principle, we need a strong argument. Our second main claim is that
we lack such an argument. Kim suggests in a number of places that it
would be very odd if actions always had two sufficient causes.8 But why is
it odd? One answer is that it is odd if the causes are not just distinct but also
independent and absolutely sufficient. It would, of course, be an astonishing
coincidence if our actions always had two independent and absolutely sufficient
causes, and this would cry out for some explanation. But the non-reducti-
vist can deny both that the mental and physical case is a case of two
absolutely sufficient causes, and that it is a case of independent causes.
First, non-reductivism can deny that the causes are absolutely sufficient.
Physicalists will want to say, of course, that the physical causes are as close to
absolute sufficiency as any cause gets. For example, they may say that given
determinism and the transitivity of causation, the state of the universe at the
Big Bang is a cause of today’s weather in Iceland. Of course, even this is only
sufficient given the laws of nature, on the usual way of thinking about these
matters. But nonetheless it is as close to absolute sufficiency as we get.
Second, the important point for non-reductivists is that the mental cause
of a physical effect is not a candidate for being sufficient in anything like this
absolute sense, given that it is dependent on its physical basis. Moreover,
this dependency explains why the mental cause is not sufficient in the
sense of being absolutely sufficient. So once we acknowledge this intimate
8 E.g. ‘[i]t is at best extremely odd to think that each and every bit of action we perform [has] two
distinct sufficient causes’ (Kim 1989, 86).
THERE IS NO EXCLUSION PROBLEM 259
relationship between the mental and the physical—assumed both by
physicalists and emergentists—there is nothing strange about our actions
having both mental and physical causes. It certainly wouldn’t be a coinci-
dence nor would it be something for which we lacked an explanation.
So what is the argument for the exclusion principle? Why should we
accept that there cannot be distinct sufficient causes except for in cases of
genuine overdetermination? It seems to us that the above considerations
seriously undermine this contention. First, we seem only too happy to
speak as if there were distinct sufficient causes where overdetermination
clearly does not apply; i.e., where the causes in question are not independ-
ently sufficient. Second, once we acknowledge that the causes in question
are not independent the reservations that one might have about allowing
distinct sufficient causes ought to let up. Curiously, however, it turns out
that Kim thinks it is precisely because the two causes are not independent that
there is a problem in acknowledging both of them:
[O]ur problem is not exactly that of causal overdetermination, although both
have to do with an overabundance of causes. It is important to see that the
problem that we face arises because the two putative causes are not independent
events. The difficulty is exactly that the causal status of the dependent event is
threatened by the event on which it depends. (1998, 53)
But how exactly is the causal status of the dependent event threatened by
the event on which it depends? Kim’s answer seems to be that given that the
causal powers of the dependent event are determined by the event on which
it depends, it couldn’t bring any additional causal powers to the picture, and
this, he thinks, means that it cannot play any causal role. (This is the basis of
what he came to call in his 1998 book the ‘supervenience argument’.) In
Mind in a Physical World, Kim claims ‘there is a real problem, the exclusion
problem, in recognizing second-order properties as causally efficacious in
addition to their realizers’ (1998, 53, our emphasis). He goes on to explain
what the problem is, as follows:
For there is nothing in the instantiation of [the second-order property] F on this
occasion over and above the instantiation of its realizer H. Given this, to think
that this instance of F has causal powers in excess of these of H is tantamount to
belief in magic. (1998, 54–5, our emphasis)
260 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
Again, the idea is that where an effect already has a sufficient cause, any
simultaneous distinct sufficient cause would have to bring some additional
causal powers to the scene. Let us call this the motivating principle.
We find this motivating principle no more plausible than the exclusion
principle. We do not have to admit that instantiations of mental properties
have additional causal powers (on any given occasion) to those bestowed on
them by their realizers. Indeed, it is difficult to see how anybody who believes
in global supervenience could think this possible. For if the thesis of global
supervenience is true, then fixing the physical facts fixes all the facts there are,
including causal facts. If we have to accept that this rules mental properties out
from being causally efficacious, we still need to know why. Just stating that a
thing could not be causally efficacious unless it brought additional causal
powers to those already determined by the physical seems straightforwardly
to beg the question against those who believe both that global supervenience
holds and that there are causally efficacious things that are neither identical nor
separate entities, but rather stand in the relation of constitution or realization
to one another. In other words, the motivating principle begs the question
against the very people against whom the exclusion argument is directed.
The comparison with particular objects is instructive at this point. Many
philosophers hold that statues are constituted by rather than identical to the
lumps of matter that make them up. Statues and lumps, they claim, are distinct
material objects that share all their matter and microphysical properties. If this
is right, then when you place a copper statue in a tub of water with the effect
that the water level rises, you place numerically distinct material objects in the
water: namely, the statue, and the lump of copper that constitutes it. What’s
more, it seems that the statue is sufficient (given the other causes) for raising
the water level and also that the lump is sufficient (given the other causes) for
raising the water level. But the statue and the lump certainly do not overdeter-
mine the raising of the water level. Nor does the statue add anything to water-
raising powers already put in play by the lump. Those who believe that statues
are distinct from lumps should take this as reason to reject Kim’s principles.9
It might be replied, of course, that we should deny that statues are distinct
from lumps: they are identical. But our present point is that we should not
9 The example can also be reconstructed in terms of property instantiations. Given that property
instantiations are individuated by the individuals instantiating the properties, the lump being submerged
in the water and the statue being submerged in the water are distinct property instantiations, each of
which is sufficient for the water level’s rising; and neither of which adds any causal powers to those of the
other one.
THERE IS NO EXCLUSION PROBLEM 261
insist that this is so simply because at any given time the statue has no causal
powers in addition to those of the lump. Because if statues were in fact
constituted by lumps we should expect them to have no causal powers in
addition to those of the lump at any given time. Note that the ‘at any given
time’ clause is important here. There being a statue can be relevant in all
sorts of ways. And on occasions where the lump does not constitute a statue,
coming to do so might well give it new causal powers. But on any given
occasion where the lump is such that it constitutes a statue, the statue brings
no extra causal powers to those already determined by the lump.
Similarly, we should not argue that mental property instantiations must
be identical to, rather than realized by, physical property instantiations on
grounds that mental property instantiations don’t have any causal powers in
excess of their realizers. For if mental property instantiations were in fact
realized by, rather than identical to, physical property instantiations, that’s
exactly what we should expect.
Our conclusion is that we have no good reason to accept the exclusion
principle. It fits badly with ordinary causal judgements, it is unsupported by
argument, and it begs the question against those who believe in consti-
tutively related causes. Since non-reductivists—non-reductive physicalists
and emergentists alike—agree that the causes in question are intimately
related, there is no reason why they should be moved by the appeal to the
exclusion principle.
5. Concluding Remarks
Once we have rejected the exclusion principle, we have disarmed the
exclusion argument. A non-reductive physicalist or an emergentist can
accept premises (1)–(3) of the argument, so long as they accept that mental
and physical causes are intimately dependent. (It is a further question of
whether this dependence can be explained—i.e., whether dependence is a
‘brute fact’ as some emergentists claim.) Intimate dependency is also de-
pendency of causal powers, and this is why non-reductivists can reject
Kim’s motivating principle that non-identical causes would have to bring
additional causal powers to those determined by the causes they depend on.
262 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
These points are illustrated by the example of constitution between distinct
objects.
What is more, we can disarm the exclusion argument without making
very many heavy-duty assumptions about causation. The only substantive
claims we are making about causation is that every effect can have many
causes, both over time and at a time, and that these causes can be sufficient
(in one of a number of senses typically appealed to by theories of causation).
According to physicalists, closure entails that the physical causes are as
absolutely sufficient for their effects as any cause can be. Mental causes
will not be sufficient in this sense: they will only be sufficient given the
other causes and other factors. But contra Kim, this is not something which
undermines the mental cause’s status as a cause. For one thing, most of the
causes we know about are sufficient ‘in the circumstances’ and not abso-
lutely sufficient. But more importantly in this context, the mental cause is
dependent on the physical cause. This is the essence of what it means to be a
non-reductive physicalist, or an emergentist in our sense. So if these forms
of non-reductivism are true, then we should not expect that mental causes
would be absolutely sufficient.
If the closure principle is true, then the physical cause suffices for the
physical effect. But since the mental cause is dependent on the physical
cause, the latter also suffices for the former. Since the mental cause is, by
hypothesis, a cause of the physical effect, this shows that the physical effect
can have more than one cause. The mental cause is sufficient for this effect
too; it’s just that it is sufficient given the physical cause. But this is the sense
in which most causes are sufficient, or at least the sense in which most
theories of causation allow themselves to talk of ‘sufficient causes’.
It would be a mistake to conclude from this that ‘all that matters’ is the
physical cause. For this would assume that the physical cause could be there
without the mental cause. But this contradicts the supposition, common to
both non-reductive views being considered, that the physical cause neces-
sitates the mental cause (see Loewer and LePore 1987; Loewer 2007). So we
should reject the idea that all that matters is the physical cause. Compare our
analogy with the statue again: just because the statue is determined by the
arrangement of the clay, this does not mean that all that matters is that there
is clay there. It matters that the clay gives rise to a statue.
Our response is available to those holding a wide variety of views about
the relata of causation. Causes can be ordinary objects, or substances (cf.
THERE IS NO EXCLUSION PROBLEM 263
Lowe 2008). They can be events as conceived by Davidson (1967) or
property-instantiations as conceived by Kim (1980) or differently conceived
by Macdonald and Macdonald (1986). They can be facta in Mellor’s (1995)
sense. Or they can be tropes, as Ehring (1999) and Robb (1997) think. None
of these views makes any difference to the way the exclusion problem
should be treated. Once the causal exclusion principle is rejected, then it
is clear that non-reductivism is not threatened by the conjunction of (1)–(3).
The correct ontology can be argued about at a later stage.10
References
Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just
Maybe, to Tract it’. Nous, 37(3): 471–97.
Burge, T. (1993). ‘Mind–Body Causation and Explanatory Practice’. In J. Heil and
A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 97–120.
Cartwright, N. (1999). The Dappled World: A Study of the Boundaries of Science.
Cambridge: Cambridge University Press.
Crane, T. (1992). ‘Mental Causation and Mental Reality’. Proceedings of the Aristo-
telian Society, 92: 1–18.
——(2001). Elements of Mind. Oxford: Oxford University Press.
Davidson, D. (1967). ‘Causal Relations’. Repr. in (2001), Essays on Actions and
Events, 2nd edn. Oxford: Clarendon Press, 149–62.
——(1970). ‘Mental Events’. Repr. in (2001), Essays on Actions and Events, 2nd edn.
Oxford: Clarendon Press, 207–24.
Ehring, D. (1999). ‘Tropeless in Seattle: The Cure for Insomnia’. Analysis, 59(261):
19–24.
Garber, D. (1983). ‘Mind, Body and the Laws of Nature in Descartes and Leibniz’.
Midwest Studies in Philosophy, 8(1): 105–33.
Gibb, S.C. (2004). ‘The Problem of Mental Causation and the Nature of Proper-
ties’. Australasian Journal of Philosophy, 82(3): 464–76.
Honderich, T. (1982). ‘The Argument for Anomalous Monism’. Analysis, 42:
59–64.
10 We would like to thank: the Leverhulme Trust and the Isaac Newton Trust for their generous
support for Steinvor Arnadottir’s research; two anonymous reviewers for OUP for their helpful
comments; and Jonathan Birch for his insightful remarks when some of this material was presented by
Tim Crane in a seminar in Cambridge.
264 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
Horgan, T. (1993). ‘From Supervenience to Superdupervenience: Meeting the
Demands of the Material World’. Mind, 102: 555–86.
Jackson, F. (1998). From Metaphysics to Ethics. Oxford: Clarendon Press.
Johnston, M. (1985). ‘Why Having a Mind Matters’. In E. LePore and
B. McLaughlin (eds.), Actions and Events: Perspectives on the Philosophy of Donald
Davidson. Oxford: Basil Blackwell: 408–26.
Kim, J. (1980). ‘Events as Property Exemplifications’. In M. Brand and C. Walton
(eds.), Action Theory. Dordrech: Reidel: 159–77.
——(1989). ‘Mechanism, Purpose and Explanatory Exclusion’. Philosophical Per-
spectives, 3: 77–108.
——(1998). Mind in a Physical World. Cambridge, MA: MIT Press.
——(2005). Physicalism, or Something Near Enough. Princeton: Princeton University
Press.
——(2010). ‘Emergence: Core Ideas and Issues’. In Essays in the Metaphysics of
Mind. Cambridge: Cambridge University Press: 66–84.
Leibniz, G. W. (1695). ‘New System of the Nature of Substances’. In
R. S. Woolhouse and R. Franks (trans. and eds.) (1998), G. W. Leibniz: Philo-
sophical Texts. Oxford: Oxford University Press: 143–52.
——(1696). ‘First Explanation of the New System of the Communication between
Substances’ & ‘Third Explanation of the New System of the Communication
between Substances’. In R. S. Woolhouse and R. Franks (trans. and eds.) (1998),
G. W. Leibniz: Philosophical Texts. Oxford: Oxford University Press, 187–93.
Lewis, D. (1986). On the Plurality of Worlds. Oxford: Basil Blackwell.
——(1994). ‘Reduction of Mind’. In S. Guttenplan (ed.), A Companion to Philoso-
phy of Mind. Oxford: Basil Blackwell: 412–31.
Loewer B. (2007). ‘Mental Causation, or Something Near Enough’. In
B. McLaughlin and J. Cohen (eds.), Contemporary Debates in Philosophy of Mind.
Oxford: Basil Blackwell: 243–64.
Loewer, B. and E. LePore (1987). ‘Mind Matters’. Journal of Philosophy, 84: 630–42.
Lowe, E. J. (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford:
Oxford University Press.
Macdonald, C. and G. Macdonald (1986). ‘Mental Causes and Explanation of
Action’. Philosophical Quarterly, 36: 145–58.
Mackie, J. L. (1965). ‘Causes and Conditions’. American Philosophical Quarterly, 2(4):
245–64.
McLaughlin, B. P. (1993). ‘On Davidson’s Response to the Charge of Epipheno-
menalism’. In A. Mele and J. Heil (eds.), Mental Causation. Oxford: Oxford
University Press: 27–40.
Mellor, D. H. (1995). The Facts of Causation. London and New York: Routledge.
THERE IS NO EXCLUSION PROBLEM 265
Menzies, P. (2008). ‘The Exclusion Problem, the Determination Relation and
Contrastive Causation’. In J. Hohwy and J. Kallestrup (eds.), Being Reduced: New
Essays on Reduction, Explanation, and Causation. New York: Oxford University
Press: 196–217.
Papineau, D. (1990). ‘Why Supervenience?’ Analysis, 50(2): 66–71.
Raatikainen, P. (2010). ‘Causation, Exclusion, and the Special Sciences’. Erkennt-
nis, 73: 349–63.
Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly,
47: 178–94.
Shoemaker, S. (2007). Physical Realization. Oxford: Oxford University Press.
Smith, P. (1984). ‘Anomalous Monism and Epiphenomenalism: A Reply to
Honderich’. Analysis, 44(2): 83–6.
Stoutland, F. (1980). ‘Oblique Causation and Reasons for Action’. Synthese, 43:
351–67.
Sturgeon, S. (1998). ‘Physicalism and Overdetermination’. Mind, 107: 411–32.
Woolhouse, R. (1985). ‘Leibniz’s Reaction to Cartesian Interaction’. Proceedings of
the Aristotelian Society, 86: 69–82.
266 STEINVOR THOLL ARNADOTTIR AND TIM CRANE
Index
Bold numbers denote reference to illustrations.
Andersen, R. 62–3, 66Anjum, R. 200 n. 6, 201anomalous monism, see monismArmstrong, D. M. 8, 28, 197, 224–9Arnadottir, S. T. 16autonomyof higher-level properties 61of the mental 60, 88, 226of mental causes 131–5of mental properties 61of the physical 20–2; see also causal
closureof psychology 23of psychological explanation 55
Bacon, J. 224 n. 3Bennett, K. 9 n. 10, 21, 81, 132, 220, 253Bjornsson, G. 176 n. 6, 276 n. 7Block, N. 68Boyd, R. 41, 42, 216, 218, 220Burge, T. 215, 256
Campbell, K. 31 n., 107, 216, 217, 224,227–9
Carnap, R. 227causationagent 9, 162–3, 171, 175, 176, 180–90,
234, 239–47continuant 16, 234–47contrastive 93–4counterfactual 11, 19, 93, 174–5, 201, 203as difference-making 14, 60, 72 n. 5, 75,
77–8, 84–5, 253as transference 11, 12, 201 n. 9, 229event 6, 9, 10, 24 n., 167, 169, 186fact 9–10generalist 11higher level 126–7macroscopic 14, 137, 149 126–51nomological 12; see also genereralist
causation; principle of thenomological character of causality
occurrent 16, 234–47psychophysical 2, 5; see also principle of
causal interactionpowers theory of 15, 194, 196–203, 204,
206, 210, 212pre-emptive 72 n., 93, 95probabilistic 63, 128, 130–4, 159, 178, 188property 92, 111–23singular 6, 11substance 15, 153–71, 183, 185–6, 190supervenient 22and thermodynamics 129–30, 133
causal asymmetry 127–134, 141causal closure 1–5, 12–13, 70, 122, 149–50,
167–9, 193–213, 216–21, 249–53,255, 263
principle of 2, 149, 168causal efficacy 5, 22–3, 46, 50, 64, 71,
134, 148causal exclusionargument from 14, 60, 61–5downward 61, 77–8, 80, 84–5Kim’s argument from 14, 65, 84–5, 88,
90, 253Kim’s principle of 64new principle of 64, 68–9, 71principle of 14, 77–81, 90–1, 193, 252problem of 16, 250revised principle 77–8upward 61, 77–8, 85
causal network 90, 96causal non-overdetermination 2–5, 13
principle of causal non–overdetermination 2
causal overdetermination 2–16, 39, 91,169, 252–5
argument from 2causal power, see powerscausal process 14, 58, 94, 129, 138–42causal profile 36, 38, 39, 41, 43–54, 81–2, 98,
101, 110, 117–20, 146, 225causal reductionism 176–7
causal sufficiency 63, 67, 69, 71–2, 77,82–5, 96
Chalmers, D. 219Clapp, L. 41Clarke, R. 175 n. 3, 180–1colors 30, 39, 224co-manifestation, see mutual manifestationcompatibilism 61, 81–4, 185 n. 19, 242, 246continuant 233–7; see also substancesconstitution 55, 101, 104, 105, 118, 119, 220,
255, 261, 263as co-ordination of the small 121and grounding 121, 123physical 100property 112property instance 112, 119
conventionalism (about exactresemblance) 25
counterfactualdependence 19–21, 94, 115, 174, 175,
201–3, 208, 215, 253circumstances 196co-variation 175, 177
counterfactuals 20, 74, 76–80, 84–5, 94, 116,144, 146, 147, 174, 253
and difference making 85, 94Crane, T. 16, 23, 251, 251
Daly, C. 222Davidson, D. 6–9, 22–9, 32, 188 n. 22,
219–21, 251, 256 n., 264Descartes 18–20, 24 n. 2, 58, 65–9, 248–9determinable-determinate 117, 220, 224disembodied 219double prevention 15, 194, 196–213
Ehring, D. 97, 103, 222, 228, 229, 250, 264Ekstrom, L. W. 179 n. 14emergent dualism 89
epiphenomenal 118–19emergentism 16, 190, 252, 254, 260, 262–3epiphenomenalism
of the mental 21, 37, 60, 62, 65, 111, 149,190, 193
of properties 7–8, 69of types 21, 22
eventsDavidsonian 26, 222 n., 251 n. 5identity of mental and physical 3–7, 22,
24–6, 168, 195, 251 n. 4, 253Kim-style 26, 28, 222, 225–6, 253, 264
mental 2–7, 10, 15, 18–26, 154, 169,194–5, 208–12, 220–1, 251 n. 4
monism 4, 7physical 2–4, 6–10, 12, 18, 21–7, 168–9,
193–5, 203–4, 206–12, 249as property instantiations 4, 8; see also
Kim-style eventsexplanation 51–2, 89, 93, 108, 114, 120, 129,
144, 147–8, 176, 181causal 72, 80, 249complete 256contrastive 187–8and control 175, 186–90and double preventers 212intensional 228 n. 6mental 148physical 139, 141, 144, 256problem of 187psychological 19–20, 23, 55rational 165scientific 31, 187thermodynamic 129, 136
explanatory exclusion 256explanatory power 174
Fodor, J. A. 21, 59, 113, 114,135, 216
Freedommetaphysical 15, 175, 176, 190of action 163, 167, 234, 241–2, 246agent-causal account of 174and moral responsibility 173–4and neo-Humeanism 174–7problem of 239of will 163, 169–171, 242; see also the will,
willingfunctionalism 142–43, 148
Gibb, S. C. 8, 15, 201 n. 9, 223–4, 250Gibbons, J. 21, 94, 105Groff, R. 187 n. 20
Heil, J. 10, 14, 31, 109, 177 n. 8, 190 n. 24,197, 215, 225–7
holistic constraints 219Honderich, T. 7 n. 5, 228, 241
Identityconditions 10, 11, 154–5of causes 253and causal profile 81
268 INDEX
of events 4, 195genidentity 237of the mental and physical 4, 22, 37,
65–6, 215–29, 253–4; see alsopsychophysical identity
and ontological dependence 154primitive 97psychophysical 38, 53, 59–60, 216–17of property instances 4, 92, 96–108,
111, 253relative 227of properties 10, 79, 92, 195, 215–29token 22, 25–7type 22, 25–7, 28, 54, 99, 135, 138, 150,
215–29immanence 220–1incompatibilism 185individuation 31, 93, 102, 104, 153, 156–7,
182, 222, 261 n.intentionalaction 1, 9agents 58, 175, 190behaviour 81, 83cause 185content 165, 195indeterministic causes 185
intentionality 190, 225
Jackson, F. 219, 220, 249Jacobs, J. 15, 177 n. 9, 182 n. 18, 186 n.James, W. 210–13Johnson, W. E. 234
Kane, R. 180 n. 14Kim, J. 6–7, 14, 20–1, 26–8, 38, 46, 58, 60,
64, 69–71, 88–92, 99–100, 127, 149n., 168 n., 190 n. 23, 216, 222, 225–6,249–50, 252–4, 257–64
Kornblith, H. 21Kripke, S. 219
laws 140basic, see fundamental lawscausal 11, 21, 23, 50–1, 113, 170–1and causal profile, see causal roleand causal relevance 113and causal role 115, 117ceteris paribus 20, 113, 114dynamical 126–7, 140exceptionless, see strict lawsfundamental 29, 63, 67, 119, 206
as linguistic items 29macroscopic 141of nature 174, 178, 180–81, 227 n., 259and patterns of variation 113physical, see laws of physicsof physics 55, 59, 63, 67, 89, 206, 248physically reducible 141and powers 206and properties 89psychological 101, 112psychophysical 7, 206as regularities 11, 177strict 7, 29, 113, 177and supervenience 140and thermodynamics 129–30, 141
law-statements 29Leibniz 248–9LePore, E. 21, 132, 135, 263levels 67; see also unilevellersof being 10, 13of description 11fundamental 69hierarchy of 48, 69, 83macro to micro 70ontological 11and supervenience 70
Lewis, D. 36, 74–5, 112, 130–1, 174–5, 196,224–5, 234, 249, 253
libertarianismagent- causal 184–6event-causal 177, 179
Libet, B. 209–10Lipton, P. 187List, C. 61, 78, 80, 93, 95, 115–16, 132,
135, 149Locke, J. 154, 160, 164Loewer, B. 21, 132, 135, 215, 253, 263Lowe, E. J. 2 n. 2, 5 n. 4, 9, 10, 15, 183–4
193–5, 207 n., 208, 220–1, 250,258, 264
Macdonald, C. 21, 27–8, 220, 250, 264Macdonald, G. 21, 27–8, 220, 250, 264McGinn, C. 6Mackie, J. L. 258McLaughlin, B. 22, 42, 251Malcolm, N. 219manifestationsof continuants 233of dispositions, see manifestations of
powers
INDEX 269
manifestations (cont.)mutual 184, 197–8, 200of active powers 159, 161of passive powers, see non-causal powersof powers 164, 199, 200, 201, 203–4,
207, 215of non-causal powers 160, 183of two–way powers 164type 153, 156–61unmanifested 197of volition 160, 163, 183
Markosian, N. 185 n. 19Martin, C. B. 31, n., 118, 177 n., 194 n. 2,
197, 207Maurin, A. 227Mellor, D. H. 23, 255, 264mental
causal relevance of the 71–7, 193dependence on the physical 20distinct from the physical, see autonomyefficacy of the 5, 22, 23, 35, 37, 56,
65–71, 179, 180, 220–1, 226,228 n. 6, 248
privacy of the 219subjectivity of the 8, 15supervenes on the physical 22, 27 n., 60,
65–6, 85, 126, 132Menzies, P. 14, 93, 95, 115–16, 132, 135, 149,
174 n. 2, 205 n., 250metaphysical necessity 89, 111, 118metaphysics: see also ontology
causal powers 178, 182–6, 189–90neo-Aristotelian 15, 174, 177–9, 182, 185,
186, 190neo-Humean 174–7, 178, 179, 190substance causal 190trope 88, 92, 103, 105–9
microphysical realization 40, 43–5, 47, 53,56, 137
Mill, J. S. 258–9Mills, E. 21Mind-body problem 18–20, 32, 59, 246–7Molnar, G. 156 n. 7, 194 n. 2, 216monism 29
anomalous 6–8, 22, 251event 4, 7mind/body 235, 242property 29, 30trope 217of the whole 121
monistic naturalism 235
multiple realizability 4, 9, 36, 40, 53, 59, 60195, 219, 220, 223, 225
argument from 3, 30, 195, 216–18,219, 222
Mumford, S. 177 n., 194, 200 n. 6, 201
naturalism 237generative 235monistic 235, 246
necessitate 99, 101, 120, 178, 236, 239causally 2, 13, 159, 168, 263metaphysically 92, 100, 110, 113, 118–20,
126necessitation-basemetaphysical 89–90, 91–4, 96, 112,
115, 118minimal 111–14, 120
necessity 44, 157, 177of indeterminism for freedom 176metaphysical 89, 111, 118nomological 89
non-causalism 180, 184non-reductive physicalism 5, 10–15, 19–25,
39, 54, 60–1, 67, 88–90, 106, 111,123, 135, 138–42, 150 n., 193,216–19, 257
Noordhof, P. 14, 220nothing over and above 59, 112, 220
objectivechance 166, 175, 178, 187and physical 222preference 28probability 159propensity 181, 187relations 11similarities/differences 25
occurrents 233–47O’Connor, T. 15, 162 n. 11, 181–2, 185Oliver, A. 216ontologicalassay 236basis 108, 235category 6, 8, 10, 20, 220dependence 5, 9difference 228independence 154levels 11; see also levelssystem 6, 8
ontologicallybasic 245
270 INDEX
dependent 154distinct 132emergent 186fundamental 121, 180, 186independent 132, 139, 154innocent 218secondary 246serious 29, 196–7
ontology 1, 8, 13, 224, 253, 264emergent dualism 118of events 27fundamental 250neo-Aristotelian 184New 184–6non-reductivist 20powers 15, 89, 117–19of properties 215, 217, 225, 229trope 88, 215–29and universals 98, 227
Papineau, D. 2 n. 2, 14–15, 128, 249, 250particulars 6, 115, 119, 154, 255; see also
individual substancesas concrete 154, 157, 175 n. 4and constitution 255as basic 154
Paul, L. 132 n., 151 n.Pereboom, D. 21, 81, 132, 218, 220physicalism 18, 30, 49, 52–3, 58–61, 63,
67–9, 85, 88–9, 99, 120, 135–6, 149,167–9, 235, 250, 252, 254
about the mental 59type 35
powers 194, 196–7, 201active/passive 153, 158–9, 160, 161,
162, 180agent-causal 153–4, 160–7, 173–90causal 11, 15, 21, 38, 46–7, 51, 56, 68–71,
81–3, 98–110, 117–20, 122–3, 138–9,153, 157–8, 160, 161–71, 177–8,182–90, 205, 223, 228 n., 244,258–62
of choice and action 182conditional 222dualism of 218–19first-order 156fundamental 186higher-order 156, 220individuated by manifestation type 153intentional 66intrinsic 177mental 218–21, 226
non-causal 153, 157–8, 160physical 31, 169, 220properties as 216–18, 223, 225, 229pure 31quality/power distinction 32rational 154, 165and causal relevance 51spontaneous 15, 153–4, 160–1, 164–5,
169, 183as tokens 155–7as truthmakers 29two-way 15, 154, 164–6as types 155–7, 226as universals 227–8
powers metaphysicsevent-causal 184–6, 189–90substance-causal 184–5, 189–90
Price, H. H. 227principle of the anomalism of the mental 7principle of causal closure 2–3, 5, 12–13,
149, 168–9, 206principle of causal exclusion 14, 256 nprinciple of explanatory exclusion 256; see
also exclusionprinciple of causal interaction 7; see also
causation; psychophysicalprinciple of causal non-
overdetermination 2–3, 5, 13; seealso causal non-overdetermination
principle of the nomological character ofcausality 7
principle of suffient reason 188principle of transmission of causal sufficiency
across realization 69, 83problem of control 189propertieshigher order (or: higher level) 41, 46,
50, 56mental, see qualities of experienceMSE-property 36, 40, 48–54MSE*-property 54–5natural 174, 177, 224as particulars 10phenomenal 13; see also qualities of
experiencephysical 4, 11, 18–20, 22–4, 27, 31, 37–9,
53–5, 59, 63, 65, 68, 89–91, 95,97–101, 105–6, 112–13, 116–23, 126,135, 195, 201–2, 219, 227, 251–4
qualitative 30–2, 118, 197, 221, 229; seealso qualities of experience
second-class 224, 228
INDEX 271
properties (cont.)universals 8–10, 26, 97–8, 103, 105, 107,
223–4, 226–8tropes 6, 9–10, 15, 26 n., 97, 106–9,
217–29, 264higher order 221resemblance classes 106–9, 112,217–18, 223–5, 226, 228
property dualism 4, 7, 65, 193, 217, 219principle of 4–5, 9
property-instantiationsidentity of 4, 253as events, see eventsas facts 256 n.mental 39, 255, 262microphysical 40physical 262as states of affairs 8as states of substances 9
Putnam, H. 3
qualitiesof experience 30–2powerful 30–2primary 30secondary 30
qualia 21, 30–1, 225, see also qualitativeproperties; qualities of experience
realizationphysical 39microphysical 40, 43–5, 47, 53, 56, 137property 39–43, 46, 47, 53, 55–6subset view 41, 56
reciprocal disposition partners 197–200resemblance 107, 223, see also tropes as
resemblance classesexact 106–7, 109, 223nominalism 106–8rough 106–7, 224–5
Robb, D. 9, 15–16, 108, 216–17, 219, 225–6,250, 264
roll-back argument 163, 189
Salmon, W. 72Shoemaker, S. 14, 21, 41–2, 61, 81–3, 85, 98,
101, 132, 177 n. 8, 197 n., 220,222–3, 254
Sidelle, A. 227Simons, P. 16, 325 n.Sosa, E. 22, 130–1
Spinoza 24Stoutland, F. 22, 251Strawson, P. F. 154, 173substance dualism 5, 58, 68–9, 194substances 4, 8and agents 162, 186Aristotelian 8as bare particular 8; see also particularsas causes 183–6, 190individual 6, 9, 153–5, 157–9; see also
particularsindividuated by powers 154kinds of 154ontologically emergent 186ontologically independent 154as property bearer 26–9, 195reducible to bundles of properties 8as state of affairs 9as substratum 8
sufficient causes 90–1, 257–9, 263absolutely 258–9independent 254, 257dependent 16, 254distinct 257–60
Tooley, M. 130–1tropes: see propertiestruth conditions 73–8truthmakers 24, 28–9, 31, 110, 216,
227, 240truthmaking 110–11, 123type diversity 25type-type reduction 136
unilevellers 109–11, 117; see also levelsuniversals 8–10, 26, 97–8, 103, 105, 107, 112,
223–8
Watkins, M. 41Whittle, A. 98, 222–4, 227Williams, D.C. 217, 224, 227the will 153–4, 160–7willing 161–2, 164–7, 183Wilson, J. 118, 220–1, 223Woodward, J. 19, 21, 61, 72, 75, 115, 174 n.
2, 216
Yablo, S. 21, 51, 72, 76, 82–4, 95, 132–3,215–16, 219, 220
zombie 219
272 INDEX