Open Book Publishers logo Open Access logo
  • button
  • button
  • button
GO TO...
Contents
Copyright
book cover
BUY THE BOOK

1. Reflections on Moore’s ‘Proof’

© 2017 Mark McBride, CC BY 4.0 https://doi.org/10.11647/OBP.0104.03

Dogmatism, immediate justification, and basic knowledge are to be exposed to a pattern of objection. I begin from Wright’s (1985) discussion of Moore’s (1939) ‘Proof’, and his introduction of the notion of failure of warrant transmission (related to epistemic circularity). I argue, against the standard view, that Wright’s position does not enjoy an advantage over dogmatism.

0.1 Suppose one has a visual experience as of having hands, and then reasons as follows:

(MOORE)

(1) I have hands.

(2) If I have hands an external world exists.

(3) An external world exists.

Suppose one’s visual experience gives one defeasible perceptual warrant, or justification, to believe (1) — that is, one’s experience makes it epistemically appropriate to believe (1).1 And suppose one comes to believe (1) on the basis of this visual experience. The conditional premise (2) is knowable a priori.2 And (3) can be established by modus ponens inference. If one reasons thus, say one is engaged in (MOORE)-reasoning.

What, if anything, is wrong with (MOORE)-reasoning? I consider two prominent responses to this question — the dogmatists’ and Crispin Wright’s. Each finds fault in (MOORE)-reasoning, but on different grounds.

0.2 Let’s start with the following accounts of (non-)transmission of warrant:

(α) Non-transmission of warrant

Epistemic warrant is not transmitted from the premises of a valid argument to its conclusion if the putative support offered for one of the premises is conditional on its being antecedently and independently reasonable to accept the conclusion. (Davies 2004: 221)3

(β) Transmission of warrant

A valid argument [transmits warrant] if […] to have warrant for its premises and then to recognise its validity is to acquire — perhaps for the first time  a warrant to accept the conclusion. (Wright 2003: 57)

Wright’s principle (β) can be contrasted with a closure principle for (an inclusive notion of) warrant:

(WC) If one has warrant for p, and knows or justifiably believes that p entails q, then one has warrant for q.

On the face of it, the key difference between (β) and (WC) is that the latter says nothing about “for the first time”. Acceptance of (WC) for an inclusive notion of warrant does not require acceptance of a closure principle for each more specific kind of warrant.4

0.3 Dogmatists, like James Pryor (2000),5 can be taken to employ a notion of negative entitlement: in order for one’s visual experience to generate warrant for (1), there is no need to have antecedent warrant for (3); in the absence of defeaters one has an entitlement — negative in nature — not to adopt the attitude of doubt towards (3).6 For the dogmatist, then, there is (or can be) warrant transmission in (MOORE)-reasoning: by recognising the entailment from (1) to (3) the dogmatist is in a position to ground a belief in (3) on the very warrant that grounds his belief in (1). If there is (or need be) no transmission failure in a piece of (1)-(3) reasoning, the dogmatist must explore other avenues for finding error in (MOORE)-reasoning.

Wright, meanwhile, employs a notion of positive entitlement: in order for one’s visual experience to generate warrant for (1), there is need to have antecedent warrant for (3); and we do in fact have antecedent warrant — an entitlement — for (3).7, 8 So warrant is — in the absence of defeaters — generated for (1); it is just not transmissible to (3). For Wright, then, there is transmission failure in (MOORE)-reasoning and, resultantly, no alternative avenues need be explored for finding error in (MOORE)-reasoning.

In sum, philosophers locating the defectiveness of (MOORE)-reasoning in transmission failure, as Wright does, typically take (MOORE)-reasoning to suffer from a form of epistemic circularity: it is only if one already has warrant for (3) that one can have a warrant for (1). If (MOORE)-reasoning suffers from transmission failure in this way, we can say that propounding (MOORE) cannot offer a new route to its conclusion. As dogmatists deny that it’s only if one already has warrant for (3) that one can have a warrant for (1), for them warrant for (1) can cross over the conditional to provide a first-time warrant for (3).

1. The Standard View

1.1 Nevertheless, the dogmatist thinks (MOORE)-reasoning can be problematic: is it really this easy to get warrant for the falsity of a global sceptical hypothesis? Assume, then — with the dogmatist and Wright — that something (at least) seems wrong with (MOORE)-reasoning. Consequently, it’s a strength of any account of the structure of perceptual warrant(s) that it can give a plausible account of why something seems wrong with (MOORE)-reasoning. Call the particular problem of giving a plausible account of why something seems wrong with (MOORE) reasoning given it can transmit warrant, the (MOORE)-transmit problem.

I want to focus on the (MOORE)-transmit problem. It’s seemingly faced by the dogmatist, but not by Wright: in (MOORE)-reasoning warrant is (or can be) transmitted according to the dogmatist, but cannot be transmitted according to Wright. Call this view of the (MOORE)-transmit problem the standard view. My claim is that the standard view is wrong. The (MOORE)-transmit problem is not just the dogmatist’s problem, it’s Wright’s too: there’s a sense in which there’s transmission of warrant in (MOORE)-reasoning for Wright. And I make a stronger claim: not only is the (MOORE)-transmit problem Wright’s problem (contra the standard view), it’s furthermore more troubling for Wright than for the dogmatist. So, at least, I shall argue.

Why is opposing the standard view important? Assume answering the (MOORE)-transmit problem is not straightforward. Then, if the standard view is correct, a strength of Wright’s account of the structure of perceptual warrant(s) is that it doesn’t face the (MOORE)-transmit problem. If the standard view is correct, then to that extent we would have some motivation to adopt Wright’s account over the dogmatist’s. If it’s wrong, however, this motivation disappears.

1.2 To fully understand the standard view note that, for Wright, (3) lies outside the domain of cognitive achievement: it’s “outside the domain of what may be known, reasonably believed, or doubted” (1985: 470–71). Wright so concludes upon consideration of the following pattern of sceptical argument (437–38):9

(a) All our evidence for particular propositions about the material world […] depends for its supportive status on the prior reasonableness of accepting [(3)].

(b) For this reason [(3)] cannot be justified by appeal to such evidence.

(c) [(3)] cannot be justified any other way.

(d) [(3)] may be false.

In responding to this argument, Wright accepts (a) and (b), but questions (c) — which calls (d) into question. Wright notes that (c) would be falsified if “it could be reasonable to accept [(3)] without reason; that is without evidence” (450). And if(f) such acceptance is reasonable one has non-evidential warrant for, or epistemic entitlement to, (3). For Wright, in (1985), it can be reasonable to accept (3) without evidence because (3) (and similar ‘cornerstones’ — see Wittgenstein (1969: 341–43)) lies outside the domain of truth-evaluability: (3) is, for Wright, not up for grabs as either true or false — it is beyond doubt.10

Wright now (2004: 167, n. 2) remarks, however:

The major strategic contrast with the present proposal is in how such warrant is conceived. In [1985] I proposed that [(3)] might be regarded as defective in factual content and that accepting [it] might accordingly be freed from the requirements of evidence that I took to be characteristic of the factual. In the present discussion non-factuality is not assigned a role in making a case that rational acceptance need not be evidence-based.

It’s one thing to say that non-factuality isn’t assigned a role in making a case that rational acceptance of (3) needn’t be evidence-based, and another to say that (3) is no longer to be regarded as non-factual. Wright here commits to the former (justificatory) claim, but makes no commitment to the latter (semantic) claim. The claims are distinct, and we can forbear from ascribing the latter claim to Wright, while respecting his shift with regard to the former.

Entitlement to accept a proposition P, for Wright (2004: 175), is unearned: I may have it “even though I can point to no cognitive accomplishment in my life, whether empirical or a priori, inferential or non‐inferential, whose upshot could reasonably be contended to be that I had come to know P, or had succeeded in getting evidence justifying P”. It’s in this context that Wright introduces the notion of ‘mere acceptance’ of (3), as distinct from a more full-blooded acceptance, involving belief:11

We may propose the notion of acceptance of [(3)] as a more general attitude than belief, including belief as a sub-case, which comes apart from belief in cases where one is warranted in acting on the assumption that P or taking it for granted that P or trusting that P for reasons that do not bear on the likely truth of P. Of course one may — sometimes irrationally — also believe P on such occasions, in the sense implicit in a conviction that one knows that P. Successful sceptical arguments may then embarrass such convictions […] Such scepticism may prove to carry no challenge, nevertheless, to the corresponding acceptances and […] warrant to accept — rather than to believe — cornerstone[s] may be enough to block the sceptical paradoxes that attend arguments to the effect that there is no such thing as getting evidence to believe them. (2004: 177)

Wright recognises a burden to delimit the scope of what we have non-evidential warrant to accept: “This is a good result […] only if it is selective — only if the entitlements generated turn out to be cornerstones of our actual ways of thinking about and investigating the world and do not extend to (what we would regard as) all manner of bizarre and irrational prejudices.” (2004: 195)

1.3 We’ve now considered the standard view of the (MOORE)-transmit problem. We can go further and consider how to answer it. A dogmatist answer has been proposed by Davies (2004: 238–43). Davies (now) claims that there is (or can be) warrant transmission in (MOORE)-reasoning, while conceding that something (still) seems wrong with (MOORE)-reasoning. But, if warrant transmission in abstracto isn’t the problem, what is?

Davies suggests (2004: 240–41):

[T]here is a kind of epistemic project whose conduct is conditioned by suppositional doubt, namely, the project of settling the question whether a particular proposition is true […] Imagine that I undertake the project of settling the question whether or not there is an external world as ordinarily conceived […] by deploying the warrants I have for believing the premises of Moore’s argument. I begin the project by regarding the question of the truth of the conclusion of Moore’s argument as open pro tem. So my conduct of the question-settling project is conditioned by the initial supposition that Moore’s conclusion is, or may very well be, false12 […] [T]hen I could not rationally regard my experience as constituting a warrant for believing [(1)]. In short, I cannot settle the question whether or not [3] is true […] by deploying the epistemic warrant I have for believing [(1) and (2)].13

This answer to the (MOORE)-transmit problem delimits the epistemic situations in which (MOORE)-reasoning transmits warrant. If Wright faces the (MOORE)-transmit problem, it seems he can’t avail himself of this answer. Davies’s answer to the (MOORE)-transmit problem rests on suppositionally doubting (3). But (3)’s status as a cornerstone, for Wright, precludes its truth being doubted — suppositionally or otherwise. (3) isn’t — on pain of irrationality — eligible for doubt; it’s to be unquestioningly accepted. Put differently: according to the ‘cornerstones’ conception of (3) in (MOORE)-reasoning, there’s no question-settling project that can even be attempted. So, for Wright, the problem with (MOORE)-reasoning cannot, it seems, be that the argument cannot be deployed in settling the question whether or not (3) is true — for there’s no such project.14

This is my (tentative and separable) stronger claim: not only does Wright face the (MOORE)-transmit problem; he faces it in a particularly debilitating way. I consider this stronger claim in section 3. First I need to demonstrate — contra the standard view — that Wright indeed faces the (MOORE)-transmit problem.

2. The Standard View Contested

2.1 Wright’s analysis of (MOORE)-reasoning15 seems internally consistent. But there’s a wrinkle. Stephen Schiffer (Wright 2004: 177, n. 8) has argued that Wright faces what Wright himself terms the leaching problem:

How exactly does [my analysis] promise to shore up the possibility of justified belief in [(1)]? We are proposing to concede, after all, that we may indeed have no (evidentially) justified belief in [(3)] — that maybe we can point to no cognitive accomplishment of which the effect is a reason to take it that [(3) is] more likely to be true than not — but countering that we may nevertheless be rationally entitled to accept [(3)]. But if standard closure principles govern justified belief, then the counter comes too late to do any good. Standard closure principles will have it that justified belief in [(3)] will be a necessary condition for justified belief in anything one knows to entail it. To surrender the former will therefore be to surrender justified belief in [(1)] […] Maybe an entitlement to accept them nonetheless can be salvaged. But the idea was to use entitlement to save justification, not to replace it. (Wright 2004: 177–78)

In short: if justified belief is closed under known entailment, and if in addition, as on Wright’s view, we cannot justifiably believe (3), then — problematically — we cannot justifiably believe (1).

2.2 The leaching problem is a nice one. And Wright has a nice answer thereto:

[I]t cannot be that evidentially justified belief is closed under (known/justifiably believed) entailment […] But if we let ‘warrant’ disjunctively cover both evidential justification and entitlement, it can still be that warrant, inclusively so understood, obeys closure principles suitable to do justice to our strong conviction that ‘justification’ — pre-theoretically understood — should do so. (2004: 178)

In sum: we can preserve justified belief in (1) without it following that we justifiably believe (3). This involves abandoning the closure of evidentially justified belief under known/justifiably believed entailment. But, according to Wright, “[t]hat is not so remarkable once one notices that evidential relations themselves are not so closed” (178) — indeed Wright takes this to be the “minimal lesson” of Dretske’s (1970) ‘zebra’ example. Closure is maintained, however, for Wright’s inclusive ‘disjunctive’ notion of warrant (cf. principle (WC) introduced at 0.2): evidential warrant for (1), given (2) is known/justifiably believed, entails some kind of warrant — evidential or otherwise — for (3). And, for Wright, there’s merely non-evidential warrant for, or entitlement to, (3).

The leaching problem originated as a closure-based problem for Wright. I want to use Wright’s response thereto to generate the transmission-based (MOORE)-transmit problem for Wright. In brief, Wright’s response to the leaching problem introduces a disjunctive notion of warrant; and reflection on that disjunctive notion of warrant prompts consideration of (non-)transmission principles specific to the evidential disjunct (see 2.6–2.9 infra). Consider (abstracting, for now, from these specific (non-)transmission principles): in virtue of an entitlement to (3) I can justifiably believe (1). The conditional premise in (MOORE) is knowable a priori. So I can believe that. And there is a simple inferential route to (3). So it seems I’d do well, epistemically speaking, to believe (3). This prompts the question: if we start out with an entitlement to (3), and then engage in (MOORE)-reasoning, should we, resultantly, come to justifiably believe (3)?16 I think the answer is ‘yes’. I’ll argue (contra Wright) that evidential warrant does transmit in (MOORE)-reasoning for Wright. The resultant justified belief in (3) means Wright faces the (MOORE)-transmit problem. In essence, the point here is that there is transmission failure when there is epistemic circularity; but, I’ll argue, there is no circularity in (MOORE)-reasoning involving specifically evidential warrant.

2.3 Let’s fix rules of engagement with Wright. What am I granting Wright? First, his (non-)transmission of warrant theses (see 0.2); second, his notion of positive entitlement (see 0.3); third, his (putatively epistemic) notion of non-evidential warrant (see 1.2); and, finally, his (stipulated) disjunctive notion of warrant and its associated closure principle (see 2.2).17

2.4 What am I not granting Wright? First, I leave the question whether evidentially justified belief is closed under known/justifiably believed entailment open pro tem (see 2.2).18 Here this is cashed out in terms of whether there is an entailment from evidentially justified belief in (1) — assuming knowledge/justified belief in (2) — to evidentially justified belief in (3).

Second, recall Wright takes (3) to be non-factual (see 1.2). So, for Wright, the question: if we start out with an entitlement to (3), and then engage in (MOORE)-reasoning, should we come to justifiably believe (3)?, is a closed one. Even if — pace Wright — (MOORE)-reasoning transmits (evidential) warrant, there’s a sense in which, for Wright, one still shouldn’t — indeed couldn’t — come to justifiably believe (3). But to grant Wright this at the outset would be to foreclose the possibility of warrant transmission in (MOORE)-reasoning for Wright. We must, rather — in order not to foreclose this possibility — find a way to make it possible to come to justifiably believe (3) by (MOORE)-reasoning while respecting as much of Wright’s analysis of (MOORE)-reasoning as possible. Here are three non-conditional routes to achieving this:

(A) Leave it an open question whether (3) is non-factual.

(B) Grant that (3) is non-factual, but allow that one can come to justifiably believe (3) by (MOORE)-reasoning.

(C) Deny that (3) is non-factual, and allow that one has an entitlement to (but not justified belief in) (3) prior to (MOORE)-reasoning.

2.5 Each of these three strategies is worthy of extensive discussion and each faces possible objections. In brief, however: (A) has the merit of not denying that (3) is non-factual, but is vulnerable to the objection of not properly engaging with Wright, for whom (3)’s status as non-factual is seemingly not an open question. (B) respects Wright’s claim about (3)’s non-factuality, but, by allowing a justified belief in (3) to be generated by (MOORE)-reasoning, it threatens inconsistency. (C), meanwhile, promises to generate more questions than it answers: if (3) is — pace Wright — factual, on what epistemic basis could we have an entitlement to (3)?19 I thus assume merely the following conditional, which I take to be the barest commitment we must make:

(D) Even if (3) is non-factual this doesn’t automatically rule out the possibility of coming to justifiably believe (3) by (MOORE)-reasoning.

That is, (3) may (or may not) be non-factual, but, if it is, this doesn’t preclude coming to justifiably believe (3) by (MOORE)-reasoning. By endorsing (D) we can postpone the question of which of (A)-(C) is to be endorsed and proceed to engage with Wright. A corollary of the minimal nature of (D) is that it allows engagement with positive entitlement theorists other than Wright, who may share Wright’s notion of entitlement yet not hold that (3) is non-factual (cf. n. 8 supra). Our focus, though, remains on Wright.20

Suppose, however, perhaps predictably, Wright refuses to accept (D). Four (related) responses are in order. First, consider the following remarks of Kent Bach (online manuscript):

There are many everyday *beliefs* and *statements*21 that are not true or false, or at least not straightforwardly true or false (to borrow a phrase from Field [1994]). There are many common one-place predicates, such as ‘large’, ‘poisonous’, ‘tasty’, ‘interesting’, ‘disgusting’, and ‘illegal’, which do not express one-place properties but which we often use as if they do. We use them in ordinary subject-predicate sentences to make statements and thereby express beliefs. Although we say ‘Fido is large’ to mean that Fido is large for a dog, ‘That mushroom is poisonous’ to mean that it is poisonous to humans (we certainly do not mean that it is poisonous to all creatures), and ‘Anchovies are tasty’ to mean that they are tasty to oneself, these utterances, taken strictly and literally, are neither true or false. They are true or false only relative to something, something that these utterances do not make explicit (there are many sorts of relativity, e.g., category-, argument place-, location-, time-, reference frame-, and norm-relativity). They would be straightforwardly true only if that something were made explicit. (My emphases added)

Bach’s remarks detail a substantive position in the philosophy of language, and I concede that there is no clear analogy between the statements Bach considers and statements involving cornerstones.22 Nonetheless, Bach’s remarks reveal there is conceptual space (at least) for justified beliefs in the realm of the non-factual; and such conceptual space is all that is required for a defense of (D). Second, we noted at 1.2 that, for Wright, non-factuality is no longer assigned a justificatory role in making a case that rational acceptance of (3) need not be evidence-based. Given this downplaying of the justificatory role of (3)’s non-factuality it might seem dubious for Wright to then give it the devastating role of automatically ruling out the possibility of coming to justifiably believe (3) by (MOORE)-reasoning. Third, we can grant Wright, consistently with endorsing (D), that one cannot come to justifiably believe (3) simply by redeploying one’s non-evidential warrant for (3):23 (D)’s consequent only speaks of not automatically ruling out coming to justifiably believe (3) by (MOORE)-reasoning. Finally, a different component of (D)’s minimal nature can be emphasised: its consequent only speaks of not automatically ruling out the possibility of coming to justifiably believe (3) by (MOORE)-reasoning. Endorsing (D) leaves open not ultimately coming to justifiably believe (3) by (MOORE)-reasoning. In sum, rejection of (D) does not seem dialectically advisable for Wright (recall, also, how much of Wright’s analysis of (MOORE)-reasoning we granted him at 2.3). Nonetheless, should Wright resolutely refuse to accept (D) our focus would shift to an engagement with positive entitlement theorists other than Wright, who may share Wright’s notion of entitlement yet not hold that (3) is non-factual (again: cf. n. 8 supra).

2.6 Recall, by (α) and (β) (see 0.2), and operating with Wright’s disjunctive notion of warrant, warrant fails to transmit, for Wright, in (MOORE)-reasoning: one needs antecedent (non-evidential) warrant for (3) in order for one’s visual experience to provide warrant for (1).

While Wright operates with a disjunctive notion of warrant, we need to separate the disjuncts again for our analysis of (MOORE)-reasoning. To see why (MOORE)-reasoning does transmit warrant for Wright, we need to set out two more specific, plausible accounts of (non-)transmission of warrant, catering for (non-)transmission of evidential warrant. Consider, initially:

(γ) Non-transmission of evidential warrant

Evidential epistemic warrant is not transmitted from the premises of a valid argument to its conclusion if the putative evidential support offered for one of the premises is conditional on its being antecedently and independently reasonable to evidentially accept the conclusion.

Theses (α) and (γ) are independent theses: (MOORE)-reasoning fails to transmit warrant by (α), yet doesn’t fail to transmit warrant by (γ). So there is non-transmission of (disjunctive) warrant, but there isn’t non-transmission of evidential warrant, in (MOORE)-reasoning. In other words, thesis (α) is met, while thesis (γ) is not.24

2.7 So the names of the theses can mislead, for satisfaction of (α) doesn’t, as the names might suggest, entail satisfaction of (γ). How counterintuitive a result is it that (MOORE)-reasoning fails to transmit (disjunctive) warrant but doesn’t fail to transmit evidential warrant? Warrant is a positive epistemic relation a subject can have to a proposition. Is it not therefore contradictory to say — as it might seem I want to — that (MOORE)-reasoning fails to provide one with a positive epistemic relation (disjunctively construed) to (3), but doesn’t fail to provide one with a proper subset of that positive epistemic relation to (3)?25

We should recognise that, while (MOORE)-reasoning doesn’t fail to transmit evidential warrant by (γ), nor does it transmit evidential warrant by (β). If the matter were left there, this would be an unhappy result. But we can construct a more specific, plausible condition for the transmission of evidential warrant (similar in form to (β)):

(δ) Transmission of evidential warrant

A valid argument transmits evidential warrant if to have warrant for its premises (including evidential warrant for at least one of its premises) and then to recognise its validity is to acquire — perhaps for the first time — an evidential warrant to accept the conclusion.

Now we have the result that (MOORE)-reasoning both doesn’t suffer from non-transmission of evidential warrant by (γ), and does transmit evidential warrant by (δ). This is perfectly compatible with (MOORE)-reasoning suffering from non-transmission of (disjunctive) warrant by (α) and failing to transmit (disjunctive) warrant by (β).

2.8 Here’s the dialectical position. We’ve four prima facie plausible theses of (non)transmission of warrant: (α), (β), (γ), and (δ). By (α) and (β) (jointly) there’s non-transmission of (disjunctive) warrant; by (γ) and (δ) (jointly) there’s transmission of evidential warrant. (Wright can’t just elect to use (α) and (β), rather than (γ) and (δ). I’m putting all four forward as plausible theses. It’s up to Wright to show those theses disfavourable to him to be implausible.) But what is counterintuitive about this? Sure, it is counterintuitive if reasoning with an argument can provide evidential warrant for its conclusion but no (disjunctive) warrant for its conclusion. To allow this is to allow a contradiction, akin to: x is a married man but not a man. But this isn’t the result these four theses deliver. If a subject has evidential warrant for an argument’s conclusion it follows he has (disjunctive) warrant for that self-same conclusion. No work to which these four theses can be put places this result in jeopardy. The only thing the four theses mandate is that evidential warrant will, while (disjunctive) warrant will not, transmit in (MOORE)-reasoning. Specifically, one’s warrant for having hands — a visual experience as of having hands — will transmit insofar as it’s (quite naturally) treated as evidential warrant, but will fail to transmit insofar as it’s treated as (disjunctive) warrant. Maybe that’s a little puzzling; but it’s not contradictory.

2.9 To emphasise: evidential warrant doesn’t fail to transmit by (γ) as the putative support offered for one of (MOORE)’s premises is only conditional on its being antecedently and independently reasonable to non-evidentially accept (MOORE)’s conclusion. This seems right. Why should having an antecedent non-evidential warrant to accept (3) preclude evidential warrant for (1) transmitting across the conditional to generate — crucially, for the first time — evidential warrant for, and justified belief in, (3)? A principle on which one might base an objection is:

(γ*) Non-transmission of evidential warrant

Evidential epistemic warrant is not transmitted from the premises of a valid argument to its conclusion if the putative support offered for one of the premises is conditional on its being antecedently and independently reasonable to accept — evidentially or non-evidentially — the conclusion.26

But (γ*) is implausible. Consider Wright operating with (γ*) when analysing (MOORE)-reasoning. He’d point out that having non-evidential warrant to accept (3) is a precondition for one’s visual experience evidentially warranting (1). Fine. But (MOORE)-reasoning generates for the first time (Wright’s phrase) a justified belief in (3). The only thing presupposed is the reasonableness of a mere acceptance of — that is, a non-belief in — (3). So it’s unclear why the putative generation of a justified belief in (3) for the first time should suffer from non-transmission of evidential warrant.

My opposition to (γ*) is anchored by:

(MOORE-EW) One’s having non-evidential warrant for (3) is no bar to one’s acquiring evidential warrant for (3) in virtue of competent inference from (1), for which one has evidential warrant.

Note the confined scope of (MOORE-EW). Perhaps its moral is extendable, mutatis mutandis, to reasoning by means of a more general class of arguments. One might, then, attempt to anchor (MOORE-EW) by:

(EW) One’s having non-evidential warrant for the conclusion of a valid argument is no bar to one’s acquiring evidential warrant for that conclusion in virtue of competent inference from premises,27 for (some of) which one has evidential warrant.

But, given our focus is on (MOORE)-reasoning, I commit only to (MOORE-EW).

2.10 As a final dialectical point, suppose I’m taken to have failed to make out my case for the implausibility of (γ*). As we’re now separating the disjuncts of disjunctive warrant, we’d have two prima facie plausible (sufficient) theses for non-transmission of evidential warrant, (γ) and (γ*), the former mandating (in conjunction with (δ)) transmission of evidential warrant, the latter mandating non-transmission of evidential warrant. This would be a puzzling state of play; but such a state of play is more troubling for Wright than for me. The onus is on Wright to show (γ) to be implausible. And so long as (γ) (and its positive cousin, (δ)) is plausible — notwithstanding (γ*), arguendo, is prima facie plausible too — Wright faces the (MOORE)-transmit problem.

2.11 In sum: (γ) is not satisfied. There’s no non-transmission of evidential warrant: by starting off with non-evidential warrant for (3), and then engaging in (MOORE)-reasoning, one generates — for the first time — an evidential warrant to accept the conclusion, (3). In other words, (δ) is satisfied.

This is what Wright must say (cf. n. 20 supra). He’ll recoil from saying this: (3) is non-factual, and we shouldn’t be in the business of generating justified beliefs in (3). Nevertheless, evidential warrant transmits for Wright, and justified belief in the existence of an external world has been delivered via a visual experience and a simple a priori inference.28 The standard view has it that the (MOORE)-transmit problem targets only the dogmatist, but, with a more refined view of (non-)transmission of warrant, we can see it targets Wright too.

3. The Potentially Devastating Nature of the (MOORE)-Transmit Problem

3.1 Let’s take stock of the dialectical situation: the dogmatist says that (MOORE)-reasoning transmits warrant — it can provide a first-time warrant to justifiably believe (3). So the dogmatist faces the (MOORE)-transmit problem. On the standard view, Wright doesn’t face that problem because he says that (MOORE)-reasoning doesn’t transmit warrant; rather it’s an example of transmission failure. But I’ve shown Wright can, and should, allow that, beginning from an unearned warrant to trust in (3), (MOORE)-reasoning provides a route to a first-time earned warrant (based on perception plus inference) to believe (3) — provided, that is, we operate with assumption (D) (see 2.5). So Wright (too) faces a version of the (MOORE)-transmit problem.

3.2 One response — Pryor’s — to the (MOORE)-transmit problem on behalf of the dogmatist is to say that (MOORE)-reasoning seems wrong because it’s dialectically unconvincing. As Wright is committed to the idea that (MOORE)-reasoning suffers from a genuinely epistemic defect — viz. transmission failure — Wright cannot locate its defectiveness merely in its dialectical unconvincingness.

On Davies’s (2004) view, (MOORE)-reasoning suffers not merely from dialectical unconvincingness; it also suffers from a kind of epistemic limitation. (MOORE)-reasoning cannot be deployed in the service of a particular kind of epistemic project that he calls settling the question. Can Wright say something along the same lines? Earlier I suggested not. Recall, Davies accepts there need be no transmission failure in (MOORE)-reasoning, but argues that, in circumstances in which one suppositionally doubts (3), warrant no longer transmits: one’s visual experience won’t provide a rationally deployable warrant for (1) or (3). But, given Wright’s ‘cornerstones’ conception of (3) in (MOORE)-reasoning, Wright isn’t in a position, on pain of irrationality, to doubt (3)’s truth — suppositionally or otherwise. The question-settling project is conditioned by suppositional doubt, but Wright cannot allow that (3) is subject to doubt — not even to suppositional doubt. Wright (1985: 470) does say: “[Cornerstones] may, in a different context, take on a more purely hypothetical role; and […] our confidence in them, in such a context, may be defeasible by empirical or theoretical considerations”. I am not sure which contexts Wright has in mind, but while the inputs to the question-settling project can be “purely hypothetical” — in (MOORE)-reasoning I suppose, just for the purposes of the project, that (3) is false — its outputssettling questions — are clearly not.29

For Wright, then, there’s seemingly unfettered transmission of evidential warrant in (MOORE)-reasoning — a curious result. Let’s briefly consider three responses to this peculiar outcome. The first two are possible responses from Wright; the third is a possible response from a hypothetical positive entitlement epistemologist.30

3.3 Response 1. To summarise the dialectic: (i) we outlined Wright’s view that (3) is non-factual and so isn’t eligible to be justifiably believed; (ii) we saw that (MOORE)-reasoning does, for Wright, transmit evidential warrant, and that, despite Wright’s views about (3)’s non-factuality, Wright ends up with a justified belief in (3); and finally (iii) we observed that, because of Wright’s ‘cornerstones’ conception of (3) in (MOORE)-reasoning (which, for Wright, entails (3)’s non-factuality), he’s not able to doubt (3) — suppositionally or otherwise. Wright might seize on steps (ii) and (iii). He might contend that, in step (ii), I compromised one of his key semantic claims about cornerstones like (3) — viz. that they’re non-factual — by pinning a justified belief in (3) on him. But then in step (iii), he might contend, I inconsistently relied on that self-same semantic claim by denying him the possibility of doubting (3).

To answer this (and the third) response we need to have recourse to theses (A), (B), and (C) (see 2.4). In light of this response, (B) seems optimal. To adopt either (A) or (C) would disable us from taking step (iii) — doubt about (3) would then be possible. Only with (B) in hand can we consistently take both steps (ii) and (iii).

3.4 Response 2 (to step (iii)). Wittgenstein remarks: “Doubting and non-doubting behaviour. There is a first only if there is a second.” (1969: 354) Drawing on this, Wright might claim: In step (ii) you lumbered me with a justified belief in (3); (only) once you’ve done that, I’m now in a position to doubt (3).31 Having anticipated this possible reply I leave it to Wright to speak for himself.

3.5 Response 3. The claim that Wright cannot avail himself of Davies’s answer to the (MOORE)-transmit problem relies on Wright’s view about (3)’s non-factuality. So we might, to concretise things, consider a positive entitlement theorist employing Wright’s notion of entitlement yet adopting (C) (cf. n. 8 supra). Such an epistemologist would be able to adopt Davies’s answer to the (MOORE)-transmit problem — he would not be precluded from doubting (3). But adoption of (C) seems unmotivated, given entitlement is a putatively epistemic notion: when (3) is ex hypothesi factual, don’t epistemic bases for accepting (3) require evidence, whether empirical or a priori (cf. n. 19 supra)?

3.6 There may be other ways for Wright to answer the (MOORE)-transmit problem.32 But in the absence of any developed suggestions, there’s a potentially devastating problem here. This is what I described earlier (1.3) as ‘my (tentative and separable) stronger claim’. We might view the problem in the form of a dilemma. Limb 1: maintain both the non-factuality of (3) and (3)’s ineligibility for justified belief. However, we’ve seen (MOORE)-reasoning generates a justified belief in (3) while respecting as much of Wright’s analysis of (MOORE)-reasoning as possible. Limb 2: allow (3)’s eligibility for justified belief (whether by denying the non-factuality of (3) or not), but maintain one can only assess (MOORE)-reasoning by means of theses (like) (α) and (β) (and (γ*) and (δ*)), and not-possibly-additionally (γ) and (δ). But there’s no argument for such a strong modal claim.

4. Conclusion

4.1 We began with the question: what, if anything, is wrong with (MOORE)-reasoning? We’ve seen that one reason, according to the standard view, for favouring Wright’s answer over the dogmatist’s doesn’t stand up to scrutiny: the (MOORE)-transmit problem is Wright’s problem as much as the dogmatist’s. In fact we can go further, though in a provisional spirit: the problem is, seemingly, Wright’s in a peculiarly devastating fashion. So much for the standard view.


1 This experience can be — indeed often for Wright it is (e.g. 2004: 170) — taken to feature as a premise. If it is so taken we will construe the transition from having this experience to coming to believe one has hands as (inductively) inferential. This is relevant to Wright’s — but not the dogmatists’ — focus on the claimability of warrant. I think this potential disconnect can be safely bracketed here.

2 This conditional need not — indeed for Wright it does not — feature as a premise.

3 Wright (2003: 57) has a similar sufficient condition for non-transmission. My ensuing claims go through, mutatis mutandis, operating with Wright’s condition for non-transmission.

4 Further understanding of why, for Wright, (WC) is exceptionless while (β) has exceptions must await the introduction of Wright’s notion of entitlement (see 1.2 infra). Endorsement of (WC), coupled with rejection of a closure principle specifically for evidential justification, is relevant to Wright’s response to the leaching problem (see 2.2 infra) — the problem providing the stimulus for my claims to come.

5 Other candidate dogmatists are Burge (1993, 2003), Davies (2004, 2009), Peacocke (2003), and Pollock (1986). As a positive matter, dogmatists take the warrant one gets for (1) on the basis of one’s visual experience to be immediate — that is, it does not rest on antecedent warrant for any other propositions.

6 The thesis preceding the semi-colon and the thesis following it are both-ways independent. Indeed, Pryor himself only explicitly commits to what precedes the semi-colon — the notion of negative entitlement is Davies’s (2004) (cf. Dretske (2000)). At this point in the chapter no particular substantive notion of entitlement itself is yet operative (cf. 1.2 infra where Wright’s notion is introduced). I characterise negative entitlement as the conjunction of these two theses, but this is merely stipulative. Moreover, one might endorse the thesis preceding the semi-colon while maintaining one in fact has antecedent warrant for (3) — see Silins (2007).

7 Again, the thesis preceding the semi-colon and the thesis following it are both-ways independent. I characterise positive entitlement as the conjunction of these two theses, but this is merely stipulative.

8 Cf. Cohen (1999: 76) and White (2006: 552–53) for such views. Cohen’s notion of non-evidential rationality, in particular, bears clear similarities to Wright’s notion of non-evidential warrant (see 1.2 infra).

9 I thus highlight Wright’s Humean, rather than Cartesian, pattern of sceptical argument.

10 Wright, following Wittgenstein, also refers to cornerstones as ‘hinge propositions’, but I bracket this since talk of propositions is not easy to square with non-truth-evaluability. Finally, for an interesting view that Wright’s and Wittgenstein’s conceptions of hinge propositions come apart, see Pritchard (2005a: 204–05).

11 Pryor (2004: 355–56) ignores this aspect of Wright’s epistemology “to keep our discussion manageable”. Unfortunately, without more, manageability is here achieved at the cost of distortion.

12 Davies adds elsewhere (2009: 369): “[A] fuller treatment of the project of settling the question would have to allow for the case where I begin by supposing that it is as likely as not that Q is false”. So principled or mandated agnosticism (see Wright 2007) about Q — in which one assigns a credence of 0.5 to both Q and its negation — might seem to be a form of suppositional doubt about Q (cf. Chapter Three, n. 9).

13 Pryor (2004, 2012) suggests the problem with (MOORE)-reasoning is not epistemic, but rather dialectical.

14 Wright and Wittgenstein differ here, in that for the latter, but not the former, doubt about cornerstones is not meaningful (and hence is not rationally possible). For Wright, doubts about cornerstones are meaningful (and can be responded to by showing that they rest on a mistaken conception of warrant for cornerstones). But for such doubts to be meaningful is not yet for them to be rationally possible. More on this later, in Section 3.

15 Wright’s (2007: secs. II and IV) is a general strategy to deal with “justificational triads” resembling (MOORE) whose subject-matter include “other minds, the laws of nature, the future, and the substantial past” (all of which can be “called into (Humean) sceptical doubt”).

16 Cf. Davies (2004: 222–23) who describes such reasoning as epistemic alchemy.

17 See also my third response to Wright at 2.5 infra for a further concession to Wright.

18 I also leave open pro tem whether “evidential relations themselves are […] so closed” (Wright 2004: 178). There is a complication here. In Wright’s presentation of the leaching problem, he talks of closure under ‘known’ entailment. Then, in his response to the leaching problem, he puts ‘known/justifiably believed’ in brackets when considering the entailment. Finally, in a footnote (178: n. 9), he appears to be focusing on a closure principle where the entailment needn’t be known or justifiably believed. Throughout, I focus on the most plausible such principles in which the entailment is known/justifiably believed (cf. n. 28 infra).

19 See Wright (2004: 205, 2008: 506, n. 3) for inchoate answers not committed to (3)’s non-factuality.

20 It must be conceded that, if Wright holds to various (Wittgensteinian) aspects of his (1985) view, then he is not open to the (MOORE)-transmit problem. But it is far from clear that Wright does now hold to all those aspects. (D), then, can be viewed as a minimal claim offered as a starting point for engagement with a Wright-like character who does not hold to all those aspects of his (1985) view (cf. also (MOORE-EW) and (EW), introduced later at 2.9, for further minimal claims with a similar function).

21 Bach uses the asterisks “to leave it open whether *statements* and *beliefs* in a given area really are true or false and qualify as genuine statements and beliefs”. So it is noteworthy that the asterisks are removed when Bach next uses ‘statements’ and ‘beliefs’, in the second italicised portion of this passage.

22 On a natural reading, Bach’s point is that utterances of these sentences are not literally true or false, yet these utterances express thoughts or beliefs that are true or false. The sentences, even as uttered on a particular occasion, are not literally true or false because they do not specify something that is crucial for truth evaluation. Yet we use the sentences to express propositions that are true or false, propositions to which we adopt attitudes such as belief or doubt — and to which we adopt these attitudes sometimes with, but sometimes without, justification or warrant. So Bach’s point depends on the distinction between sentences (or utterances thereof), on the one hand, and propositions (or attitudes towards propositions), on the other hand. It seems clear, however — our concession — that neither Wright’s nor Wittgenstein’s notion of ‘non-factual’ relates to sentences that are neither true nor false being used to express propositions or thoughts that are true or false.

23 Cf. Wright (2004: 176). To allow for non-evidential warrant to believe (3) would complicate things considerably.

24 It cannot be that thesis (γ) is met while thesis (α) is not: (simplifying) if it must be antecedently and independently reasonable to evidentially accept the conclusion, it must be antecedently and independently reasonable to accept the conclusion. And so theses (α) and (γ) are not both-ways independent.

25 Recall, transmission failure is the result of circularity. There is circularity of inclusive (disjunctive) warrant, but there is no circularity of evidential warrant. There is, I shall show, not first-time inclusive warrant for (3); but there is first-time evidential warrant for (3).

26 (γ*)’s twin transmission principle reads:

(δ*) Transmission of evidential warrant

A valid argument transmits evidential warrant if to have warrant for its premises (including evidential warrant for at least one of its premises) and then to recognise its validity is to acquire — perhaps for the first time — a warrant (disjunctively-construed) to accept the conclusion.

27 Of course, the number of ‘premises’ may be one: the relevant argument may be single-premise (cf. n. 2 supra).

28 So we finally oppose Wright’s view that: (MOORE) is a locus for a counterexample to evidentially justified belief being closed under known/justifiably believed entailment (in part) for the reason that we cannot justifiably believe (3). Consistently with this, we may ultimately wish to join Wright in rejecting this (doxastic) closure principle. We also finally oppose Wright’s view that (MOORE) is a locus for a counterexample to “evidential relations themselves [being] so closed” (cf. n. 18 supra).

29 If Wright were to adopt the view that cornerstones are no longer to be regarded as non-factual, room for (suppositional) doubt might open up. But insofar as cornerstones are still presuppositional in our various cognitive projects, it’s not clear how such doubt could be explained.

30 These responses are, concededly, only given summary treatment here, and merit more detailed consideration. The claim of this final section is, though, as we’ve already noted, tentative and separable.

31 Strictly, to make this claim, the final sentence of Wittgenstein’s must be: there is a second only if there is a first.

32 A (undeveloped) candidate: for the dogmatist, (MOORE)-reasoning provides a route to a wholly new warrant to believe (3). But for Wright it only provides an earned warrant to believe something that one already had an unearned warrant to trust.