Van Inwagen’s modal skepticism

Van Inwagen’s modal skepticism

Peter Hawke

Published online: 16 February 2010

� Springer Science+Business Media B.V. 2010

Abstract In this paper, the author defends Peter van Inwagen’s modal skepticism.

Van Inwagen accepts that we have much basic, everyday modal knowledge, but

denies that we have the capacity to justify philosophically interesting modal claims

that are far removed from this basic knowledge. The author also defends the

argument by means of which van Inwagen supports his modal skepticism, offering a

rebuttal to an objection along the lines of that proposed by Geirrson. Van Inwagen

argues that Stephen Yablo’s recent and influential account of the relationship

between conceivability and possibility supports his skeptical claims. The author’s

defence involves a creative interpretation and development of Yablo’s account,

which results in a recursive account of modal epistemology, what the author calls

the ‘‘safe explanation’’ theory of modal epistemology.

Keywords Skepticism � Modality � Modal epistemology � Yablo �van Inwagen

I

A possibility-claim is a claim of the form ‘‘it is possible that p’’ (sometimes

abbreviated as �p). In his 1998 paper ‘‘Modal Epistemology’’ (Van Inwagen 1998),

Peter van Inwagen expresses doubt that many of the ambitious possibility-claims

frequently held by philosophers can adequately be justified, although he is happy to

regard our normal, everyday possibility-claims as justified. That is, he defends a

brand of modal skepticism that is not aimed at everyday, or even scientific, modal

claims, but rather at the ‘‘maverick’’ pronouncements of the philosopher. My interest

P. Hawke (&)

Department of Philosophy, University of the Witwatersrand,

Private Bag 3, WITS, Johannesburg 2050, South Africa

e-mail: [email protected]

123

Philos Stud (2011) 153:351–364

DOI 10.1007/s11098-010-9520-5

in his paper stems chiefly from its final section, where van Inwagen argues that his

modal skepticism is supported by a recent and compelling account of the basis of our

modal knowledge due to Stephen Yablo (Van Inwagen 1998, pp. 76–81).

The aim of the present paper is to provide an adequate defence of van Inwagen’s

skeptical conclusion, in the face of a potentially serious objection to his skeptical

argument (an objection I outline in detail). In order to achieve this defence, I will do

three things. First, I will expand and elaborate upon Yablo’s somewhat vague

account of modal epistemology, in the process developing a more precise account

which I call the ‘‘safe explanation’’ theory of modal epistemology. This theory of

modal epistemology has two interesting features: for one thing, it is foundationalist

in nature, implying that the justification of possibility-claims is recursive and, for

another, it grounds the justification of possibility-claims by way of a combination of

conceivability and empirical methods (essentially, an a posteriori justification).

Secondly, I will revise van Inwagen’s argument, taking into account the insights of

the ‘‘safe explanation’’ theory, and show that this revision is immune to the

objection I lay out. Finally, I will conclude by arguing that we are not justified in

asserting the possibility of ghosts or ghouls (zombies), two typical possibility-

claims advanced by philosophers of mind.

II

Van Inwagen sets out to expose a philosophical culture that he sees as having grown

accustomed to accepting far-out possibility claims on the basis of mere intuition. He

concedes that there is a class of everyday modal statements, what he calls basicmodal statements, that our modal intuitions, honed by experience, can claim a

certain authority over. However, for more ‘‘remote’’ modal statements, those far

removed from actuality, he claims that our intuitions cease to be of much use and

more rigorous techniques are required, and rarely applied. With this in mind, here is

a formulation of van Inwagen’s skeptical argument:

P1. For any proposition p, one is justified in asserting the possibility of p only

ifsomeone has imagined a world that the imaginer takes to verify that p

(Yablo).

P2. For any proposition p, someone has imagined a world that the imaginer takes

to verify p only if someone has imagined a world in a sufficient amount of

detail, relevant to p, so as to rule out the compatibility of the specified details

of that world with :p:P3. If proposition p belongs to the class FP (where a proposition is a member of

this class iff it is a philosophical claim1 that is far-removed from everyday

experience), then no-one has imagined a world in a sufficient amount of detail

relevant to p.

C. So, if proposition p belongs to class FP, then one is not justified in asserting

the possibility of proposition p.

1 A ‘‘philosophical claim’’ I take to mean, roughly, an incredibly general claim concerning fundamental

concepts. Perhaps ‘‘metaphysical’’ claim would serve equally well here.

352 P. Hawke

123

Let us consider the above premises in turn. Van Inwagen defends P1 and P2 by

appealing to Yablo’s influential account of ‘‘philosophical conceivability’’ (Van

Inwagen 1998, pp. 76–81). I will call the claim that conceivability acts as a reliable

(if fallible) guide to possibility the c–p principle. Yablo argues that we have no

reason to doubt the c–p principle so long as it is understood that by ‘‘conceivable’’

we mean ‘‘philosophically conceivable’’ (what I will sometimes calls ‘‘Yablo-

conceivable’’) (Yablo 1993), a notion of conceivability which is cashed out as

follows:

p is [philosophically] conceivable for me if[f] I can imagine a world that I take

to verify p (Yablo 1993, p. 29).

Yablo argues that when it comes to judging possibility, we need only concern

ourselves with types of conceivability that advertise themselves as guides of

possibility—that is, that provide the appearance that p is possible. For a type of

conceivability to involve the appearance of possibility is for the state this

conceivability produces to be one such that (i) this state is veridical only if possibly

p (that is, my conceiving that p is true only if it is possible that p), and (ii) being in

this state moves one to believe that possibly p. Yablo claims that his ‘‘philosophical

conceivability’’ fulfills these requirements: the stipulation that the imagined world

appear to ‘‘verify’’ p is tantamount to ensuring that the state be veridical only if a

possible world exists of which p is true. Yablo suggests that an imagined scenario is

only a verification of p if that scenario is not compatible with the falsity of p. In

other words, the veridicality of the imagined scene must be such that it can only be

fulfilled by p being the case. Yablo gives the example of imagining a super-

computer print out a number n which is hailed as a counter-example to the Goldbach

conjecture (Yablo 1993, p. 31). This imagining does not suffice as an example of

Yablo’s philosophical conceivability, and hence acts as no evidence in favour of the

possibility of the Goldbach conjecture being false, since this scenario is compatible

with the falsehood of the claim that there exists a counter-example to the Goldbach

conjecture (the scenario can be filled out in the following way: no counter-example

exists; the super-computer is flawed and expert observers have simply been too lazy

to check its result).

Now, let us consider P3 in Van Inwagen’s argument. In its defence, Van Inwagen

takes a sample of two relatively conservative ‘‘far-out’’ possibility-claims—namely,

that a naturally purple cow is possible and that transparent iron is possible—and

argues that, in these two cases, Yablo-conceivability fails.

It would seem that we can imagine a naturally purple cow quite easily. Just

picture one bounding around a meadow. However, this kind of imagining does not

suffice for Yablo-conceivability, van Inwagen says (Van Inwagen 1998, p. 78): such

a scene is compatible with the falsity of the proposition that there is a naturally

purple cow—the cow in our imagined field might just have been dyed. In order to

imagine a situation that verifies the existence of a naturally purple cow, van Inwagen

suggests, we would need to imagine a chemically possible purple pigment, such that

this pigment is produced and properly placed in the cow’s coat by structures that

exist in some DNA that counts as cow DNA. Van Inwagen is doubtful that anyone

has, or can, perform this imaginative exercise. Likewise, van Inwagen thinks it isn’t

Van Inwagen’s modal skepticism 353

123

enough to, for instance, imagine a scientific conference in which leading scientists,

to great acclaim, announce the existence of transparent iron to an applauding

audience, while holding up a chunk of what looks like transparent glass (Van

Inwagen 1998, p. 79). This imagined situation is compatible with, for instance, the

background scenario that transparent iron does not exist, but the scientific

community has somehow been deceived into thinking that it does. In order to

verify the imagined existence of transparent iron, again an insight into the micro-

structure of such a substance is required. And again, there is no evidence that

anyone has, in fact, attempted this kind of detail when imagining this scenario.

Van Inwagen is right to pursue an inductive defense of P3. The class FP will

probably be difficult to define precisely and it certainly isn’t feasible to attempt to

run through every proposition that falls in this class. Demonstrating the truth of P3

for a small, yet representative, sample of FP-type propositions will be more than

adequate. Unlike van Inwagen, however, it strikes me as important to pick

genuinely significant philosophical propositions as our sample class. In this paper, I

will consider P3 as adequately defended if it can be shown that it is not possible to

imagine a world in the requisite detail in which either a ghost or a ghoul exists.

A ghost is an immaterial, non-corporeal, conscious being, generally depicted as a

mind/soul separated from its body. A ghoul I will take to be equivalent to a zombie:

a being whose body operates, but has no mind and thus no conscious states (in

fiction, they tend to distinguish ghouls from persons by way of their—murderous—

behaviour, but this is hardly necessary: the ghoul’s behaviour may match an

ordinary person’s). Now, consider the following two possibility-claims:

1. It is possible that ghosts exist.

2. It is possible that ghouls exist.

Some philosophers have taken these claims very seriously, on the basis of

possibility-conceivability arguments. Claim 1 is crucial to the argument for

cartesian dualism. Claim 2 forms the basis for a more recent argument in favour of

dualism (Stalnaker 2002): if it is possible that our bodies operate precisely as they

do and yet no minds are ‘‘housed’’ in our bodies, then it appears that mind and body

cannot be identified.

Are ghosts and ghouls Yablo-conceivable? I will delay answering this question in

detail until later in the paper. For now, it is worth noticing that, for all that has been

said so far, it isn’t obvious that they are.

III

I now consider a potentially serious objection to van Inwagen’s skeptical argument.

The objection goes like this: either P2 is true, or it isn’t. That is, either it is necessary

to imagine a situation that is incompatible with (and therefore rules out) :p in order

to Yablo-conceive of p, or this condition isn’t necessary for Yablo-conceivability.

Now, if P2 is true, then this means that Yablo-conceivability is an extremely

demanding imaginative exercise (as pointed out by Geirrson in a paper discussing

van Inwagen’s scepticism (Geirsson 2005)). So demanding, in fact, that it seems

354 P. Hawke

123

completely impractical to Yablo-conceive of propositions that we take, as part of

daily life, to be obviously possible, since for (at least many of) such propositions it

is extremely difficult to imagine a situation that is genuinely incompatible with the

falsity of that proposition. This, however, casts serious doubt on P1 in the skeptical

argument, as the apparently excessive demands of Yablo-conceivability makes it

extremely unlikely that this kind of conceivability is a necessary condition for being

justified in asserting possibility-claims, given that there must be many ordinary

possibility-claims that we are justified in asserting despite our not being able to

Yablo-conceive of the propositions in question. On the other hand, if P2 isn’t true,

then presumably we are free to understand Yablo-conceivability as making less

strenuous demands in terms of what is required of one’s imagination when Yablo-

conceiving of p. But in this case, it is difficult to see why P3 is true. After all, it

seems that we are able to imagine ghosts, ghouls and other fantastical beings and

events in a fair amount of detail (as the proliferation of fantasy and horror movies

and books professes). In summary then: either P2 is true or it isn’t. If it is, then P1 is

likely to be false. If it isn’t, then P3 is likely to be false. Either way, key premises in

the argument will have to be rejected.

In order for this objection to be taken seriously, more needs to be said to justify

the claim that the truth of P2 renders Yablo-conceivability excessively demanding.

That is, some reason needs to be given to believe that there is a mundane proposition

p that is obviously possible and yet it is not easy to see how one is meant to imagine

a situation in which the denial of p is ruled as incompatible with that situation. To

this end, consider the mundane proposition ‘‘I prepare myself a cup of tea in the

kitchen, in my favourite tea mug’’ and the accompanying possibility-claim ‘‘it is

possible that I prepare myself a cup of tea in the kitchen, in my favourite tea mug’’. I

take it that this is the kind of possibility-claim that is obviously true (although the

proposition is, as it happens, false). It also seems that I am able, with ease, to picture

myself carrying out this exercise: I walk to the kitchen, get out the mug and prepare

the tea. Nevertheless, on reflection it is clear that I have not imagined this situation

in the amount of detail required by P2: nothing about what I have pictured is

incompatible with, for instance, the scenario that the mug in my imagining is not my

favourite mug—but rather another mug that has precisely the same observable

properties as my favourite mug. For all I have imagined, the background to my

imagining might be that my favourite mug never entered my possession, perhaps

having been carelessly dropped and smashed before ever being put up for sale (and I

rather purchased an identical mug on the day that I—in the actual world—purchased

my favourite mug).

What sort of imagining would it take to rule out this potential background

scenario? This is difficult to say, particularly since I have no idea of the essential

properties (if any) of my favourite mug (such as the origin of the material that was

used to make this particular mug) that would help to distinguish it from a double in

an imagined scenario. Of course, I could simply stipulate that I am imagining my

favourite mug, not an identical mug, but this kind of stipulation appears to miss the

point of the verification so central to Yablo-conceivability (one could also stipulatethat the super-computer has printed out a genuine counter-example).


123

IV

I now aim to develop and motivate a reading of Yablo’s conceivability that will (a)

expand the insights provided by this version of conceivability into something more

precise and (b) give us a surer footing in evaluating the objection to van Inwagen’s

argument. In the process, I hope to take Yablo’s ideas further than what is offered in

his original discussion.

Yablo’s account will, firstly, benefit from being phrased in a way that is as free of

modal concepts as possible. Here is a revision:

p has been philosophically conceived by someone iff someone has imagined a

world that the imaginer takes to verify p.

Now, Yablo is essentially asking us to do two things: (i) imagine a world and (ii)

verify that proposition p holds in that world. Each instruction is somewhat

ambiguous. Let us consider them in turn.

What is it to imagine a world? Yablo clearly intends for one to imagine a possibleworld. We need to be careful here, however.

I will assume in this paper that there is a contrast between fictional worlds and

possible worlds. I will also assume three common ways of drawing the distinction

between the two. One seemingly obvious distinction between the two is that

fictional worlds are constructions—a fictional world is created by someone, called

its author—while possible worlds are not constructed—they exist entirely

independently of human thought and activities. In this sense, fictional worlds are

mind-dependent while possible worlds are not. Had there not been any minds, there

would not have been any fictional worlds. The same cannot be said for possible

worlds. Another distinction between the two is that fictional worlds are not completewhile possible worlds are. Roughly, this idea of completeness points to the fact that

fictional worlds are, of practical necessity, under-described: there are certain

propositions that are neither true nor false for the fictional world, since the author of

the world did not bother to outline features of the world that would entail the truth or

falsity of those propositions (surely there is no fact of the matter as to how many

hairs there are on Captain Ahab’s head). On the other hand, possible worlds are fully

described—for any proposition and possible world, that proposition is either a true

description or a false description of that world. Thirdly and finally, it seems that

fictional worlds can be inconsistent (consider stories in which there are ‘‘holes in the

plot’’). This seems a direct result of the fact that they are constructed. On the other

hand, it is generally taken to be true that possible worlds cannot be inconsistent. Put

more technically, it is possible that in the set of propositions associated with a

fictional world (taken as true in that world), that there is some proposition where

both it and its negation are in the set. On the other hand, it is not possible that both a

proposition and its negation occur in the set of propositions associated with a

possible world.

With all this in mind, consider again Yablo’s instruction to imagine a world.

Surely, by imagining a world, one is actually constructing a fictional world. If Yablo

intends for us to imagine a possible world, then we need to have some method for

constructing fictional worlds that at least increases the probability that that fictional

356 P. Hawke

123

world approximates a possible world (where ‘‘approximate’’ here means that the

propositions associated with the fictional world are a subset of the propositions

associated with some possible world). Some philosophers (Sosa 2000) appear to

have doubts that we are ever justified in claiming that a fictional world approximates

a possible world. However, it seems to me that there is a way of constructing a

fictional world that, no evidence to the contrary, increases the probability that this

world approximates a possible one. First, one ought to construct the world so as to

be consistent; second, one ought to make the world as detailed as possible; third, one

ought to create the world so that none of the stipulated facts of the world aremodally controversial. This last requirement might strike one as suspicious—if we

are constructing fictional worlds in order to attain knowledge of possibility, how can

we always assume that we are already in possession of uncontroversial possibility-

claims? I will return to this matter. For now, let us assume that there is always some

store of uncontroversial possibility-claims that we can use in the construction of a

fictional world.

In total, Yablo’s definition of conceivability ought to read:

p has been [philosophically] conceived by someone if[f] someone has

imagined a consistent fictional world, which corresponds to a possible world

(as a partial representation of that possible world), that is taken by the

imaginer to verify p.

Now that we are agreed that Yablo is instructing one to imagine a fictional world

(which approximates a possible world), his further instruction to imagine the world

so that it can be verified that p holds for that world becomes problematic. For one

thing, it is potentially trivial to construct a fictional world so that some proposition

holds for that world—but trivial in a very problematic way. Since one constructs a

fictional world, one seemingly is able to stipulate that proposition p holds for that

world. This surely counts as a trivial verification that p holds for the world. Indeed,

it is not clear if there are any useful limits as to what can be stipulated to hold for a

fictional world: clearly, inconsistencies can be stipulated to hold for fictional worlds

(‘‘holes in the plot’’), as can other propositions whose metaphysical possibility is

dubious (for instance, that George W. Bush is a Soviet robot). In other words, it is

extremely easy to stipulate that modally controversial propositions hold for a

fictional world and, as we noted earlier, constructing the world so that one stipulates

the truth of modally controversial propositions in that world weakens the case that

that fictional world is an approximation of a possible world. Now, since we

presumably use conceivability techniques to test the status of modally controversial

claims, it seems that verifying the truth of such a claim in a fictional world by mere

stipulation will only be counter-productive.

In what sense can we ‘‘verify’’ that a proposition holds for a constructed world, if

we are barred from stipulating that p holds? It seems that what we require is that the

proposition not itself be stipulated, but rather that it is a consequence of otherfeatures of the fictional world(where these features may themselves have been

stipulated) and that we come to realize this fact. What is more, in order to avoid

stipulating modally controversial features of the fictional world, it is a requirement

that these other features—from which the truth of p ‘‘follows’’—are not themselves


123

modally controversial features. Hence, it seems that what ‘‘verification’’ means in

this context is to come to realize that modally controversial proposition p results

from (possibly stipulated) modally uncontroversial features of the world.

Several options present themselves as to what this amounts to:

Option 1: To verify that proposition p holds in a fictional world is to realize that

the modally uncontroversial, fictionally true propositions p1, p2, …, pn

provide evidence for proposition p.


the features described by modally uncontroversial, fictionally true

propositions p1, p2, …, pn are sufficent for causing the features p

describes.

Option 3 To verify that proposition p holds in a fictional world is to realize that


logically imply that p.



can be used as a basis for explaining that p is fictionally true.

The most plausible of these suggestions is, in my opinion, option 3. The other

options are either problematic, or seem to collapse into option 3. Consider option 1.

If by ‘‘evidence’’ we here mean ‘‘conclusive evidence’’ then surely this amounts to

option 3. On the other hand, if by evidence we mean ‘‘inconclusive evidence’’, then

this way of taking ‘‘verification’’ will not do in the context of imagining fictional

worlds. Imagined evidence is simply too cheap: I can imagine plenty of evidence

that Clarabell is a naturally purple cow: Clarabell’s purple tone never fades;

Clarabell was born of two equally purple cows; and so on. None of this seems the

slightest bit persuasive however, with regards to convincing one of the possibility of

a naturally purple cow. This lack of persuasiveness is no doubt due to the fact that

imagined evidence is made up, and it seems crucial to the persuasiveness of

inconclusive evidence that it be discovered.

As for option 2, this seems to again collapse into option 3. If p1, p2, …, pn are

sufficient for causing p, then surely the propositions ‘‘p1, p2, …, pn causes p’’ and

‘‘if p1, p2, …, pn hold and p1, p2, …, pn are sufficient for causing p, then p’’ are

fictionally true and logically imply that p is fictionally true. What is more, it is

preferable to use option 3, as it draws out the fact that the causal propositions partly

sufficient to imply p need themselves to be modally uncontroversial. As for option

4, it seems to me that if p1, p2, …, pn can serve as a basis for explaining that p is

fictionally true, then either this option collapses into option 2 (and then into option

3) or straight into option 3, depending on what one means by an ‘‘explanation’’. If

there are subtler interpretations of the notion of ‘‘explanation’’ that might be of use

in this context, it is not clear to me what they are (and certainly will take us too far

astray into debates concerning the nature of explanation).

Putting these thoughts together, here then is a revisionary and clarificatory

account of philosophical conceivability that emerges when taking a hard look at

Yablo’s conceivability:

358 P. Hawke

123

p is philosophically conceived by person x iff that person is familiar with a

consistent, reasonably detailed fictional world (possibly constructed by person

x) of which some set of propositions p1, p2, …, pn are fictionally true, such

that (i) p is a logical implication of p1, p2, …, pn, (ii) p1, p2, …, pn are all less

modally controversial than p and (iii) person x knows that (i) and (ii) hold.

Notice that this account means that if p has been philosophically conceived by

me, then I have imagined a world in a sufficient amount of detail, relevant to p, so as

to rule out the compatibility of the specified details of that world with :p. Hence,

this account supports the second premise, P2, of van Inwagen’s argument. It

explicitly denies, however, that this necessary condition is also sufficient.

Let me introduce some handy terminology. If both (i) p is a logical implication of

p1, p2, …, pn and (ii) p1, p2, …, pn are all less modally controversial than p, then I

will refer to p1, p2, …, pn as jointly constituting a ‘‘modally safe’’ explanation of p. I

think the use of the word ‘‘explanation’’ is justified in this context. A serviceable

account as to what an explanation is that it is some set of statements such that the

thing to be explained can be logically inferred from those statements and the

acceptance of those statements removes the problematic or surprising character of

the thing to be explained. These conditions are, respectively, fulfilled by (i) and (ii).

While I think the use of ‘‘explanation’’ is under the circumstances natural, I will not

attempt to find a place for ‘‘modally safe’’ explanations within any detailed theory

of (scientific) explanation.

Notice that the above account of modal epistemology leaves room for varying

degrees of justification, and offers some insight as to why modal error can occur. A

fictional world might fulfill all the given criteria and yet still fail to match a possible

world, for reasons that are not immediately transparent to author of that world.

Sometimes filling out further details of the world, or reflecting on the implications

of already stipulated details, can reveal an inconsistency, for instance.

V

There is a nagging worry about the ‘‘safe explanation’’ theory: the account calls for

justification of possibility-claims in terms of other, already justified possibility-

claims. As it stands, this could either lead one in a circle or upon a path of endless

justification.

In order for an infinite regress to be halted, justification needs to halt somewhere.

Thus, there need to be propositions whose possibility does not need to be justified by

imagining a consistent fictional world whose details provide a ‘‘modally safe’’

explanation for that proposition. Hence, there need to be ‘‘foundational’’ uncon-

troversial possibility-claims for the ‘‘safe explanation’’ theory to work. If such

foundational possibility claims are explicitly acknowledged, then the fear of a

vicious circularity is abated. The purpose of providing a ‘‘modally safe’’ explanation

would be to link the non-foundational claims to the bedrock of foundational claims.

All that remains is to give some account as to what these foundational claims are,

and what justifies them. First, notice that the idea of two such distinct categories of


123

possibility-claims is hardly an unfamiliar one and is suggested by van Inwagen. I

think we tend to see a difference between our everyday modal claims, the kinds of

claims that allow for normal living and that we generally assert and believe without

question, and the more ambitious modal claims brought to our attention by the arts

(particularly in genres such as science fiction or fantasy), philosophy or science. To

use van Inwagen’s terminology, we may call the former ‘‘basic’’ modal claims

(examples being: ‘‘it is possible that my living room furniture be rearranged’’, ‘‘it is

possible that I got out of bed late this morning’’, ‘‘it is possible that this window

pane should shatter’’), while the latter are ‘‘non-basic’’ (‘‘it is possible that God

exists’’, ‘‘it is possible that alien life exists on another planet’’, ‘‘it could have been

the case that the ice age didn’t occur and dinosaurs still roam the earth’’). What is

the intuitive distinction being drawn here? In the first place, basic modal claims are

somewhat sacrosanct (somewhat unsurprising considering how important they are to

normal life): for instance, a theory of modal epistemology or modal metaphysics is

likely to be viewed with suspicion if it suggests that we are not justified in believing

basic modal claims. Non-basic modal claims, on the other hand, are, apparently,

open to revision, and overly confident affirmation of the truth of one is liable to be

met with suspicion. In the second place, an intuitive distinction between the two

types of claim is their ‘‘distance’’ from actual experience. Basic possibility-claims

are undoubtedly closer to our everyday experience of the actual world. A world in

which I got out of bed late this morning is hardly very different to the actual world.

What is more, things like that in fact happen quite frequently. Non-basic claims, on

the other hand, are typically far-removed from the actual state of things.

We thus have an intuitive idea of a distinction between possibility-claims that are

basic and non-basic, foundational and non-foundational, uncontroversial and

controversial. Can a more in-depth account be given for our access to basic modal

claims? I suggest the following: given both our penchant to cite experience as the basis

for our basic modal claims, and the undeniable ‘‘closeness’’ of basic modal claims to

the actual world, it may prove fruitful to consider if experience of the actual world does

indeed back up such claims. I make the following suggestion: basic modal claims are

those that are justified by the actuality principle and the similarity principle.

The actuality principle is as follows: if p is actual, then p is possible. This

principle enjoys little controversy. As for the similarity principle, this requires a

little introduction. Although this principle is seldom given a name, I think it is in

wide use. Consider the following common answers to questions like ‘‘how do you

know that this drinking glass could break if dropped?’’ or ‘‘how do you know that

you could leave this room by the door?’’ After a moment of puzzlement, a layman is

likely to respond ‘‘well, because things like that happen all the time’’, or perhaps,

simply, ‘‘experience’’. That is, it seems that we often justify our modal beliefs

(chiefly our most basic ones) according to our experience of the actual world. Yet

this is not an application of the actuality principle. What is going on here? My best

guess is that something like the following principle is being applied:

If two things (situations, objects) are similar in some respects, then the

possibilities (relevant to the similarities) concerning those things are likely to

be the same.

360 P. Hawke

123

This principle can be utilized thus: in the past, I have left this room by the door.

Thus, under those circumstances, it was possible for me to leave the room by the

door (by the actuality principle). Now, the current situation in which I find myself is

similar in many respects to the one which occurred in the past (the door is not

locked, my body is functioning normally, and so on). Therefore, by the similarity

principle (in conjunction with the actuality principle), it is likely to be possible,

under the current circumstances, that I leave the room by the door, since this

possibility is relevant to aspects which the two situations share.

The actuality principle is basically beyond reproach and is clearly a fruitful

source of modal knowledge: what better way to convince someone that a certain

state of affairs, or kind of object, is possible, than to actually produce the state of

affairs or object? Nevertheless, the actuality principle, in isolation, is obviously not

satisfactory for accounting for all our basic modal knowledge. I know I could go for

a cup of tea right now; but, I don’t; so the actuality principle cannot ground this

exceedingly trivial possibility-claim. Hence I suggest we also bring the similarity

principle into the picture. The similarity principle, if used in conjunction with the

actuality principle, justifies (basic) possibilities that are non-actual.

The similarity principle is not meant to be foolproof. It can sometimes lead us

astray. Chiefly, this can occur because we are not aware of a relevant difference

between two situations or objects that are being compared (one might conclude that

it is possible that a certain man could fall victim to a heart-attack, since one knows

of no relevant differences between him and the many people who have fallen victim

to a heart-attack. However, it turns out that this fellow is a fitness and health fanatic,

with a history of strong hearts in his family). What counts as a ‘‘relevant’’ similarity

when it comes to making judgements of possibility? It would seem that a similarity

is relevant to the possibility of p if that similarity stands in some kind of causal or

determining relation to the advent of the states of affairs that make p true. All this

suggests that the similarity principle has much in common with analogical

arguments. It also suggests the role that we take causal factors to play in making

possibility judgements (and makes one wonder about the relation between causal

determinism/indetermism and metaphysical possibility). Indeed, much more needs

to be said to properly explore and evaluate the similarity principle. I don’t have the

space for this here but I offer the following inductive argument to justify confidence

in the similarity principle.

Thanks to the actuality principle, there is a conclusive test which confirms the

possibility of a proposition. To confirm that p is possible, make it so that p is actual.

For instance, the similarity principle concludes that it is possible that I can leave the

room right now. I can test this by actually leaving the room. Examples like this are,

clearly, innumerable and the similarity principle, I am sure it is agreed, tends to fair

very well in the face of such tests. What this amounts to is that significant evidence

exists for the truth of the similarity principle. Thus, one may conclude inductively

that the similarity principle is true.

Let us now give a more precise definition for a ‘‘basic’’ possibility-claim: a basicpossibility-claim is one that is (directly) justified by either the actuality principle, or

the similarity principle in conjunction with the actuality principle. A non-basic


123

possibility-claim is a possibility-claim that is not basic. Thus, we have the following

as a final formulation of the ‘‘safe explanation’’ theory:

Person x is justified in asserting the non-basic possibility claim ‘‘it is possible

that p’’ iff that person is familiar with a consistent, reasonably detailed

fictional world (possibly constructed by person x) of which some set of

propositions p1, p2, …, pn are fictionally true, such that (i) p is a logical

implication of p1, p2, …, pn, (ii) either �pk (1 B k B n) is a basic possibility-

claim, or �pk is a non-basic possibility-claim that has been justified to some

acceptable degree and (iii) person x knows that (i) and (ii) hold.

Thus, we end this section with a recursive account as to how one can be justified

in asserting possibility-claims. The kind of conceivability touted here I will continue

to refer to as ‘‘Yablo-conceivability’’, in order to emphasize its roots.

VI

With a more sophisticated take on Yablo-conceivability at hand, let us now re-

evaluate the objection to van Inwagen’s skeptical argument. Recall the basic

structure of the objection: either P2 is true or it isn’t. If P2 is true, then Yablo-

conceivability is too demanding and so P1 can’t be true. If P2 isn’t true, then P3

must also be rejected, along with P2.

I noted earlier that my expanded and elaborated account of Yablo-conceivability

appears to support the truth of P2. Are we then forced to reject P1? Indeed, I think

that the rejection of P1 is a consequence of the safe explanation theory. This is

hardly surprising, however: the safe explanation theory explicitly states that Yablo-

conceivability is not necessary for justifying all possibility-claims, since the entire

class of basic possibility-claims need not be justified in this manner.

Given that I apparently agree with the objection, where does this leave van

Inwagen’s skepticism? I propose that the main thrust of van Inwagen’s argument is

not compromised by this objection since a simple revision of the argument, one

which follows naturally from the ‘‘safe explanation’’ theory of modal epistemology,

will suffice to shield the argument from the objection. Since the conclusion of van

Inwagen’s argument is clearly directed at non-basic modal claims, why not simply

rephrase the entire argument in terms of non-basic modal claims? This gels well

with the ‘‘safe explanation’’ theory, as this theory is committed to the following

proposition:

One is justified in asserting the truth of a non-basic modal proposition ‘‘it is

possible that p’’ only if someone has imagined a world the imaginer takes to

verify the proposition p.

Now, it does not seem likely that the truth of P2 can be taken as a good reason to

deny the above proposition (in contrast to P1). Taking ‘‘non-basic’’ to refer,

intuitively, to those modal propositions that are far-removed from actual experience,

it is far from clear what other means we might have for assessing the truth of these

possibility-claims, other than imagination-centered techniques that have been

362 P. Hawke

123

carefully formulated to avoiding the weaknesses of ‘‘mere’’ imagining. Experience-

based techniques, such as those formed around the actuality and similarity principle,

are (by themselves) obviously inadequate.

Hence, van Inwagen’s skepticism is placed in a defensible position, once we have

revised his argument in the following way:

P*1. One is justified in asserting the truth of the non-basic possibility-claim ‘‘it is

possible that p’’ only if one is familiar with a fictional world such that a ‘‘modally

safe’’ explanation for p is embedded in the details of that fictional world.

P*2. If proposition p belongs to the class FP (where a proposition is a member of

this class iff it is a philosophical proposition that is far-removed from everyday

experience), then ‘‘it is possible that p’’ is a non-basic possibility-claim and no-

one has imagined a fictional world such that a ‘‘modally safe’’ explanation for p is

embedded in the details of that fictional world.

C*. So, if proposition p belongs to class FP, then one is not justified in asserting

the possibility of proposition p.

The only matter that needs to be settled now, before we can safely support the

conclusion of this argument, is to defend P*2 above. As promised, my defense will

be an inductive one: I will cast doubt on the claim that anyone has Yablo-conceived

of ghosts or ghouls (I will assume, incidentally, that it is clear that ghosts and ghouls

do not, in fact, exist).

It is easy to come up with a fictional world in which a ghost exists (if your

imagination fails you, borrow the fictional world from The Sixth Sense). However, it

is required, for this imagining to count as Yablo-conceiving, that all the

controversial modal facts in this world have a ‘‘modally safe explanation’’ that

emanates from the uncontroversial modal facts. To this end, put the claim that

ghosts exist in this world, on hold. Now that it hasn’t been stipulated that ghosts

exist in this world, what might lead one to conclude that ghosts exist in this world?

Remember, being able to conclude, unequivocally, that ghosts exist in this world,

implies having a world that is incompatible with the claim that ghosts don’t exist.

What, however, in this imagining compels one to assert the existence of a ghost? A

supernatural realm cannot be imagined as part of the world—this is just as

controversial a claim as the fact that the ghost exists. Ought we to construct the

world so that there are events in that world that are meant to be consequences of the

ghost interacting with the physical world? Hollywood movies favour showing

objects that move, yet have no physical cause behind their moving. Yet an object

that moves with no physical cause is itself a controversial modal claim. Indeed, it is

of the same ilk as imagining somebody moving an object with nothing but the power

of their mind. Surely, however, few are willing to accept the possibility of

telekinesis very easily. In short, I am not convinced that anyone has imagined a

fictional world in which ghosts exist according to the conditions required to satisfy

Yablo-conceivability.

Now, imagine a situation in which there is a ghoul. The point about this ghoul is

that, based on outward appearances, there is no difference between himself and a

conscious human being. He displays all the behaviour that we normally use to

conclude (unobservable) consciousness. Yet, the fictional fact is that there is no


123

conscious experience for the ghoul. Now, in the interests of ensuring that our

fictional world verifies the existence of this ghoul, let us put on hold the stipulation

that this human being is not conscious. Given this, is there anything about this

situation that would entail that this human being is not conscious? Indeed, the

fictional facts of our world are more likely to lead one to believe that our ghoul is

conscious, since he displays both the behavioral tendencies of conscious being, and

any (fictional) medical examination of his brain will reveal a normally working

brain. There is little reason to think this human being is not conscious, let alone

verification that he is not conscious.

With this, my inductive defense of the second premise of the revised skeptical

argument is complete. Hence, I advocate the truth of the conclusion of the revised

skeptical argument.

Acknowledgements I’d like to express my thanks to David Martens, of the Wits University philosophy

department, for bringing van Inwagen’s paper to my attention and for providing invaluable comments

over the course of the construction of this paper.

References

Geirsson, H. (2005). Conceivability and defeasible modal justification. Philosophical Studies, 122, 279–

304.

Sosa, E. (2000). Modal and other a priori epistemology: How can we know what is possible and what is

impossible? The Southern Journal of Philosophy, XXXVIII(Supplement), 1–16.

Stalnaker, R. (2002). What is it like to be a Zombie? In: T. S. Gendler, Hawthorne, J. (eds.),

Conceivability and possibility, (pp.385–400). Oxford: Oxford University Press.

Van Inwagen, P. (1998). Modal epistemology. Philosophical Studies, 92, 67–84.

Yablo, S. C. (1993). Is conceivability a guide to possibility? Philosophical and PhenomenologicalResearch, 5(1), 1–42.

364 P. Hawke

123

Documents

Van Inwagen’s modal skepticism