Ladyman - Science, metaphysics and method.pdf

8/13/2019 Ladyman - Science, metaphysics and method.pdf

1/21

Science, metaphysics and method

James Ladyman

Published online: 11 May 2012 Springer Science+Business Media B.V. 2012

Abstract While there are many examples of metaphysical theorising being heu-

ristically and intellectually important in the progress of scientific knowledge, many

people wonder how metaphysics not closely informed and inspired by empirical

science could lead to rival or even supplementary knowledge about the world. This

paper assesses the merits of a popular defence of the a priori methodology of

metaphysics that goes as follows. The first task of the metaphysician, like the

scientist, is to construct a hypothesis that accounts for the phenomena in question. Itis then argued that among the possible metaphysical theories, the empirical evidence

underdetermines the right one, just as the empirical evidence underdetermines the

right scientific theory. In the latter case it is widely agreed that we must break the

underdetermination by appeal to theoretical virtues, and this is just what should be

and largely is done in metaphysics. This is part of a more general line of argument

that defends metaphysics on the basis of its alleged continuity with highly theo-

retical science. In what follows metaphysics and theoretical science are compared in

order to see whether the above style of defence of a priori metaphysics is successful.

Keywords Metaphysics IBE Underdetermination Scientific realism

1 Introduction

Many philosophers are not content to think of metaphysics as the mere explication

of our conceptual scheme or schemes, as in Strawsons descriptive metaphysics,

J. Ladyman (&)

Department of Philosophy, University of Bristol, 9 Woodland Road, Bristol BS29YS, UK

e-mail: [email protected]

1 3

Philos Stud (2012) 160:3151

DOI 10.1007/s11098-012-9910-y


2/21

but rather regard it as aiming to describe the truth about objective reality.1 Prima

facie it is puzzling that although we have successful empirical science, philosophers

also carry out a separate form of a priori enquiry into the nature of things. If there

are supernatural entities and realms, or if the world is ultimately mental or spiritual

in nature, then it is reasonable that mere thought or at least non-sensory experiencemight access it. However, contemporary analytic metaphysics does not usually

presuppose such ideas but rather seems to treat of the world under the same

assumptions about it that are made in everyday and scientific investigation. Science

gives us an evidentially and practically supported description of the phenomena of

extraordinary richness for philosophical reflection. The elaborate and complex

nature of the world, for example, the structure and effects of the periodic table of

elements, is much more interesting, and even more far removed from common

sense, than the ancient ideas of substance as water, fire, or some combination of

elements. The a priori speculations of the ancient Greek atomists are a goodexample of how reflection and metaphysical theorising can be of great value, and,

arguably science without it would quickly atrophy. However, truth is stranger than

fiction in matters of metaphysics, for our contemporary atomists talk about hidden

dimensions, virtual particles, ghost fields, holographic projections, and the

mathematical structure of all these things that makes ancient numerology seem

utterly banal; how could a priori speculation produce ideas that are crazy enough to

be true (the transmutation of base metals into gold being a lucky guess)? More

importantly, science seems to compete with a priori metaphysics when it comes to

the nature of life, matter, mind, space and time and even perhaps causation andmodality. While there are many examples of metaphysical theorising being

heuristically and intellectually important in the progress of scientific knowledge,

many people wonder how metaphysics not closely informed and inspired by

empirical science could lead to rival or even supplementary knowledge about the

world.2

This paper assesses the merits of a popular defence of the a priori methodology of

metaphysics that goes as follows. The first task of the metaphysician, like the

scientist, is to construct a hypothesis that accounts for the phenomena in question.

This usually means deploying some key terms in definitions and axioms, and this is

non-trivial since one must arrive at a consistent theory this is also compatible with

the data, and it must be parsimonious enough in its concepts and claims to be

cognisable.3 To achieve the latter, as in science, abstraction and idealisation is

necessary, and as in science the elegance with which all this is done is part of the

prize. It is then argued that among the possible metaphysical theories, the empirical

evidence underdetermines the right one, just as the empirical evidence underde-

termines the right scientific theory. In the latter case it is widely agreed that we must

1 See Strawson (1959) for his distinction between descriptive and speculative metaphysics. For a

contemporary metaphysician who is clear about the ambitions of metaphysics in this regard see Sider(2011).2 Of course, others such as Van Fraassen (2002) and Price (2009) reject metaphysics altogether.3 In a symposium in this journal on van Fraassens The Empirical Stance I argued that he understates

the difficulties in constructing a coherent metaphysical theory that is compatible with known science

(2004, p. 133).

32 J. Ladyman

1 3


3/21

break the underdetermination by appeal to theoretical virtues, for example,

simplicity or explanatory power, and this is just what should be and largely is

done in metaphysics.4 This is part of a more general line of argument that assumes

scientific realism and then defends metaphysics on the basis of its alleged continuity

with highly theoretical science. In what follows metaphysics and theoretical scienceare compared in order to see whether the above style of defence of a priori

metaphysics is successful. In the next section, the relationship between scientific

realism and metaphysics is introduced, and it is argued that the recent rise of the

latter is linked to the development of the former. In Sect. 3 the content and

methodology of analytic metaphysics is briefly considered, before metaphysics and

theoretical science are compared in respect of the nature of the underdetermination

in each case, and briefly also other arguments, in Sect. 4. It is argued that the

explanationist defence of metaphysics is at least problematic on the basis of a

number of disanalogies between theoretical science and the kind of hypotheses atissue in much contemporary debate in analytic metaphysics. Finally, in Sect. 5, it is

suggested that metaphysical theorizing ought to be fruitful for the rest of our thought

especially science, and that metaphysics should be naturalized; following Ladyman

and Ross (2007), a distinctive creative task for naturalistic metaphysicians is

recommended that goes beyond engaging in the metaphysics of particular sciences,

important though that is.

2 Metaphysics and scientific realism

There is a lot of controversy abouthow to define scientific realism and a fair amount

about how to define metaphysics.5 For the moment, take the latter to be enquiry

concerning the most general questions about the nature of reality including, for

example, questions about the nature of matter, abstracta, fundamentality, space

and time, and causation, law, necessity and probabilitythat at least captures

metaphysics pretty well in extension. A preliminary definition of scientific realism

is that it is the claim that we have knowledge of the unobservable entities and

processes posited by our best scientific theories. However, this is not precise enough

since those who agree with this but also regard the ontology in question as ideal or

socially constructed are not properly regarded as scientific realists. (Hence, when

Kant says that he is an empirical realist, the qualification is enough to exclude him

from the camp of scientific realists, and of course many of his followers were

prominent critics of scientific realism including notably Poincare.) The spirit of

scientific realism requires a background commitment to something like mind-

independence or external realism about the physical world in general. Once this is

assumed scientific realism can be defined as a lack of epistemic discrimination

against unobservables, so that truths about electrons are known just as truths about

tables are. The problem with this way of thinking about scientific realism is that it

does not address the seeming incompatibility of everyday and scientific ontology; it

4 A good source for this view is Swoyer (1983) as well as Paul in this volume.5 Of course, there is also lots of controversy about the status of metaphysics, see Chalmers et al. (2009).

Science, metaphysics and method 33

1 3


4/21

is arguable that the table of common sense cannot co-exist with its scientific

counterpart and one or other must be reduced or eliminated so that we must chose

between realism about the manifest and scientific images of the world. In order not

to beg answers to such complex questions it is best to define scientific realism as a

self-standing thesis (though the comparison with everyday realism is importantwhen it comes to the arguments for scientific realism once defined as we will see

below).

The standard definition of scientific realism involves metaphysical, semantic and

epistemic components.6 The former is roughly the aforementioned commitment to

the mind-independence of the unobservable world; the semantic component requires

that theoretical terms and statements be taken literally as putatively referring to

unobservable entities, properties and processes, and that the ordinary conception of

truth as correspondence to reality is operative although perhaps understood in a

deflationary sense;7

finally, the epistemic component is the claim that we haveknowledge of scientific claims involving unobservables which are at least

approximately true.8

The fortunes of metaphysics often run in parallel to the fortunes of scientific

realism. This is not surprising because scientific realism is often taken as equivalent

to the claim that science can deliver metaphysical knowledge about, for example,

the nature of space and time, the nature of matter and perhaps mind, causation, and

so on. Even if that is going too far, scientific realism is at least a necessary condition

for science to deliver metaphysical knowledge. Historically, empiricists have been

the chief critics of scientific realism, and they have derived their criticisms of it, orof particular posits within theories, from their antipathy to metaphysics. Similarly,

in current philosophy of science, van Fraassen argues that scientific realism is

objectionable just because it is metaphysical (1985, 1989), and other prominent

critics of scientific realism are equally opposed to metaphysics (Stanford 2007).

However, it is possible to defend one without the other. For example, scientific

realists among scientists may be hostile to metaphysics, and some metaphysicians

do not look to science for either metaphysical knowledge or methodological

guidance. Historically, it is undoubtedly the case that, while the revival of

metaphysics in late twentieth century philosophy had many causes, it coincided with

the resurgence of scientific realism as both recovered from the assault to which they

were subjected by positivism in both philosophy and science.

The philosophical programme of logical positivism defined itself in part by the

elimination of metaphysics. The positivist reaction to the excesses of metaphysics

6 See Psillos (1996). The discussion of scientific realism in this paper is much expanded in Ladyman

(2002) and Ladyman and Ross (2007).7 There are scientific realists such as Brian Ellis who adopt pragmatic accounts of truth but then much of

the debate about scientific realism becomes moot, since, for example, the question of whether theoretical

virtues are epistemically as well as pragmatically valuable collapses.8 The notion of approximate truth is vital to the defense of scientific realism since no sensible realist

believes that our best current theories are absolutely true. However, it is proved very difficult to arrive at a

theory of approximate truth. We return to this issue below but for now note that it is not at all clear what

approximate truth in metaphysics would amount to whereas in the scientific case we can always fall back

on approximate empirical adequacy.

34 J. Ladyman

1 3


5/21

was to seek a criterion of empirical significance that would reveal metaphysical

hypotheses to be meaningless, or at least to lack cognitive content and so be

irrelevant to science. Perhaps the folklore about the demise of positivism directly

influenced attitudes among late twentieth and early twenty-first century philosophy

more than the real intellectual history. According to simplified narratives thefortunes of logical positivism and its critique of metaphysics turned on the

verification principle and the analytic-synthetic distinction. The former failed to do

its job because it was a metaphysical principle itself being neither tautological nor

empirically verifiable. If metaphysics cannot be avoided we may as well do it

openly and not be in denial about it. Quines critique of the analytic-synthetic

distinction then also spelled disaster for the key empiricist tenet of the positivists

that necessity could be reduced to the analytic and the way was clear to the

metaphysics of modality. Kripkes work in possible world semantics and Lewis

work using possible worlds for a variety of projects in epistemology and semanticsbut also in metaphysics made the latter seem philosophically exciting to many

philosophers, as did itsrevival in Australia by those using it to articulate their views

about mind and matter.9

However, the demise of positivism was also due to the rise of scientific realism in

response to the problems that beset the logical positivists attempt to account for the

use of theoretical terms in science. The nineteenth century positivist Mach thought

that science should not posit unobservable entities like atoms for to do so is to leave

the empirical world for one constructed by metaphysical theorizing free from the

constraints of evidence from the senses. He argued that where neither confirmationnor refutation is possible, science is not concerned (1893[1960, 587]). While, the

logical positivists were greatly inspired by this, and indeed many scientists today

would agree with this statement and testability remains a popular criterion for

science, it remains a slogan without a rigourous account of confirmation and

refutation. The logical positivists contributed greatly to inductive logic and did

pioneering work in statistical inference and probabilistic epistemology, however,

their parallel project to reduce the meaning of the theoretical terms of scientific

theories to their potential observational consequences fared poorly. In particular,

there were two major problems: dispositional terms and theoretical terms especially

those of what I shall call high theory. Examples of the former include catalyst,

and examples of the latter include scalar curvature.

Originally Carnap wanted to define explicitly theoretical terms, but he gave up on

this in the 1930s and settled for partial definition (see his 19361937a, b). The

problem with the latter is that for any theory in which some terms are defined only

partially there will be lots of models and so there is no unique interpretation for it,

hence it would seem to make no sense to talk of it being true or false of the world.

Hempel (1963, p. 695) and Carnap (1939, p. 62) solved this problem by stipulating

that the theory is to be given an intended interpretation; theoretical terms are

interpreted as (putatively) referring to the entities and so on appropriate to the

9 A version of this folklore is presented with due recognition that it is a great simplification of the real

history in Callender (2011). Price (2009) criticizes the claim that Quine made metaphysics respectable.

Friedman (1999) develops a new historiography of the demise of logical positivism.


1 3


6/21

normal meanings of them in scientific (and everyday use). The picture of the

referent of electron as somehow related to classical point particles and apt to be

deployed in an account of the constitution of macroscopic matter is important in

determining to what it is to that the theory of electrons refers. With partial definition

the meanings of the theoretical terms that do not have testable consequences arenonetheless important in determining their reference. The logical empiricists

became committed to this excess or surplus meaning of theoretical terms over

and above the meaning given by the partial interpretation in terms of what can be

observed. Feigl recognised this explicitly and argued for the view that theoretical

terms genuinely refer to unobservable entities as a consequence (1950).

Carnap (1939) eventually gave up on the requirement that all theoretical terms be

related to an observational basis altogether, though he continued to seek an account

of the cognitive significance of theoretical terms (1956). Meanwhile, of course the

development of science in the twentieth century eventually made Machianreservations about unobservable entities seem idle as the new experimental

technology of bubble chambers, X-ray spectroscopy and electron microscopy

opened up a new world of phenomena with atoms at their heart. The view that

theoretical terms could be analysed in terms of observations and empirical evidence

came to seem irrelevant as the working commitment to atoms and the sub-atomic

world became routine in experimental science and engineering. Scientific realism is

still contested but its critics do not deny its semantic component above; both sides in

the debate now take theoretical language literally as describing an unobservable

world.The resurgence of scientific realism also led to metaphysics coming to be seen as

legitimate by the subgroup of philosophers to whom it was previously most

abhorrent namely philosophers of science. This aspect of the revival of metaphysics

is not so well known. Apart from the aforementioned reasons for no longer

regarding metaphysics as different in principle from high theoretical science, there

were a number of factors that led to metaphysics acquiring a positive valence among

philosophers of science.

1. The surplus content of theoretical terms: As mentioned above it was widely

agreed that to explicate fully the meaning of theoretical terms required morethan systematizing how in a given theory they could be used to relate

observables; it also came to be accepted that the excess content of terms such as

atom was important to theory development and heuristics for example via the

role of analogy and metaphor in model building. This directly led to the large

amount of current work on models in science through the work of early pioneers

such as Black (1962), Hesse (1966) and Suppes (1961).

2. The continuum between high theory and metaphysics: As mentioned above, the

positivist attempt to demarcate meaningful, scientific, and therefore for them

empirical discourse, from meaningless metaphysics floundered in part becauseof the impossibility of making an explicit the observational basis for claims

highly theoretical claims in science. As well terms like field and force,

physics in the twentieth century became progressively more abstract and made

statements about the structure of space and time, and incorporated symmetry

36 J. Ladyman

1 3


7/21

principles and conservation laws. It seems that there is a continuum between

high theory and metaphysics.

3. Analogous to the continuum in the terms of observational science, theory and

metaphysics, Quine influentially argued for a holism about confirmation in the

same article in which he allegedly destroyed the analytic/synthetic distinction(1951). This may suggest that metaphysics is part of the web of belief that gets

tested along with everything else. For a philosopher of science like Lakatos

(1978), metaphysics is part of the hard core of a research programme and that is

like being at the centre of the web of belief. Accordingly, when a research

programme progresses this is tantamount to confirmation for metaphysical

hypotheses, and when one degenerates metaphysical hypotheses are effectively

refuted.

4. Putnam and causal theories of reference: Scientific realism faced a challenge

that was absolutely nothing to do with the positivist demand for testability inthe form of Thomas Kuhns historiography of science. Philosophers of science

became more aware of the history of science in part because of the challenges to

their realism and its axiology of approximate truth, convergence, cummulativity

and progress that came from historians and sociologists of science inspired by

Kuhn. In response to the concern that the meanings of scientific terms such as

atom in different theories are incommensurable, Putnams (1975) influentially

incorporated the causal theory of reference into his version of scientific realism

and generated a wealth of work on reference in philosophy of science, and the

metaphysics of causation, rigid designation and possible worlds became part ofphilosophy of science.

5. The explicit engagement with metaphysical issues in science: Perhaps the most

significant factor that contributed to the recent rise of metaphysics is that the

sophistication of twentieth century science gave rise to intricate issues at the

intersection of science and metaphysics. Probably the most important example

is the debate about the quantum mechanics that led to the Einstein associating

his defence of scientific realism with specific metaphysical claims such as

individuality, determinism, locality and perhaps most importantly of all

possessed values for counterfactual measurement outcomes.10 (This latter

claim fully associates scientific realism with realism about the external world

since it is seen to be analogous to believing in possessed values for everyday

perceptual experiments that we merely could perform and the basis of both is

inference to the best explanation, of which more below.)

In the context of general relativity, the classic metaphysical issue of substan-

tivalism versus relationalism was revived in the context of General Relativity as

Einstein at first sought a Machian theory in which the structure of space and time

would be entirely determined by the distribution of matter, but actually developed a

theory that treats spacetime as a dynamical entity and so seems to grant it physicalstatus. This raises metaphysical issues about determinism and the identity and

10 The EPR paper made explicit these issues that are beautifully explained in Redhead (1987).


1 3


8/21

individuality of spacetime points that have again interacted with the philosophy of

modality.11

In relation to quantum particles issues of identity and individuality have also

came to the fore and the standard view is that the principle of the identity of

indiscernibles was false for quantum particles rendering them non-individuals.12

Questions of identity over time and transworld identity are also discussed.

Furthermore, quantum mechanics was important to philosophical theorising about

causation as the idea that the world might be fundamentally indeterministic

combined with the increasingly widespread use of statistics in the special sciences

promoted the development of theories of probabilistic causation beginning with

Reichenbach (1956).

However, it was not only in physics that important metaphysical issues arose as

science developed. In biology the development of theories of the unobservable

structures of DNA, genes, fitness and function, information, and natural selectionraised many metaphysical issues about essence, identity and individuality, kindhood

and teleology.13 Other areas of new science that are metaphysically rich include the

theories of information and computation, complexity science and cosmology.

All these fascinating subjects for metaphysical enquiry are empty on a narrow

instrumentalist construal of science. Scientific realism is a rich source of

metaphysical inspiration, and it suggests if not requires that metaphysical

components to scientific theorising are important and not to be dismissed as

lacking cognitive content. By 1974 John Watkins gave a presidential address to The

British Society for the Philosophy of Science entitled Metaphysics and theAdvancement of Science, in which he says, I have the impression that it is now

almost universally agreed that metaphysical ideas are important in science as it is

that mathematics is. (p. 91) For all the reasons explained above, philosophers of

science came to believe that metaphysics is part of science and contributes to

scientific progress. However, Watkins warns, this idea is liable to misuse just as

mathematics has been abused in the service of pseudo-scientific work in the social

sciences: The likelihood of such misuse might be reduced by a methodological

account of the proper role of metaphysical ideas in science. (p. 92) The implication

is that Watkins does not want to grant free reign to metaphysical theorising that they

way to avoid doing so is to ensure that the latter is tied to its role in scientific

methodology. For Watkins, as for Zahar, whose most recent book is called Why

Science Needs Metaphysics (2007), metaphysics is rehabilitated not as an

autonomous discipline, but rather only in so far as it contributes to the progress

of science. Not surprisingly what Watkins and Zahar mean by metaphysics is very

different from what is often discussed by contemporary metaphysicians. More on

this below; first consider current metaphysics.

11 See, for example, Maudlin (1990) and Hoefer (1996).12 See French and Krause (2006) and for a dissenting view Saunders (2006).13 See Hull (1967) for an early influential account. The radical metaphysical implications of evolutionary

biology for individuation were noted earlier by in.

38 J. Ladyman

1 3


9/21

3 The content and methodology of analytic metaphysics

The problems with which much work in metaphysics is now concerned have drifted

away from any mooring to science. The paradigmatic subject matter has become the

special composition question, perdurantism versus endurantism, universals versustropes, and the question as to whether physical objects are identical with the matter

of which they are made.14 To these questions we can add those concerning the

existence of mathematical objects and the metaphysics of modality. There is a

widely shared emphasis on truth-makers and a truth-making relation that seems to

have absolutely nothing to do with anything in science. The first debate concerns

whether or not there are composite objects composed of whatever the alleged

ultimate partless parts or simples there are, and if so on what principle governs the

relationship between such wholes and their parts. The literature on this issue is

largely if not completely lacking any engagement with scientific accounts ofcomposition. Instead, the problem is taken to be a completely general and abstract

one, the solution to which is imagined to consist of principles that make no mention

of any scientific theories. Likewise, while the debate about the persistence of

material objects, which is largely about whether their three-dimensional existence

endures, or they are in fact temporally extended entities with temporal as well as

spatial parts, might have been inspired in part by consideration of the implications

of relativity theory, the current arguments about it do not engage closely if at all

with the physics of space and time, nor with any science about how entities manage

to cohere. Similarly, as Maudlin (2007) points out, the problem of universals isaddressed without reference to what science tells us about properties and relations.

Finally, it is hard even to imagine a way in which science could be relevant to the

debate about whether a statue is identical with the lump of clay out of which it is

fashioned.15

There are various metaphysical packages, and sometimes it seems as if the choice

from the metaphysical menu depends only on quasi-aesthetic criteria and taste. Even

metaphysical views that virtually everybody regards as scoring very badly on

accordance with common sense and intuition must be taken seriously if they are not

inconsistent and are accompanied by arguments that are ingenious enough. Positions

in logical space with little to recommend them other than their boldness and novelty

are liable to be entertained if they can avoid the logical snares set for them by the

critics.

As, we have seen, those who embraced metaphysics because of its importance to

science, make room for a kind of remote empirical support for metaphysics, at least

in so far as there are periods in which a particular metaphysics seems to be refuted

14 Of course this is only the first-order subject matter and much work in metaphysics addresses higher-

order questions.15 An anonymous referee objects that high-level ontology is important in high-level science and hence

that the debate about statues and lumps of clay ought not to be dismissed so lightly. The relationship

between the ontologies of the special sciences and fundamental physics is indeed important and

interesting but the metaphysical debate about whether statues co-exist with lumps of clay is liable to shed

no light on the former issue since the latter debate pays no attention to dynamics, regime and scale all of

which are fundamental to accounts of composition in science (see Ladyman and Ross 2007, Chap. 1).


1 3


10/21

because the science that goes with it is rejected. This happened with Descartes

metaphysics of matter without vacuum as his plenum theory entailed that all action

was action by contact, but Newtons postulation of a universal gravitational force

was just too successful for the mysterious action at a distance it seemed to describe

stand in the way of its acceptance. The Cartesian metaphysic was in decline bythe time the electrostatic force was introduced by Coulomb. Other examples are

discussed below, but supposing that we are considering hypotheses that have no

even indirect connection to science, what methodology can we use to choose among

competing theories? Furthermore, in the face of the poor track record of a priori

insight into the nature of reality why suppose there is any such method at all? For

traditional rationalism, the intelligibility of the world to the human mind was

evidence of the beneficence of the creator, but absenting such an appeal to a positive

reason for faith in a priori methods metaphysical seems liable to lack any

naturalistically respectable method or justification.However, there is a common line of defence of metaphysics that is worth

examination. It involves the idea that the metaphysical method should be like the

methods of empirical science only without the connection to testability that

scientists always like to keep in mind. As noted above, even constructing a logically

consistent system of the world with something to say about all its phenomena and

their kinds and how they interrelate is hardly easy. So metaphysics begins with the

task of constructing a coherent theory of the world and integrating it with

philosophical logic. This is no mean feat if it is to be simple enough to be

expressible reasonably briefly, however, of course there may be many such systemsso some way must be found to chose among them. The solution is to undertake a

cost-benefit analysis of the various theories, and to take the costs and benefits to be

closely modelled on those weighed in theoretical science when choices must be

made between theories that are all consistent with the data, but some are argued to

be simpler or more explanatory, and from that is inferred their superiority. Consider,

for example, the standard story about Special Relativity and its superiority to

Lorentzs theory that avoided radical revision of space and time by positing forces

to distort matter in the measurement device so that the motion of the Earth through

the ether would not affect the observed speed of light. Special Relativity is more

parsimonious and less ad hoc and so superior even though empirically equivalent in

accounting for the failure to detect the Earths motion through the ether. Similarly,

while Poincare(1905) argued that the geometry of the world can never be measured

because it is always possible to reproduce the effects of curvature with universal

forces, General Relativity triumphs over such theories because of its overall

simplicity in comparison with them, or at least for some other reason to do with the

pathological and ad hoc nature of universal forces that might be subsumed by

arguing that they are not genuinely explanatory.

This line of argument is reinforced by the recognition that the only way to defeat

sceptical hypotheses about induction, other minds and the external world seems to

be on the grounds of the super-empirical virtues of our normal hypotheses.

However, unlike in science, in metaphysics the cost-benefit analysis often proceeds

with the assumption that giving up common sense or contradicting intuition is a

cost, as is any ontological or ideological commitment. So, for example, the claim

40 J. Ladyman

1 3


11/21

that tables do not strictly speaking exist is a cost because it violates common sense,

the claim that there are other concrete possible worlds is an ontological cost, and the

claim that there is natural necessity is an ideological cost. On the other hand, an

explanation of something is a benefit, and it often seems so too is being intuitive,

natural or similar.Ontology and ideology are necessary to discharge the costs that ensue from

commitments to claims that originate with common sense and everyday experience,

such as that there are whose states are largely independent of our beliefs and desires,

but may also include commitments to laws of nature and menageries of scientific

species. The idea is that our ontology and ideology should explain the ordinary and

scientific facts that we accept, and those we revise. For example, necessitarians such

as Armstrong (1985) and Bird (1998) argue that inference to the best explanation

(IBE) provides a route to metaphysical knowledge of natural necessity because it is

needed to explain the rationality of induction and to underpin the rest of theexplanations we have of the phenomena we observe. As Anjan Chakravartty puts it

the basic methodology of much work in metaphysics is to seek trade-offs involving

ontology and explanation (Chakravartty2007, xiii).

Bas van Fraassen is the latest in a long line of opponents of scientific realism who

are also sharply critical of metaphysics, indeed he is more scathing about the latter

than the former. However, in the case of both the fundamental basis of his critique is

his rejection of the idea that explanatory power has any probative force. Van

Fraassens empiricist critique of explanation and IBE is often dismissed on the

grounds that his view leaves no positive grounds for everyday realism or induction.To many critics agnosticism about unobservable entities seems to be a form of

mere philosophical doubt analogous to scepticism about the existence of the

external world. Hence it is argued that van Fraassen is a selective sceptic who ought

similarly to be a sceptic about all inductive knowledge even of observable objects.

Hence, the explanationist defence of scientific realism and metaphysics according

to which there is a seamless connection between the rationality of everyday

induction, inference to the existence of unobservable entities in science, and

inference to the best explanation in metaphysics. In the next section, the

explanationist defence of metaphysics is considered in more detail.

4 Comparing metaphysics with theoretical science

The purpose of this section is to assess the arguments about scientific realism to

inform a comparison between the appeal to explanatory power in science and in

metaphysics. The most directly relevant argument is the underdetermination

argument. As discussed above, it is often argued that metaphysical hypotheses are

no different from theoretical scientific claims in being underdetermined by the data

in strict empirical terms, and that in both cases explanatory power is what solves the

problem. However, it is questionable whether explanatory power is really as

important in the epistemology of science as the above argument suggests. It is also

not established that explanation in science is the same or similar enough to

explanation in metaphysics. If the explanatory power of scientific theories plays the


1 3


12/21

role it does in theory choice because of the relationship between theoretical

explanation and the empirical virtues of scientific theories, but the alleged

explanatory power of some metaphysical hypotheses is decoupled from any

prospect of empirical success, then we would have no reason to think that kind of

explanatory power is worth pursuing on the basis of the value of pursuing it incommon sense and in science. Accordingly there follows an examination of the role

of explanatory power in scientific theory choice. Sections 4.1and4.2concern the

underdetermination problem, and Sect. 4.3 offers reasons to think that the

importance of explanatory power in science and in explaining its history is limited.

In Sect.4.4theoretical science and metaphysics are briefly compared in respect of

the pessimistic meta-induction and other arguments. Section4.5argues that there is

a difference between the kind of metaphysics that is important to science and its

counterpart in current analytic metaphysics.

4.1 Varieties of underdetermination

The standard form of the argument for the underdetermination of theory by data is

as follows:

(1) There are empirically equivalent theories.

(2) Such theories are evidentially equivalent

Therefore theory choice is underdetermined.

The response to the scientific underdetermination problem that suggests a similardefence of metaphysics is to deny (2) on the grounds that there are superempirical

theoretical virtues such as simplicity, elegance, coherence with background beliefs,

and explanatory power. Hence, it is argued that evidence properly understood does

not reduce to what empirical facts a theory entails, rather how those empirical facts

follow from the theory and how the theory otherwise is are part of the evidential

basis for it. However, it is widely acknowledged that it is difficult to argue that

simplicity or elegance are direct evidence for the truth of a theory so realists have

tended towards the view that all the superempirical virtues are subsumed under

explanatory power so that the solution to the underdetermination problem is IBE.16

The most sustained defence of inference to the best explanation is that of Lipton

(2004). He argues that in cases of everyday inductive reasoning where many

hypotheses could account, for example, for what appear to be bear tracks in the

snow, the hypothesis we prefer is the one that best explains the phenomena. Perhaps

the tracks appeared by chance, perhaps there is a practical joker around, perhaps

there is a new kind of creature with tracks indiscernible from bears, perhaps there

are no tracks at all but some kind of illusion, perhaps, the world came into existence

10 min ago and so on. All these hypotheses are consistent with the phenomena, but

that there is a bear is a simple hypothesis that is unified with background beliefs, and

16 See Psillos (1996) for an account of the explanationist defence of scientific realism that originates with

Richard Boyd and others. The underdetermination argument and realist responses to it are discussed with

great care in Kukla (1998). My own account is in Ladyman (2002, Chap. 6).

42 J. Ladyman

1 3


13/21

with background metaphysical assumptions such as that tracks dont come from

nowhere.

(1) is also contestable depending on what empirical equivalence is taken to be

and depending on to what kind of theories it is taken to apply. The latter distinction

is that between localand globaltheories, in the sense that theories may be putativetheories of everything or merely theories in some specific domain such as optics (see

Hoefer and Rosenberg1994). Almost all the theories that have been proposed in the

history of science have been understood by all to be local in their scope, though of

course a theorys scope may be expanded as when the laws of optics came to be

understood to apply to electromagnetic radiation other than visible light, or when

thermodynamics was applied to the whole universe. Theories may also be

empirically equivalent with respect to all the observations we have carried out up

to now (weak), or with respect to all possible observations (strong). (Clearly the

modality here may be of different strengths too.)The problem of induction is a form of weak (and usually local) underdetermin-

ation. Universal generalisations are underdetermined weakly by the observed

regularities since the latter are also consistent with counter-inductive claims

according to which the future will not resemble the past in respect of the behaviour

of colliding billiard balls or whatever. Inductive projections and counter-inductive

projections are clearly not equivalent with respect to all possible observations, but

only with respect to those we have made so far. In The Scientific Image(1981), van

Fraassen discusses the underdetermination of the different hypotheses about the

state of relative constant motion of the centre of mass of the universe with respect toabsolute space in the context of Newtonian mechanics. This is a case of strong

underdetermination since because the covariance of Newtonian mechanics under

Galilean transformations means that the facts about the matter could make no

possible difference to the outcome of any mechanical experiment. However, critics

argue that van Fraassens disavowal of IBE means he has no solution to the problem

of weak underdetermination either and hence should be a sceptic about all inductive

knowledge.

According to many realists, the difference between the inference to the existence

of unobserved and the existence of the unobservable is of no epistemic significance.

In some cases of the former we may never check directly so the entities in question,

for example, dinosaurs, may just as well be unobservable. We believe in them, just

as we believe in unseen bears, because their existence is the best explanation of the

phenomena, and realists argue belief in electrons is similarly justified.

However, explaining the difference between the reliability of projections of

regularities into the future by appealing to an ontology of everyday and scientific

objects does not licence all forms of explanation by posit. The underdetermination

of metaphysical hypotheses by data is clearly of the strong and global kind.

However, as far as science goes strong, global underdetermination is a problem in

principle not in practice since it is not a problem that has ever been faced: we have

never had a globally empirically adequate theory and it remains an option to adopt

Newton-Smiths (1981) arrogance response, according to which the two theories

that are globally empirically adequate theories are just notational reformulations of

each other, if such a case did arise. This response also does not arise with the kind of


1 3


14/21

underdetermination we face in the case of the bear-tracks just as it does not arise in

science. The hypotheses that we chose among in everyday life and in science are at

most weakly underdetermined by the empirical facts.

There are familiar philosophical examples of global strong underdetermination

by construction. Descartes evil demon hypothesis, or Russells suggestion that theworld came into being 5 min ago being globally empirically equivalent to the

common sense views they rival. However, these hypotheses are entirely parasitic in

their empirical content on the beliefs they are proposed to call into question. In the

context of the underdetermination problem in science such artificiality leads to

doubt about whether global underdetermination should be taken seriously.

4.2 Defeating the underdetermination problem in science

Duhem (1906) was concerned to explain how science solves weak, localunderdetermination. This is a practical problem of theory choice. He invokes

scientific good sense to explain how it is resolved in practice, but he thought that

this is only necessary temporarily because subsequent scientific developments will

find some additional domain of observations with respect to which otherwise weakly

empirically equivalent theories will differ in their predictions (Ivanova2010). Even

strongly empirically equivalent theories may turn out to differ when they are

extended to new domains or conjoined with different background theories and

auxiliary assumptions (Laudan and Leplin1991). Science itself may be what tells us

that theories are empirically equivalent by demarcating the limits of the observable:for example, free quarks cannot be observed according to the quark confinement

model, and colour charge is unobservable in principle for theoretical reasons

according to QCD. Accordingly, science may tell us that what is observable has

changed, as when distant galaxies or blood flow in the brain become observable

because of new technology. Crucially, the genuinely scientific underdetermination

problems like the problems of making inductive generalisations in everyday life arise

for agents situated at particular times facing choices about which particular theories

to develop, and later observations will or at least could show which was correct.

As mentioned above, in response to the underdetermination argument in the

context of science, there is no need to invoke explanation if the first premise is

denied. As well as the considerations just mentioned concerning the definition and

status of empirical equivalence, Laudan and Leplin also argue that, at least, in the

case of strong and global underdetermination, there are no genuine scientific

examples to establish the general claim that (1) makes. The examples we do have are

often not even strong, for example, underdetermination between wave and particle

theories of light which was resolved by the observation of interference effects, and

the underdetermination of Special Relativity, which was only weak because there

was later confirmation of it by the Compton effect and the inter-convertability of

mass and energy, and in any case it became moot when General Relativity issued

quite new empirical predictions. None of these theories were global in any case. The

example of global underdetermination mentioned above namely the underdetermin-

ation of uniform motion with respect to absolute space in Newtonian mechanics, is

best understood as evidence that there is no fact of the matter because surplus

44 J. Ladyman

1 3


15/21

structure is being attributed to spacetime. Hence, Galilean spacetime is the right one

for Newtonian physics in which there is no underdetermination because there is no

absolute space, but still a notion of absolute acceleration (Penrose 2004, chapter 17).

On the other hand, Laudan and Leplin argue that the philosophical arguments that

purport to establish (1) in the global strong case all involve cheap tricks or logico-semantic trickery. For example, one way of establishing (1) so interpreted is to use

the observable/unobservable distinction to define a theory T as the observable

phenomena are as if T but T is not true; this is clearly ad hoc in every sense that

Popper and his followers elaborated.

In the light of all these arguments many philosophers of science are not

persuaded that there really is a global and strong underdetermination problem in

science. Hence, the scientific and metaphysical underdetermination problems are

different, and even in the scientific case it is questionable that inference to the best

explanation is needed. Even IBE is vindicated in overcoming underdetermination,all actual cases in science and common sense are at best strong and local, but not

strong and global as in the case of the kind of metaphysical hypotheses surveyed in

3 above. There is no analogue of the distinction between genuine and artificial

examples in such examples since all the hypotheses in question are the products of

philosophers tricks.

4.3 Explanatory power and scientific theory choice

Philosophers of science such as Alan Musgrave and Ernan McMullin argue thathistorical or diachronic virtues of theories are needed to explain how underdeter-

mination in science is resolved. According to McMullin, Dichronic virtues of

theories include their lack of ad hocness, their fecundity for future development and

extension, and the degree to which they enjoy novel predictive success. According

to, Musgrave (1974), the evidential support a hypothesis receives from observations

depends in general on the temporal relationship between them and is not a purely

logical relation. As mentioned above, it is often claimed that both Special and

General Relativity are to be preferred over their empirically equivalent rivals on

account of their simplicity and explanatory power. However, this is tendentious. The

rival hypotheses of Lorentz and Fitzgerald were not merely ad hoc and had good

theoretical motivations. A detailed investigation of why Einsteins programme

superceded Lorentzs was carried out by Zahar (1973a,b,1988). He found that the

most compelling grounds for preferring relativity theory is that the metaphysical

principles that formed the positive heuristic in the hard core of Einsteins research

programme were ones that had great fecundity for the future development of

science. In particular, special relativity has had two great offshoots. The first is

General Relativity and the second is the relativistic quantum theory of Dirac, and

the programme of Lorentz-covariant quantum field theory. Both of these have

enjoyed significant predictive success of the strongest variety. If explanatory power

of these stories it is as a method for the development of future science.

Another example of a central scientific theory about which it is claimed that it

was chosen on the grounds of its explanatory power is the theory of evolution by

natural selection. However, evolutionary biology has been highly predictive and


1 3


16/21

subsequently confirmed.17 Explanatory power in science is closely connected to

predictive power, and often both explanandum and explanans are very specific and

precise. Such explanatory and unifying power is usually achieved in part by

mathematics. For example, the inverse square law of Newtonian gravitation unified

terrestrial and celestrial mechanics, and Maxwells field theorys highly abstractmathematics unified electricity and magnetism. Furthermore, quantum theory in its

orthodox form was chosen despite its complete lack of an explanatory basis for the

nature of superpositions, wavefunction collapse, and the correlations between

entangled states. (It is worth noting that both Arthur Fine and Bas van Fraassen

argue powerfully that the pursuit of explanations of the latter is completely futile,

and Jeffery Bub argues that there cannot in principle be an explanation of quantum

measurement.)

Even those who value and pursue explanatory power in science, do not consider

it enough. Hence, the scepticism among many contemporary physicists about stringtheory, and the fact that dark matter is not regarded as proven without empirical

success over and above its explanation of the phenomena it was introduced to

explain. It is a lesson of Poppers critique of Marxism and psychoanalysis that

explanatory power is sometimes cheap and does not carry the probative force of

prediction especially novel prediction. In any case, in so far as explanatory power is

supported by its use in science and in everyday life it is coupled to empirical and

practical success because it is never global and strong underdetermination that is

there at issue. In sum, in the case of scientific realism, technological success,

intervention and detection and crucial in motivating the idea that theoretical virtuesin general and explanatory power in particular, are epistemic and not merely

pragmatic. Explanatory power plays the role it does in theory choice because of the

relationship between theoretical explanation and the empirical virtues of scientific

theories. We have inductive grounds for believing that pursuing simplicity and

explanatory power in science will lead to empirical success, but no such grounds

where we are dealing with distinctively metaphysical explanations, since the latter is

completely decoupled from empirical success.

4.4 Arguments from theory change among others

The arguments that bother most scientific realists are based on the history of theory

change in science rather than on abstract and artificial worries about global strong

underdetermination. The pessimistic meta-induction may be severely weakened if

17 An anonymous referee objects that explanationists paradigm cases are from (largely) non-predictive

sciences such as evolutionary biology. It is certainly true that much of evolutionary biology is not

predictive and seeks to reconstruct and explain the history of life on Earth. However, evolutionary

biology is replete with successful predictions. For example: Darwin himself successfully predicted thathuman ancestors arose in Africa based on homologies with African apes; theory successfully predicted

higher mutation rates for organisms in heterogeneous and rapidly changing environments (Oliver et al.

2000); theory predicts that genealogy there should be all manner of transitional forms subsequently

unearthed; theory predicts that homologies in phylogeny will be accompanied by homologies in

genealogy; and finally more prosaically, evolutionary biology predicts there will be no evidence pre-

Cambrian rabbits.

46 J. Ladyman

1 3


17/21

the kind of empirical success insisted upon for realist commitment is novel

predictive success. However, it seems that there are cases of novel predictive

success based on theoretical terms that do not refer, and or on theoretical

explanations that are not correct. The response of scientific realists has been to make

the criteria for epistemic assent to theoretical claims extremely restrictive. Notably,Philip Kitcher proposes an account of selective confirmation according to which

those parts of science that are idle in the production of novel empirical success

enjoy no support from it. A similar account is proposed by Psillos (1996), whose

divide and conquer strategy crucially involves regarding the metaphysical content

of theories such as the caloric theory of heat, according to which heat is composed

of particles or a fluid of some material kind, as dispensable. In general, the bar is set

very high for epistemic commitment in the scientific realism debate. The most

developed and defensible forms of scientific realism are ones that would not endorse

our believing in the metaphysical content of contemporary physics on the basis of itsexplanatory power.

Other disanalogies between scientific realism and metaphysics include that there

is no analogue of entity realism for metaphysical hypotheses concerning universals,

tropes, composition, and the like, while in philosophy of science, entity realism and

the associated inference to the most probable cause as opposed to inference to the

best explanation, are what make the likes of Nancy Cartwright and Ian Hacking

realists. There is also no analogue of the debate concerning whether or not theoretical

entities such as atoms, dark matter, sub-atomic particles and so on are in fact

observable and observed. In respect of theory change again, there is no analogue ofthe way that successful scientific theories that are abandoned are retained as limiting

cases in their successors, nor for the other structural relations between theories that

inspired structural realism. The historical case against the accuracy of a priori

metaphysical speculation is stronger than that against theoretical science.

4.5 Metaphysics versus high science

Recall that one of the reasons for the recent rise of metaphysics is the fact that

metaphysical issues were explicitly discussed in the context of debates in theoretical

science. The marks of this kind of metaphysics are as follows:

(i) extreme generality

(ii) untestable

(iii) continuous with paradigmatically metaphysical subject matter

Some examples were mentioned briefly above. Other important cases in which a

metaphysical doctrine has formed the central heuristic of a successful research

programme in science are as follows:

Atomismthe idea that all matter is composed of tiny particles and that their

dynamical and structural properties might give rise to the apparent properties of

everyday material objects such as their colours and textures was first articulated as

purely metaphysical speculation with no observational implications, but of course it

led to chemical atoms, the kinetic theory of gases and ultimately to the sub-atomic

world.


1 3


18/21

Locality principlesthe idea that all action was action by contact was an

important heuristic in the development of the mechanical philosophy although

Newtons law of universal gravitation violated it. Subsequently the development of

field theory restored locality to the theories of light and electromagnetism and

ultimately to gravity as well. Locality principles can be formulated in both spatialand temporal terms and are now routinely expressed in terms of the spatio-temporal

metric of relativity theory, hence it is said that the causal past of any event must lie

within its backwards lightcone.

The principle of the conservation of energy has its origin in the idea of nothing

being created from nothing and the ancients articulation of the idea of substance. In

its modern form it began with the conservation of kinetic energy in systems without

friction and the conservation of momentum more generally, but became funda-

mental to all science after the experiments on the conversion of heat and work in the

nineteenth century. It is now a unifying principle throughout all science (thoughthere are issues about it in General Relativity).

Physicalismarose out of late nineteenth and early twentieth century controversy

about whether physical processes could account for biological and chemical

phenomena without the need for chemical and vital forces (Papineau 2001). It is

now regarded as fundamental to our understanding of biology that biological

processes are essential chemical and physical processes and that and the basic

taxonomy of atoms embodied in the periodic table of elements is a fundamental

presupposition of all the natural sciences.

The central dogma of molecular biologyis effectively a metaphysical claim sinceit places a condition on all causation and explanation of the traits of organisms

according to which it is not possible for information to be transmitted from acquired

characteristics of the organism to the genetic basis for the inheritance of traits by the

next generation.

Other examples, include the imposition of Lorentz invariance of quantities and

covariance of laws, and the methodological use of other symmetry principles in

quantum field theory both of which have led to the standard model of particle physics.

In all these cases the explanatory power of the hypotheses is coupled to their

fecundity for the development of local theories that are empirically adequate and

crucially predictive. Now consider the debate about special composition, or tropes

versus universals. The purported explanations offered are decoupled from anything

but the most general and common empirical content and bear no relationship to any

research programmes in current science. These disconnections break the continuum

between high theory and metaphysics.

5 Conclusion: in defence of metaphysics

There is a lot more that can be said in defence of a priori metaphysics other than the

explanationist argument criticised above. One may argue that exploring logical

space is valuable intrinsically just as it is in logic and pure mathematics. This may

be so but there are two important disanalogies between metaphysics and its a priori

cousins in the mathematical sciences. Firstly, logicians and mathematicians have

48 J. Ladyman

1 3


19/21

much more claim to be genuinely exploring an objective structure because their

results once established are forever part of the edifice of knowledge because of the

apodictic nature of proof in those subjects. If this is true at all in metaphysics and

not only that part of it that could be called philosophical logic, then it is to a lesser

extent. Secondly, even if logic and pure mathematics are entirely constructed, it isfair to say their structure is usually of far greater intricacy and intellectual beauty

than that of metaphysical theories.

Of course, if metaphysics is regarded as descriptive rather than speculative in the

sense of Strawson, then it is perhaps possible or even necessary to do some of it

a priori. Alyssa Ney (this volume) defends a kind of descriptive metaphysics. Against

this no argument has been given in this paper. The Canberra Plan proposes a kind of

descriptive metaphysics that is not completely a priori because it investigates the

implicit commitments of our beliefs by Ramsifying our theories to distil their

cognitive content. This programme brings us back to the positivists for it was Carnapwho first elaborated a theory of scientific knowledge based on the Ramsey sentence

approach. Worrall (forthcoming) argues that his structural realism does not face the

underdetermination problem because two Ramsey sentences with all the same

empirical consequences cannot be incompatible, hence he adopts the aforementioned

arrogance response by declaring any differences between theories with the same

Ramsey sentence to be merely notational. However, he believes that the cognitive

content of theories is contained in their Ramsey sentences that leave out their

metaphysical content. However, the metaphysical content of theoretical terms can

play a productive role in the progress of science, as, for example, with Worralls ownexample of Fresnel and the hypothesis that the ether is a mechanical solid.

There are also utilitarian defences of metaphysics. For example, it may be argued

that metaphysical debate is worthwhile because it keeps alive empirically dormant

metaphysical issues that may once again become scientifically important for example,

determinism versus indeterminism, individuation and PII, and substantivalism/

relationalism. It is not possible to decide a priori on the basis of their content which

metaphysical hypotheses are scientifically valuable. So this is a possible defence of

novel work in metaphysics too. Godfrey-Smith (2006) argues that metaphysics should

be thought of as generating models of reality at a very abstract level but analogous to

scientific models. It is plausible to argue that this might help science occasionally as

the very abstract models of metaphysics may be applicable to a scientific problem, as

perhaps with functionalism in the metaphysics of mind, and also as with possible

world semantics, for there is no doubt that the debate about modality following the

seminal work of Kripke and Lewis has been of great importance for the theory of

artificial intelligence and computer science. Even if metaphysical models are not

applicable they may indirectly open up ways of thinking that allow scientists to

conceive of new types of theory and so overcome a major impasse to theory

development problem of unconceived alternatives (Stanford2006).

In any case, there are always the metaphysics problems of the special science that

it was argued were an important factor in the revival of metaphysics. It is arguable

whether they should really be called metaphysics since it is the kind that could

always be called theoretical science. Definitions aside, the interesting question is

whether there is another job for metaphysicians other than working at the highly


1 3


20/21

theoretical end of a special science. Metaphysics has often concerned itself with an

underlying unity to the world beyond the phenomena. What matters is how such a

unifying metaphysics is sought and in particular whether it is hermetically sealed

from science or integrated with it. There is a possible task for naturalistic

metaphysicians, not working in the special sciences one by one, but makingconnections between them and understanding their interface and even perhaps how

they all fit together. This is a distinctive creative role for naturalistic metaphysics:

the unification of science and the elaboration of a picture of the world that brings

together what we have learnt from physics and the special sciences. Perhaps this

cant be done, as Nancy Cartwright and John Duprebelieve, but at the moment it is

still arguably a goal worth pursuing given the success of doing so thus far. However,

the challenge of disunitarians is important because conceptual conservatism in

metaphysics is only appropriate when the concepts are at the core of a progressive

research programme and form the framework for the successful development ofeveryday and scientific thought.

Acknowledgments Versions of this paper were presented at the Metaphysics of Science conference in

Melbourne in 2007, the British Society for the Philosophy of Science conference in 2009, and at the

universities of East Anglia and Oxford. I am very grateful to Alexander Bird, George Darby, Steven

French, Leon Horsten, Phyllis McKay Illari, Alyssa Ney, Samir Okasha, Laurie Paul, Barry Loewer,

Damian Veal and anonymous referees for this journal for comments, discussions and criticisms.

References

Armstrong, D. (1985). What is a law of nature?. Cambridge: Cambridge University Press.

Bird, A. (1998). Philosophy of Science. London: Routledge.

Black, M. (1962).Models and metaphors: Studies in language and philosophy. Ithaca: Cornell University

Press.

Callender, C. (2011). Metaphysics and philosophy of science. In S. French & J. Saatsi (Eds.), The

continuum companion to philosophy of science.

Carnap, R. (19361937a). Testability and meaning. Philosophy of Science, 3, 419471.

Carnap, R. (19361937b). Testability and meaning. Philosophy of Science, 4, 140.

Carnap, R. (1939). Foundations of logic and mathematics. Chicago: University of Chicago Press.

Carnap, R. (1956). The methodological character of theoretical concepts. In H. Feigl & M. Scriven (Eds.),

The foundations of science and the concepts of science and psychology (pp. 3876). Minnesota:

University of Minneapolis Press.

Chakravartty, A. (2007). A metaphysics for scientific realism: Knowing the unobservable. Cambridge:

Cambridge University Press.

Chalmers, D., Manley, D., & Wasserman, R. (Eds.). (2009). Metametaphysics. Oxford: Oxford University

Press.

Duhem, P. (1906). The aim and structure of physical theory (Trans. by P. Wiener, 1954). Princeton:

Princeton University Press.

Feigl, H. (1950). Existential hypotheses: Realistic versus phenomenalistic interpretations. Philosophy of

Science, 17, 3562.

French, S., & Krause, D. (2006). Identity in physics: A historical, philosophical and formal analysis.

Oxford: Oxford University Press.

Friedman, M. (1999). Reconsidering logical positivism. Cambridge: Cambridge University Press.Godfrey-Smith, P. (2006). Theories and models in metaphysics. Harvard Review of Philosophy, 14, 419.

Hempel, C. (1963). Implications of Carnaps work for philosophy of science. In P. Schilpp (Ed.),

The philosophy of Rudolf Carnap. LaSalle: Open Court.

Hesse, M. (1966). Models and analogies in science. Oxford: Oxford University Press.

Hoefer, C. (1996). The metaphysics of space-time substantivalism. Journal of Philosophy, 93, 527.

50 J. Ladyman

1 3


21/21

Hoefer, C., & Rosenberg, A. (1994). Empirical equivalence, underdetermination, and systems of the

world.Philosophy of Science, 61, 592607.

Hull, D. (1967). The metaphysics of evolution.The British Journal for the History of Science, 3, 309333.

Ivanova, I. (2010). Pierre Duhems good sense as a guide to theory choice. Studies in the History and

Philosophy of Science, 41(2010), 5864.

Kukla, A. (1998). Studies in scientific realism. Oxford: Oxford University Press.Ladyman, J. (2002). Understanding philosophy of science. London: Routledge.

Ladyman, J. (2004). Discussionempiricism versus metaphysics. Philosophical Studies, 121, 133145.

Ladyman, J., & Ross, D. (2007). Every thing must go: Metaphysics naturalised. Oxford: Oxford

University Press.

Lakatos, I. (1978). The methodology of scientific research programmes: Philosophical papers (Vol. 1).

Cambridge: Cambridge University Press.

Laudan, L., & Leplin, J. (1991). Empirical equivalence and underdetermination. Journal of Philosophy,

88, 269285.

Lipton, P. (2004). Inference to the best explanation (2nd ed.). London: Routledge.

Mach, E. (1893). The science of mechanics: A critical and historical account of its development, La Salle,

IL: Open Court 1960.

Maudlin, T. (1990). Substances and spacetimes: What Aristotle would have said to Einstein. Studies in

History and Philosophy of Science, 21, 531561.

Maudlin, T. (2007). The metaphysics within physics. Oxford: Oxford University Press.

Musgrave, A. (1974). Logical versus historical theories of confirmation. British Journal for the

Philosophy of Science, 25, 123.

Newton-Smith, W. (1981). The rationality of science. London: Routledge.

Oliver, A., Blazquez, J., & Baquero, F. (2000). High frequency of hypermutable Pseudomonas

aeruginosain cystic fibrosis lung infection. Science, 288, 12511253.

Papineau, D. (2001). The rise of physicalism. In C. Gillett & B. Loewer (Eds.), Physicalism and its

discontents. Cambridge: Cambridge University Press.

Penrose, R. (2004). The road to reality. London: Jonathan Cape.

Poincare, H. (1905 [1952]). Science and hypothesis. New York: Dover.Price, H. (2009). Metaphysics after Carnap: The ghost who walks? In D. Chalmers, D. Manley, &

R. Wasserman (Eds.), Metametaphysics. Oxford: Oxford University Press.

Psillos, S. (1996). Scientific realism: How science tracks truth. London: Routledge.

Putnam, H. (1975).Mind, language, and reality: Philosophical PAPERS(Vol. 2). Cambridge: Cambridge

University Press.

Quine, W. V. O. (1951). Two dogmas of empiricism. The Philosophical Review, 60, 2043.

Redhead, M. L. G. (1987). Incompleteness, nonlocality and realism. Oxford: Oxford University Press.

Reichenbach, H. (1956).The direction of time. Berkeley and Los Angeles: University of California Press.

Saunders, S. (2006). Are quantum particles objects? Analysis, 66, 5263.

Sider, T. (2011). Writing the book of the world. Oxford: Oxford University Press.

Stanford, K. (2006).Exceeding our grasp: Science, history and the problem of unconceived alternatives .

Oxford: Oxford University Press.Stanford, K. (2007). Protecting rainforest realism. Metascience, 19, 161185.

Strawson, P. (1959). Individuals. London: Methuen.

Suppes, P. (1961). A comparison of the meaning and use of models in mathematics and the empirical

sciences, in his (1969) Studies in the methodology and foundations of science. Dordrecht: Reidel.

Swoyer, C. (1983). Realism and explanation. Philosophical Inquiry, 5, 1428.

Van Fraassen, B. C. (1981). The scientific image. Oxford: Oxford University Press.

Van Fraassen, B. C. (1985). Empiricism and the philosophy of science. In P. Churchland & C. Hooker

(Eds.), Images of science (pp. 245308). Chicago: Chicago University Press.

Van Fraassen, B. C. (1989). Laws and symmetry. Oxford: Oxford University Press.

Van Fraassen, B. C. (2002). The empirical stance. New Haven: Yale University Press.

Worrall, J. (forthcoming). Underdetermination, realism and empirical equivalence.Synthese

.Zahar, E. (1973a). Why did Einsteins programme supersede Lorentzs? (I). British Journal for the


Zahar, E. (1973b). Why did Einsteins programme supersede Lorentzs? (II). British Journal for the


Zahar, E. (1988). Einsteins revolution: A study in heuristic. New York: Open Court.

Zahar, E. (2007).Why science needs metaphysics: A plea for structural realism. New York: Open Court.


1

Documents

Ladyman - Science, metaphysics and method.pdf