2. A modal assumption

Traditionally, external world skepticism is not explicitly construed in modal terms. To see how considerations in the epistemology of modality become relevant to it, let us begin with what has become the standard version of the skeptical argument. Suppose that the Skeptic presents the modus tollens below and I become worried by its conclusion—that I am not justified to hold even my most mundane beliefs, e.g., that I have hands.

Argument SK:

What can be done now? Since the validity of the argument is unassailable, responses to it have concentrated on its soundness.Footnote ⁴ An antiskeptical strategy tried in the past has been to show that the premises of the argument SK are actually problematic.Footnote ⁵ Here, I shall pursue a somewhat similar idea. However, I will not deal with any of the premises directly; instead, I shall challenge the first one indirectly. The challenge to SK-1 will be indirect since my concern will be the following modal proposition:

The focus on this proposition—in which I take modality to be metaphysicalFootnote ⁷—is motivated by the observation that a most natural rendering of the first premise of the skeptical argument would be “Assuming that you accept that it is possible that you are the subject of a sks, you are not justified to believe that you are not in such a situation”; or, “Assuming SKP-0, SK-1.”Footnote ⁸ So, one wonders, what happens if I do not make this assumption? The answer, I take it, is that, in this case, nothing further can be said. In particular, one is no longer in the position to evaluate SK-1 at all. Similarly, if we recall the presuppositional analysis of the famous example involving the present king of France, if there is no such king, the issue of his baldness does not arise; hence, the proposition “The present king of France is bald” lacks a truth-value. I suggest that we face the same situation here. Once I doubt that SKP-0 holds, the justification issue raised in SK-1 does not actually arise, i.e., SK-1 lacks a truth-value. Thus, SKP-0 should be understood as a (Strawsonian) presupposition of premise SK-1. (A proposition p presupposes another proposition q iff p lacks a truth-value unless q is true.Footnote ⁹ Here, p = SK-1 and q = SKP-0). Therefore, SKP-0 has a special status in this debate: argument SK poses a threat only to someone who accepts it. So, if the Skeptic wants me, her opponent, to accept that SK-1 has a truth-value (i.e., that it is true), then I must first believe that SKP-0 holds.

At this point, there are two dialectical options left to explore. The first is that the Skeptic will ask me to provide the reason for which I doubt proposition SKP-0. The second is that the Skeptic will try to justify it herself in order to remove my doubt and, thus, to advance SK-1. As we shall see, neither option is satisfactory for the Skeptic.

The first move amounts to shifting the burden of proof on me, her opponent. This raises the further question about whether the Skeptic is entitled to do this, and, moreover, about how this new question should be answered and by whom.Footnote ¹⁰ So, let us ask what kind of proposition is SKP-0. It surely isn’t an a priori/necessary truth. Is it a “natural intuition” we have?Footnote ¹¹ Or, is it part of the background knowledge about the world? If any of these were the case, the Skeptic would be entitled to ask me for a reason to doubt such a proposition. Yet SKP-0 clearly does not share this status with “2 + 2 = 4,” “Stones do not feel pain,” or “Humans need food to survive,” and other similar propositions. SKP-0 is nothing of this kind; it is an unusual, strange thought, a bizarre possibility indeed.Footnote ¹²

But, on reflection, things are even more difficult for the Skeptic since—somewhat ironically—the obligation to justify SKP-0 stems from within her own stance. By the Skeptic’s own admission (embedded, as we recall, in the very proposal of a sks), it is uncontroversial that the appears-is gap exists in the perceptual domain: things may only appear a certain way without actually being that way. Applied to the present context, the appears-is distinction becomes a distinction between the genuine possibility that I am the subject of a sks and the situation in which this merely appears possible. Thus, the further challenge to the Skeptic is raised on the grounds that, contrary to what SKP-0 claims, it may only appear possible that I am the subject of a sks.Footnote ¹³ This challenge amounts, in the end, to the obligation to provide reasons to believe SKP-0, i.e., reasons which would a fortiori remove my suspicion that the possibility I contemplate is not genuine.

Before we move on, let me spell out the distinction between the appearance of possibility and genuine possibility. The epistemic phenomenon relevant here has been identified in different contexts and examined under the broad label of “modal error”:Footnote ¹⁴ modality/possibility judgments are quite often unreliable, even wrong, and there are many cases when certain states of affairs appear/ed possible, while they are/were not possible to exist. These cases show that we are entitled to be suspiciousFootnote ¹⁵ when it comes to distinguishing between “it is possible that p” and “it appears possible that p.” Yet, importantly, in what follows, this is not a matter of exceptionless principle, but applies only to what we called above bizarre propositions (like SKP-0).Footnote ¹⁶

To close this argument, it is crucial to realize that the position we start from is by no means one in which I should believe SKP-0 (by default, as it were), and hence I need to justify why I doubt it. On the contrary, since SKP-0 is at best located in a grey area on the spectrum of plausibility,Footnote ¹⁷ it is the Skeptic who has to present a justification for it.

I take these considerations to show that the Skeptic has no choice but to appeal to the second option—to (try to) justify SKP-0. Thus, the burden of proof is on her; she has to adduce reasons to believe that being the subject of a sks is a genuine possibility indeed. And now the thesis I shall defend below can be formulated simply as this: the Skeptic is not able to present such reasons.

An immediate reaction here may be that, on the contrary, the Skeptic is able to justify SKP-0, and in two ways. Call the first strategy the “Positive Conceivability” strategy. SKP-0 holds, the Skeptic maintains, simply because it is conceivable that I am the subject of a sks (be it the brain in a vat scenario or any other scenario). And, since what is conceivable is possible, SKP-0 follows. Just in case this first strategy fails, she may appeal to a second, alternative strategy. Let us label it the “Default” strategy. According to it, the Skeptic asserts that SKP-0 holds because I cannot rule out the possibility that I am the subject of a sks. We shall examine these two strategies below.Footnote ¹⁸ Both fail—and, it turns out, for virtually the same reason.

One more remark is in order before we move on to discussing the first strategy in detail.Footnote ¹⁹ A look at the literature reveals that SKP-0 has not received the attention it deserves,Footnote ²⁰ and a possible explanation for this lacuna is, perhaps, the initial plausibility of the two strategies mentioned above. With only a few but notable exceptions,Footnote ²¹ this proposition has not been investigated, even though it is a modal claim and there has been substantial debate around the justification of modal claims.Footnote ²²

3. Positive conceivability

A traditional strategy to justify SKP-0 is to invoke conceivability considerations. By appealing to the venerable Conceivability Principle—that, roughly, “conceivability entails possibility”—the Skeptic may claim that the statement “It is conceivable that one is the subject of a sks”Footnote ²³ entails SKP-0 (“It is possible that one is the subject of a sks”). Let us abbreviate this (Cartesian) principle of inference as SKP-1; moreover, let us grant that the Skeptic has identified a suitable version of it, which bypasses its well-known problems.Footnote ²⁴

Facing this threat, I will immediately note that this principle is a conditional. Thus, SKP-1 is inert unless its antecedent is shown to hold. So, the question to ask the Skeptic is: What is the justification to believe this antecedent? Note, once again, that I am entitled to place the burden of proof on her, since the antecedent is surely neither an intuition we all share nor a piece of established knowledge about the world. The question then is why should I accept the bizarre possibility that it is conceivable that I am the subject of a sks?Footnote ²⁵

Let us label the relevant claim here (the antecedent of SKP-1) as

Moreover, as we recall, this challenge (to justify SKP-2) is entirely compatible with the Skeptic’s own perspective. If things may only appear a certain way without actually being that way, then, in the present context, what appears conceivable may not be genuinely conceivable. Let us abbreviate the former claim as SKP-3:

The issue of “modal error” becomes relevant once again in the form of the distinction between cases when certain states of affairs appear/ed possible to conceive (conceivable), while they are/were not possible to conceive. It is the occurrence of these situations in the past that gives us the right to be distrustful when it comes to distinguishing between “it is conceivable that p” and “it appears conceivable that p.”Footnote ²⁶ (As we also recall, this guarded attitude applies only to bizarre propositions, and SKP-2 is among them).

But, what is it to be conceivable? My understanding of this notion here relies on a recent analysis due to Chalmers (Reference Chalmers, Gendler and Hawthorne2004). According to this analysis, naturally enough,

By substituting “I am the subject of a sks” for p in the biconditional (CON), another equivalence that we will need later—call it (CONsk)—follows immediately:

From (CON), through negating both sides, it follows that:

Furthermore, we can also introduce the correlate notion of “appearance of conceivability” as follows:

Thus, the worry here is, again, that since we can document cases of bizarre propositions p for which we have made the mistake of not distinguishing “it appears conceivable that p” from “it is conceivable that p,”Footnote ²⁸ the Skeptic needs a reason to justify the claim that the (bizarre) proposition “I am the subject of a sks” is not one of them. That is, she needs a reason to hold the stronger SKP-2, and not only the weaker SKP-3.

Now, before we continue to develop this line of reasoning, we need to take a short detour. Consider the following objection. One may remark that according to the principle (CON) above, the Skeptic could immediately show that it is conceivable that I am the subject of a sks—i.e., SKP-2, the left-hand of (CONsk) above. This is so since its equivalent holds, i.e., the right-hand term in (CONsk), namely that

But this is too quick. I reply to the Skeptic that she is in fact entitled to accept only a weaker claim, namely

(This does appear so given how we understand apriority, and also given what we take ourselves to know about brains, vats, computers, etc.)

To justify the claim that there is a difference between proposition (CONsk-rh) and proposition (a-CONsk-rh), I reiterate the argument I have already given above for the notion of conceivability: there are situations in which a proposition appears (or appeared) to hold a priori, and yet, it turns out, it does not (did not) hold a priori. There are actually many such casesFootnote ²⁹ when it is reasonable to say that for a (bizarre) proposition p,

This gap mirrors exactly the contrast between “it appears/ed conceivable that p”, and “it is/was conceivable that p.” Then, the conclusion of this detour is as stated above, that we can continue the discussion either in terms of conceivability or by spelling this notion in terms of apriority—it will not matter.

(We can see this even more clearly by substituting “I am the subject of a sks” for p in [ACON] above. This yields the following equivalence:

Since the right-hand side of this equivalence is granted, then, as objected, the Skeptic can assert, again, only the left-hand, namely

Now the detour has ended, and we are back to where we left off. Returning to the main thread, let us reiterate the conclusion: the Skeptic can only assert the weaker SKP-3 and not the stronger SKP-2. And this is of no use to her.

At this point, it looks like the Skeptic faces a trilemma.

The first horn is to say that although the example from set theory works fine, and although the gap between appears-that-p and is-p needs to be bridged for bizarre p’s, in the situation under consideration here, there is no such gap because asserting SKP-3 amounts to asserting SKP-2.

The second horn is to show that she can infer SKP-2 from SKP-3.

The third horn is a way out not yet discussed, but still available: justify SKP-2 on grounds independent of SKP-3.

All three horns present serious difficulties for the Skeptic, as follows.

According to the first horn, the Skeptic claims that, in the present case, the two notions overlap. If it appears (to me) that the state of affairs in which I am a brain in a vat (say) is conceivable, then it is so, i.e., it is conceivable (by me)—just as if it appears to me that I am in pain, then I am in pain. On this view, we, the individual epistemic subjects, are the ultimate authority in both matters. Thus, the concept “being conceivable” would behave like (or, as Wittgenstein would say, would have the same grammar as) “being in pain”—and unlike, for instance, “being provable.” (It is an obvious falsehood to say, for a mathematical proposition q: if it appears to me that q is provable, then q is provable.)

But the suggestion of an overlap is implausible on conceptual/“grammatical” grounds. Conceivability is, by definition, a capacity: one’s ability (capacity) to conceive something. It thus has clear rational-discursive, objective, and public connotations, just like provability. We saw that it is only natural to say “it appears/ed (to me) conceivable that p, but, upon reflection, turns out it is/was not.” So, although it is admittedly not entirely transparent what claiming that “p appears conceivable” amounts to, “p is conceivable” amounts to more. Footnote ³¹

To take stock: we saw that when it comes to the conceivability of bizarre propositions, there are two categories to consider. First, cases of genuine conceivability, when what appears conceivable is so; second, cases of modal error, when it is not so. Hence, the Skeptic was under the obligation to provide reasons that SKP-2 belongs to the first category. However, her first attempt to do this (the first horn) is not convincing.

Let us now discuss the third horn of the trilemma. (We will return to the second horn below). Can the Skeptic justify SKP-2 on grounds independent of SKP-3? It is rather hard to see how, but one proposal could be the following. Imagine that she asserts this:

(Since, unquestionably, what is actual is also possible). As is immediate, this helps only provided that she also indicates a concrete epistemic subject who does conceive a sks—that is, someone who not only utters the words “I am conceiving a sks,” but does the conceiving, as it were. And, moreover, suppose the Skeptic finds someone, call him Mr. P, prepared to declare that

This is perplexing but let us assume that the Skeptic finds someone willing to sincerely declare that. Yet, when queried about a specific skeptical scenario, Mr. P will have to concede that he does not have that much to tell us. Unsurprisingly, he is not in the position to offer (publicly verifiable) details about how the perfect simulation mechanism is supposed to work.Footnote ³² For instance, if we consider the specific brain-in-a-vat skeptical scenario, Mr. P cannot tell us where to place the electrodes in the brain, or how to write the software for the supercomputer, and so on. Thus, for any such concrete person that the Skeptic brings forward, I, her opponent, will point out that, absent these relevant, publicly ascertainable details, even if this person is totally sincere, it may only appear to them that they conceive a sks, while in fact they do not conceive it.

Despite that, the Skeptic may be tempted by the further thought that Mr. P’s actual failure is irrelevant—since, she claims, it is possible that such details be provided (by Mr. P, or someone else) in the future. But the same problem arises right away: namely, that this may only appear to be possible without it, in fact, being possible. This is the same difficulty once again, which, we now see, the Skeptic cannot avoid. Just as above, the absence of details functions one more time as a defeater of the “is possible” claim.

A complication has to be considered here, however. The Skeptic may insist that such (currently missing) details can be filled in, yet not by a regular person like Mr. P, but by a more powerful intellect, a superhuman one: by an Ideal Mind. Footnote ³³

Note that this is yet another bizarre possibility claim. The problem with it is evident: the very notion of such an Ideal Mind is unfathomable. It may well appear conceivable that such a mind (i.e., one possessing infinite memory, infinite computational speed, etc.) exists without this being conceivable: we just cannot tell. Moreover, is such a Mind, simply by virtue of being a mind, still subject to modal error? As these hesitations reveal, we are on very shaky ground when it comes to judging this situation. What appears conceivable to such a Mind may or may not be genuinely so (e.g., to conceive a sks). All in all, then, since the very notion of an Ideal Mind is extremely foggy, the Skeptic cannot appeal to it without begging the question.

Finally, let us consider the second horn. The Skeptic may try to show that the inference from SKP-3 to SKP-2 goes through.

Now things get complicated slightly, since the Skeptic realizes that she is within her rights to refuse to take on the quite difficult task to provide such a justification. That is, she notes that she is not under the obligation to accomplish this task—namely, to indicate concretely a reason why this inference holds. She observes that she is only committed to something that prima facie looks significantly less demanding: to justify the modal proposition that it is possible to provide such a justification:

This is a legitimate move. But I will immediately note, in light of our discussion so far, that even though SKP-4 asserts only a possibility, it will still need justification—because, once again, SKP-4 is a bizarre proposition, hence modal error becomes a threat. What we want, after all, is to show that a proof of the entailment SKP-3 ➔ SKP-2 is conceivable, and all this while being aware that there are many cases when although it may (have) appear/ed conceivable to establish a certain entailment, it is/was not possible/conceivable to do it (since it is/was a priori false that the entailment holds).Footnote ³⁴

At this point, it becomes clear that this attempted modal escape launches a (Carrollian) infinite justificatory regress—upward, so to speak. The Skeptic now realizes that propositions SKP-0 and SK-1 are getting further and further “away” from her—and thus, fortunately, from us too. It turns out that SK-1 is based not only on SKP-0, but also, indirectly, on SKP-1, SKP-2, and SKP-3, as well as on yet another modal proposition, SKP-4. And a justification for the bizarre SKP-4 possibility claim is needed. It will consist, presumably, in offering yet another conceivability argument.Footnote ³⁵ This is to be piled up on top of what we have amassed so far. So, more and more steps will have to be added to justify SKP-4. Since there is no end in sight for this exercise, the justificatory “tower” grows infinitely high, as follows. (One should read the argument below upward from line 1 and downward from line SK-1):

I take this to establish that the worrisome SK argument is beyond the Skeptic’s reach, so to speak; hence, it poses no threat to us. This is so since premise SK-1 may not be available to the Skeptic, since its presupposition SKP-0 becomes accessible to her only as the limit of an infinite series of justifications; that is, not at all.Footnote ³⁷ Moreover, since we recall that the specific form of the sks discussed here did not matter at all (the brain-in-a-vat scenario served only as a more vivid illustration), the conclusion is entirely general.

Yet, as announced at the outset, this is only half of the story. The Skeptic can appeal to another strategy to justify SKP-0. But it will fail as well, and the failure shares the same structure (the infinite ascent) as the failure of the first strategy.

4. The Default strategy

Now the Skeptic’s idea is to claim that SKP-0 holds simply because I, her opponent, am unable to rule out the possibility described in it; that is, because I cannot conceive reasons against the possibility that I am the subject of a sks.Footnote ³⁸ However, this move will fail as well. The root of all the forthcoming difficulties for the Skeptic is easy to identify: the claim supposed to justify SKP-0 is itself modal.

To begin with, note that here the Skeptic is supposed to rely on what can be called a Default Principle of Modal Justification roughly stated as:

Suppose that I accept this principle.Footnote ⁴⁰ Yet, since it is a conditional, it is effective only if its antecedent is shown to hold. The antecedent of interest here is:

As is clear, (~p ₁SKP-0) and DPMJ together entail p=SKP-0, by modus ponens. Then, the next challenge I raise for the Skeptic is to justify (~p ₁SKP-0). But, must she accept it? Yes, she must. The challenge brings in the idea of modal error one more time: although it may appear that reasons to rule out the possibility described by SKP-0 are not conceivable, maybe it is, after all, possible to conceive reasons to rule out this possibility. The relation between these claims is hard to fathom, since the propositions involved are on all accounts bizarre. Moreover, this need for justification arises since, as we recall, there are many cases when what appeared as not being possible to rule out turned out to be possible to rule out, and vice versa; we cannot tell whether we are in one of these situations right now. Hence, the burden of proof is on the Skeptic, indeed, to specify what entitles her to make the stronger claim (“it is not possible …”), rather than the corresponding weaker claim, that “it appears that reasons to rule out SKP-0 are not conceivable.” Although this latter claim is unproblematic, the Skeptic cannot substitute it as an antecedent in the DPMJ.

Now the Skeptic realizes that she may still have a way out. She may point out that she is not under the obligation to show, concretely, that what does not appear possible to do is indeed not possible to do. Instead, all she needs to establish is a weaker modal claim, namely that

If such reasons are conceivable, then, as always in logic, we can say that what was needed to be shown is in fact shownFootnote ⁴²; namely, that reasons to rule out SKP-0 are not conceivable.

The Skeptic realizes very well that she could help herself with (p ₂~p ₁SKP-0), but she also understands right away that she is not in that position yet because she still has to offer a justification for the bizarre modal claim above i.e., (p ₂~p ₁SKP-0). Why is it possible to show that? Why is that a genuine possibility?

Now, in line with the Default strategy, the Skeptic will say that such a claim is justified because, again, it is not possible to conceive reasons to rule out the possibility described in it, i.e., in proposition (p ₂~p ₁SKP-0). But this very claim (that it is not possible to conceive …) is a bizarre modal claim serving as the antecedent for a new application of the DPMJ principle. To say that it is not possible for someone to conceive reasons to rule out (p ₂~p ₁SKP-0) is to say that, as far as that subject is concerned:

We notice that the indices increase, and this shows where we are heading. The (Carrollian) iterative infinite ascent we already encountered is gradually building up once again. The Skeptic realizes that she is not automatically entitled to claim that (~p ₃p ₂~p ₁SKP-0), but, at most, the weaker (and useless) version of this proposition—that it only appears that reasons to rule out (p ₂~p ₁SKP-0) are not conceivable. In order to be in the position to hold (~p ₃p ₂~p ₁SKP-0), she needs further justification, and so on. SKP-0 and, thus, SK-1 are, just as in the case of the first strategy, getting farther and farther away from the Skeptic—and from us, her opponents.

More explicitly, in order to present SK-1 to us, the Skeptic has to “climb down” the stairs of the infinitely high justificatory “tower” below. (Recall that this is to be read upward from 1 to 4, and then downward from SK-1 to SKC; also, recall the DPMJ principle: “If reasons to rule out p are not conceivable, then p.”):

To close the discussion of this second strategy: argument SK is devastating, indeed—if the Skeptic presents it to us. But she cannot do this. The conclusion drawn at the end of the examination of the first strategy holds here as well. Since the SK argument’s presupposition SKP-0 turns out to be infinitely far away from the Skeptic, the argument is actually not available to her. Thus, despite strong appearances to the contrary, we, her opponents, do not get to confront it.

5. Further clarifications and concluding remarks

In this final section, I shall take up two issues that may be raised in regard to the line of thinking articulated above. Both difficulties result from a kind of cost-benefit analysis. On one hand, it is not hard to accept that this debate should be governed by the methodological principle that no response to radical skepticism should have major damaging collateral effects. Footnote ⁴⁴ On the other hand, the worry is that the strategy developed here seems to have two such unwelcome effects.

The first effect arises because of the oft-repeated requests made on the skeptic to justify her commitments. Yet, the objection goes, this is unfair: if one is required to justify all their commitments, then one becomes a victim of an infinite regress of justification.Footnote ⁴⁵ And this, in turn, leads to a radical form of skepticism. Hence, the end result is that the cost of responding to skepticism about the external world in this way is too high: the deployment of another form of radical skepticism.

The side effect identified above is worrisome, indeed—and fair, since I do aim to neutralize the Skeptic in this manner. But, on reflection, it is not much of a concern in the present attempt to deal with skepticism. Recall that the strategy proposed here relied on the idea that I, the epistemic subject haunted by the specter of skepticism, should not lose sight of the (skeptical!) distinction between the modal appears-that-p claims and the modal is-p claims; and, what is more, I should ask for justifications of the latter claims when the propositions p in question were what we called “bizarre,” i.e., strange, far removed from our familiar epistemic environment. Specifically, these were the kind of justifications that (indeed) launched the infinite regress and, thus, rendered the argument SK “unreachable.” The antiskeptical strategy articulated here does not impose the requirement to justify each and every commitment one (the Skeptic, in particular) makes, but only a certain category of them; for instance, there was no obligation on any of the parties to justify some important claims, i.e., SKP-1 or DPMJ, as well as any of the “appears-so” statements.

The second side effect is somewhat similar to the first. By asking for a justification of SKP-0 (a modal claim), I seem to endorse another form of skepticism—about modality. So, again, the cost of the asymptotic kind of victory claimed here seems too high: defeat external world skepticism by embracing another form of skepticism.

This is, again, a legitimate concern, and what hopefully answers it is that the suspicion about modality adopted here is quite limited in scope. The strategies above are compatible with the view that many of the modal claims we make in vernacular are, in fact, entirely unproblematic, since they regard familiar matters (e.g., it is possible to find beer in the fridge/to find who shot JFK/to count to 1 billion, and so on, are perfectly acceptable). Only some modal claims should be scrutinized, and SKP-0 belongs to this category given how bizarre, even fantastic, it is. Also, note that the antiskeptical conclusion was drawn without questioning any of the modal claims in the “appears-so” category (e.g., appears possible, appears conceivable, etc.).

To conclude: in devising the present strategy, I did not aim to show that there is something wrong with the skeptical argument SK per se. The fact that it relies on a certain presupposition (even a modal one) is not in itself a problem since all arguments have presuppositions. Nor did I try to provide a test whose result will tell us whether we are the subjects of a skeptical scenario. The novel idea was to pay close attention to where we stand in relation to skepticism; more precisely, to show that such a threat is infinitely far from us, so to speak. Thus, although impeccable, the Skeptic’s argument only appears threatening, without being so. The Skeptic cannot ask us, her opponents, to accept the first premise of her argument without first accepting its bizarre presupposition SKP-0. But once we ask the Skeptic to justify it (since she has to do this, not us), and once she sets out to provide this justification, we saw that she will never be able to “return” to us—unless she performs the impossible (super-)task of climbing down an infinitely high justificatory tower.