The Paradox of Moore's Proof of an External World

The Philosophical Quarterly  doi: ./j.-...x

THE PARADOX OF MOORE’S PROOF OF AN EXTERNAL WORLD

B A C

Moore’s proof of an external world is a piece of reasoning whose premises, in context, are true and warranted and whose conclusion is perfectly acceptable, and yet immediately seems ﬂawed. I argue that neither Wright’s nor Pryor’s readings of the proof can explain this paradox. Rather, one must take the proof as responding to a sceptical challenge to our right to claim to have warrant for our ordinary empirical beliefs, either for any particular empirical belief we might have, or for belief in the existence of an external world itself. I show how Wright’s and Pryor’s positions are of interest when taken in connection with Humean scepticism, but that it is only linking it with Cartesian scepticism which can explain why the proof strikes us as an obvious failure.

... I remember with particular vividness that on one such occasion I found him poring over a piece of paper on which was handwritten a very short argument entitled ‘Proof of an External World’. I read it with some dismay. ‘Moore,’ I said, ‘surely your proof is a simple petitio principii?’ ‘Indeed, it is not,’ he smilingly replied. ‘Ah, so then does it perhaps suffer a failure of warrant transmission?’ ‘No’, he said. Bewildered, I ventured, ‘Then is it merely dialectically ineffective?’ ‘No’, he said, smiling more mysteriously than ever. Then it came to me. ‘Moore’, I said, ‘is your proof merely completely irrelevant to the point at issue?’ ‘Yes!’ he said, clapping his hands with delight. From that point, we became firm friends. (With apologies to Jonathan Miller)

I. THE PARADOX

Normally, a philosophical paradox is a piece of reasoning which, starting from seemingly correct premises, leads us by seemingly unexceptionable means to an unbelievable conclusion. I propose to draw attention to an al- together different kind of paradoxical argument, one which employs only obviously valid forms of reasoning, starts from undisputed premises which, in context, are justified (or even known), leads to a perfectly acceptable conclusion, yet is still such that it seems evidently flawed. This, I submit, is the case with Moore’s claim to prove the existence of an external world:

©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly Published by Blackwell Publishing,  Garsington Road, Oxford  , UK, and  Main Street, Malden,  , USA  ANNALISA COLIVA THE PARADOX OF MOORE’S PROOF 

I can prove now, for instance, that two human hands exist. How? By holding up my depends, in Wright’s view, on his already having a warrant for the conclu- two hands, and saying, as I make a certain gesture with the right hand, ‘Here is one sion, since it is only in the context of such anterior information that he can hand’, and adding, as I make a certain gesture with the left, ‘and here is another’. And justifiably take his sense experience as a warrant for ‘Here is a hand’. The if, by doing this, I have proved ipso facto the existence of external things’ [of which proof, accordingly, is valid and proceeds from premises which there is no Moore takes his hands to be examples], ‘you will all see that I can also do it now in (non-sceptical) reason to deny are known, but it fails to be rationally per- numbers of other ways: there is no need to multiply examples.1 suasive: it cannot produce a first warrant for believing its conclusion. This proof is now standardly rendered as follows: James Pryor has contested Wright’s diagnosis.4 He agrees that doubts . Here is a hand about whether there is an external world would defeat the warrant supplied . If there is a hand here, then there is an external world by perceptual experience for the belief that here is a hand. But he thinks one . Therefore there is an external world. can certainly have a defeasible perceptual warrant for the premise that here is a hand, namely, seeming to see one’s hand, without needing any prior and This is an obviously valid inference. The first premise is quite clearly true, independent warrant for the conclusion that there is an external world – and indeed, we are willing to grant, known, in the kind of context in which indeed, without needing any collateral information at all, provided one does the proof is meant to be produced (one in which the thinker draws attention not already have any such doubts. Hence the proof, in his view, is not to his hand in front of his face). The second premise is a conceptual necess- epistemically circular. But, he suggests, it is dialectically ineffective: specifically, ity, knowable a priori (leaving idealist worries aside). The conclusion validly it fails in the anti-sceptical setting in which it is presented, since a sceptic will follows, and is absolutely acceptable and unparadoxical in any case. And yet already doubt its conclusion, that is, according to Pryor, will believe this it remains the case that Moore’s proof strikes (almost all of) us as an obviously (more probably) false, and hence will properly refuse to regard one’s current annoying failure, and does so immediately, on first encounter. The question is, experience as a warrant for the premise that here is a hand. But then since why? What can be wrong? by the sceptic’s lights there is no warrant for ‘Here is a hand’ in the first ff Various diagnoses have recently been o ered. According to Crispin place, there is no warrant to be transmitted from the premises to the conclu- Wright, for example, the proof fails because it is, in a certain way, epistem- sion of the argument, which will therefore fail to establish the conclusion ically circular.2 Wright does not question the first premise ‘Here is a hand’, that there is an external world, at least to the sceptic’s satisfaction. but argues that perceptual experience can provide a warrant for this premise Each of these positions, though instructive, is open to objection. But for only in a conducive informational setting which includes the thesis that there my purposes their most salient feature is that they take us into complex and is an external material world broadly manifest in ordinary sense experience. controversial issues concerning the rational architecture of our most basic ff A di erent example may make this clearer: seeming to see water falling perceptual warrants – issues on which many may well have no settled view. outside my window provides warrant for the belief that it is raining now So neither diagnosis can easily explain the more or less universal reaction of only in a setting where other sources of falling water, such as a hosepipe, are intellectual dismay on first encountering Moore’s proof. Wright’s account is excluded.3 Only in this kind of setting does the content of my current in difficulty on this point simply because it is not obvious or generally experience count as information that my belief that it is raining now is true. accepted that the underlying rational architecture of the proof is what he Hence, in the context of the proof, Wright would say that Moore’s per- thinks it is.5 Moreover, how on his proposed account are we to explain the ceptual experience of seeming to see a hand provides a warrant for the immediacy of our disappointment? After all, it is certainly not immediately premise that there is indeed a hand here only in an informational setting evident that the proof has the epistemic structure Wright thinks it has. His where it is assumed that there is an external world broadly manifest in sense account can predict our reaction only on the assumption that we are in experience. Thus Moore’s warrant for the premise that here is a hand 4 J. Pryor, ‘The Sceptic and the Dogmatist’, Noûs,  (), pp. –; ‘What’s Wrong 1 G.E. Moore, ‘Proof of an External World’, repr. in his Philosophical Papers (London: With Moore’s Argument?’, Philosophical Issues,  (), pp. –. George Allen & Unwin, ), pp. –, at p. . 5 See, for instance, beside Pryor, also H. Beebee, ‘Transfer of Warrant, Begging the 2 C.J.G. Wright, ‘Facts and Certainty’, Proceedings of the British Academy,  (), pp. Question and Semantic Externalism’, The Philosophical Quarterly,  (), pp. –; –; ‘(Anti-)Sceptics Simple and Subtle: G.E. Moore and John McDowell’, Philosophy and M. Davies, ‘Epistemic Entitlement, Warrant Transmission and Easy Knowledge’, Proceedings of Phenomenological Research,  (), pp. –. the Aristotelian Society, Supp. Vol.  (), pp. –; C. Peacocke, The Realm of Reason (Oxford 3 I would like to thank an anonymous referee for this enlightening example. UP, ).

©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly ©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly  ANNALISA COLIVA THE PARADOX OF MOORE’S PROOF  some sense aware of that structure, and that our awareness functions as a who thinks that we cannot claim warrant for that belief, since the rational reason for our reaction. Pryor’s interpretation, for its part, has difficulty in stance towards it is one of agnosticism. But if one had indeed no view about explaining the near universality of the reaction. For Pryor’s account seems the existence of an external world, then if Pryor’s reconstruction of the proof to predict that the proof should seem perfectly all right to the general were correct, one could very simply acquire warrant to believe in it. For his reader, and that only an already committed sceptic, one who already doubts account predicts that if one has no view about the existence of an external the conclusion, should find it deficient.6 But surely one need not be a com- world and has an experience as of a hand in front of one, this experience mitted sceptic (who is, after all?), nor indeed have any doubt whatsoever will give one immediate warrant to believe that there is indeed a hand about its conclusion, in order to find Moore’s proof unsatisfactory.7 The where one seems to see it. Hence, in effect, one would have to give up one’s paradox is this: how is it that an argument which on the face of it is perfectly agnosticism with respect to the existence of an external world. Therefore, on sound, being a valid deduction from warranted premises, can immediately Pryor’s account of the proof, one could not for long be a sceptic, vicariously strike both sceptics and non-sceptics alike as a failure? or otherwise, about the existence of an external world. But then how could Both Pryor and Wright have lines of possible reply. Wright could attempt we account, on Pryor’s reconstruction of the proof, for the fact that the to exploit a gap between explicit theoretical awareness of the necessary proof strikes (almost) everyone as a failure? Rather it seems that even a conditions of cogent argumentation and tacit awareness of them. After all, genuine sceptic should be persuaded by it to give up his view. thinkers may, familiarly, have a tacit sense of a fallacy in a piece of reason- Secondly, a sceptic does not start off as a sceptic. Rather, he thinks that the ing without having the explicit conceptual means to identify it and charac- arguments we may produce to claim warrant for a given belief are somehow terize it correctly. Could not something broadly analogous apply to a tacit deficient. In particular, he will first find fault with Moore’s proof, and as a sense of the shortcomings of Moore’s proof? Pryor, for his part, despite the consequence will come to hold that warrant for its conclusion is not ration- fact that he claims that only someone who really has sceptical doubts and ally redeemable. But this simply restates my initial problem: why, when thus shares a sceptic’s psychological attitude,8 would find the proof unsatis- prima facie it seems perfectly in order, is the proof after all an obvious failure, factory, might rejoin that non-sceptics find the proof unsatisfactory vicari- even when it is taken – this is what is instructive about Pryor’s account – as ously, on behalf of sceptics,9 without thereby being committed to the kind of an attempted response to a kind of philosophical worry which is likely to scepticism which he has in mind. result in scepticism? But I find both these lines of reply quite unsatisfactory: the former works only by assuming that Wright’s account is indeed the right one – a matter which remains highly controversial.10 The latter, in contrast, relies on a mis- II. THE SOLUTION conception of scepticism and its dialectic. First, a sceptic need not disbelieve that there is an external world: scepticism is not the same as idealism! In order to solve the paradox, I propose to scrutinize further the content of Rather, as I shall show, a sceptic is most fruitfully thought of as someone the sceptical challenge ‘Is our belief in the existence of an external world warranted?’. For presumably how we interpret this question must be mani- 6 Or, at any rate, one who is ‘taken in’ by sceptical arguments: see Pryor, ‘What’s Wrong with Moore’s Argument?’, p. . fested in our immediate reaction to the proof, pretty much as a certain 7 I am grateful to Sebastiano Moruzzi for helping me to see this point. understanding of the question ‘Do numbers exist?’ is implicated in impatient 8 Pryor, ‘What’s Wrong With Moore’s Argument?’, pp. –. This is why he claims responses to any attempt to settle it which runs ‘Of course they do – the (p. ) that if one did not catch that ‘disease’, or if one cured oneself from it, one would be  free to accept Moore’s proof. As will become apparent, I do not think that scepticism is a actual number of books in my house is , and only an existing number disease or a psychological attitude; rather, it consists in broaching fully intelligible questions can be the actual number of my books’. and hypotheses which bring about what Wright, in ‘Warrant for Nothing (and Foundations On one natural reading of the sceptical challenge, it can plausibly be for Free)?’, Proceedings of the Aristotelian Society, Supp. Vol.  (), pp. –, at p. , calls ‘crises of intellectual conscience’, which need to be answered, if only by diagnosing that they taken as raising what I shall term a specific question: the question whether any rest on at least unargued assumptions. of our external-world beliefs is warranted by the kind of considerations we 9 I am grateful to Crispin Wright for pressing me on this. standardly take to warrant them. Crucially, however, there is also an issue 10 Pryor, ‘What’s Wrong With Moore’s Argument?’, Davies, ‘Epistemic Entitlement, Warrant Transmission and Easy Knowledge’, and Peacocke, The Realm of Reason, all dispute it, about the level of the sceptical challenge, whether what is challenged is the although in different ways. very existence of the warrants we ordinarily take ourselves to have, or whether

©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly ©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly  ANNALISA COLIVA THE PARADOX OF MOORE’S PROOF  the challenge concerns our right to claim to have them. This distinction is unimportant when warrant is conceived along traditional internalist lines, as something whose existence is inevitably salient to a sufficiently attentive III. A NEW PUZZLE thinker and which is immediately employable to redeem the justifiability of one’s beliefs. But the distinction becomes vital if warrant may be constituted One cannot leave matters there, however, since a new puzzle seems now to in real external cognitive relationships whose obtaining need not be sub- have been generated. Wright and Pryor seemed to be conducting a well jectively salient. I cannot argue the point here, but I side with those philo- conceived debate about deep epistemological issues raised by the apparent sophers who hold that the real challenge of scepticism operates at the higher lack of cogency of Moore’s proof. But if the question at hand is just the level: what we are challenged to do is not to show how it is consistent with higher-level specific question, their debate seems to be an irrelevant side- sceptical argument that we may indeed be warranted in much the ways we show, since no such profound issues are raised: what is wrong is, as I have ordinarily think we are, but rather to redeem those warrants – to make out noted, the irrelevance of the proof to the sceptical challenge. So why does it our right to claim to have them.11 So one possible content of the specific seem that their discussion does involve potentially illuminating suggestions as sceptical challenge is as follows: ‘For any ordinary empirical belief we have, to why the proof does not work? I think an explanation of this too will and take ourselves to be warranted in having, can we rationally claim a emerge from further reflection on the content of the sceptical challenge. warrant?’. This, I submit, is the most immediate reading of the sceptical The specific interpretation just reviewed takes the sceptical challenge as challenge. And it is exactly what drives our immediate impatient response to the challenge to redeem the perceptual warrants for particular empirical Moore’s proof. For if this is how we take the question, a ‘proof’ which beliefs which we routinely take ourselves to have. This is the challenge simply assumes that our normal perceptual grounds do indeed provide which is most naturally read into Cartesian forms of scepticism. By well a warrant of one such belief in particular, viz Moore’s belief that ‘Here is a known routines involving hypotheses such as that we might be dreaming hand’, simply does not address the question. The issue is exactly whether we right now, or that we might be envatted brains, an argument is developed can somehow reflectively provide a rational reassurance that such grounds that no such redemption is possible. The basic train of thought is familiar 12 do indeed warrant this ordinary specific empirical belief (or any other), that enough. If we were in such a sceptical scenario, we would not have the is, whether we can redeem the warrant we take them to supply. perceptual warrants we think we have for our specific empirical beliefs. Similarly, someone who tried to answer the question ‘Are there num- But it is a metaphysical possibility that, for example, dreams and waking bers?’ by simply pointing out that a given number is the number of a certain perceptual experiences might be indistinguishable. Hence we have no re- collection of objects would fail to address the question he has been asked. sources, based purely on what is subjectively salient, for distinguishing when For the question asks for a guarantee that the numbers we associate with we are in a sceptical scenario from when we are not, nor therefore for any given collection of objects are indeed entities which should be counten- redeeming rationally our claim to any particular warrant based on current anced by our ontology. experience. We may indeed be perceptually warranted in believing a pro- So we react impatiently to the proof because well before entering the position like ‘Here is a hand’, but we cannot rationally lay claim to the details of the rational architecture of empirical warrant, we take the sceptical warrant. So at any rate runs this broadly Cartesian genre of sceptical argu- challenge to be the higher-level specific question – a question about the ment, whose import is that all our empirical beliefs appear to be in the same redeemable adequacy of our routine grounds for specific claims about the external predicament: each of them may have the authority of veridical perceptual world; and we realize that Moore’s proof, taken as a response to that experience, but for none of them in particular do we seem to have the question, assumes precisely the point at issue. resources for redeeming that authority, for laying rational claim to it. Thus, unlike Pryor, I do not think that Cartesian scepticism starts by doubting or 11 Wright, ‘Warrant for Nothing’, and Davies, ‘Epistemic Entitlement, Warrant Trans- disbelieving that there is an external world. Rather, the conclusion that the mission and Easy Knowledge’, repeatedly stress this point. For someone sympathetic with it, externalist moves would not represent answers to scepticism but mere refusals to engage with whole class of empirical beliefs cannot be claimed to be warranted is it. It is also worth noticing that from a historical point of view, scepticism arguably was reached by showing that none of the instances of the class can be. developed in an intellectual climate where warrant was conceived along internalist lines, where the issue of the existence of warrant was also ipso facto the issue of the rational 12 See for instance, Wright, ‘Scepticism and Dreaming: Imploding the Demon’, Mind,  redeemability of it. (), pp. –.

©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly ©  The Author Journal compilation ©  The Editors of The Philosophical Quarterly  ANNALISA COLIVA THE PARADOX OF MOORE’S PROOF 

The Cartesian genre is, however, only one kind of scepticism. As Wright is to present it as a claim about the possible status of Moore’s proof when has recently emphasized (‘Warrant for Nothing’, pp. –), there is an- the issue concerns not the redemption of warrant for the conclusion that there other form of sceptical challenge, of broadly Humean descent, characterist- is an external world – the issue, I have suggested, raised by Humean ically formulable in the template of what he calls the (I)–(II)–(III) argument – scepticism under its proper higher-order interpretation – but merely the illustrated, ironically, by Moore’s proof itself, as he reads it. This consists in possibility of our having any warrant. the challenge to produce warrant for our metaphysically most general framework or ‘hinge’ beliefs, propositions such as that there are other minds, that the world has a long history, that it is amenable to inductive IV. CONCLUSION inference, and that there is an external material world at all. In contrast with the specific reading just discussed, I shall call this the general reading of the To sum up, the Cartesian and the Humean sceptical challenges are defined sceptical question. Again, however, it is important that the sceptical chal- around two different though of course closely related questions. We may lenge may be presented at two different levels: at the lower level it concerns accordingly take Moore’s proof as an attempt to respond to either. When the very existence of warrants for framework beliefs of the kind illustrated; addressed to the specific Cartesian challenge of rationally confirming our at the higher level it is the challenge to redeem the warrants – to muster the actual standards of warrant for any particular empirical belief, the proof is resources to lay rational claim to have them. simply irrelevant, since it fails even to acknowledge the Cartesian concern. So now I shall consider Moore’s proof as an argument against Humean And for what it is worth, Moore himself, 13 as well as Wittgenstein14 and a scepticism, so understood. Suppose the circumstances are such that we do whole first generation of critics,15 thought that, so understood, the proof have a perceptual warrant for the first premise that here is a hand, because stood no chance. But on the other hand there is the resurrected Humean we are in fact seeing a hand. Still, it seems evident that in order to claim to general challenge of showing how the belief that there is an external world at have perceptual warrant for the premise, we must take it that the conclusion all can be rationally redeemed per se. It is Wright’s observation that Moore’s that there is an external world holds – more specifically, that the experience proof, so far from addressing this concern, can in fact, incongruously, be we are now having, as of a hand in front of us, is indeed produced by used to supply a template for its very formulation (thus giving rise to a whole normal sensory interaction with a world populated by physical objects. One new chapter of Moore studies). could not rationally lay claim to warrant for the first premise that here is a Whether or not Wright is correct about the actual epistemic architecture hand, yet profess openmindedness on the latter score. But if this is right, of basic empirical warrant, my suggestion is that he is right about why the then Moore’s proof fails as a response to Humean scepticism too, because in proof can offer no answer to the Humean sceptical challenge, when pro- laying claim to warrant for its premise, one would presuppose that one was perly construed at the higher order, as centred on the issue of rational already in position to lay claim to warrant for its conclusion. My suggestion, redeemability. To repeat: one cannot rationally both profess openminded- then, is that when the proof is taken as a response to higher-level Humean ness about the existence of an external world and claim that one’s current scepticism, it is plausible that it is epistemically circular for pretty much the sense experience as of a hand warrants the belief that here is a hand. reasons which Wright urged: to lay claim to its first premise that here is a 13 Most current interpretations consider the proof as an anti-sceptical argument, but Moore hand, one must presuppose the warrantedness of its conclusion that there is himself thought of it as anti-idealist: see, for instance, G. E. Moore, ‘A Reply to my Critics’, in an external world. P.A. Schilpp (ed.), The Philosophy of G.E. Moore (Northwestern UP, ), pp. –. Moore stressed that it could not be read as an anti-sceptical argument because for this he should at This is not necessarily the same as saying that in the Wright–Pryor least have proved that he was not dreaming of having a hand in front of him. For a discussion dispute the latter comes off worse. Pryor might be right that when no such of how, despite his intentions, the proof can be read as an anti-sceptical argument, see my higher-level Humean scepticism is at issue, one need not presuppose the ‘Proof of an External World: Transmission-Failure, Begging the Question or Dialectical Ineffectiveness? Moore, Wright and Pryor’, in A. Coliva and E. Picardi (eds), Wittgenstein Today warrantedness of the conclusion that there is an external world in order for (Padova: Il Poligrafo, ), pp. –. one’s perceptual experience to confer a warrant for the premise that here is 14 ’If you do know that here is a hand, we’ll grant you all the rest’: Wittgenstein, On Certainty a hand. However, this is not the place to try to explore further any possible (Oxford: Blackwell, ), §. 15 Of special relevance among them are A. Ambrose, ‘Moore’s Proof of an External World’, reconciliation of the dispute between Pryor and Wright. All I emphasize for M. Lazerowitz, ‘Moore’s Paradox’, and M. Malcolm, ‘Moore and Ordinary Language’, all in the purposes of this paper is that the only prospect I see for Pryor’s position Schilpp (ed.), The Philosophy of G.E. Moore, pp. –, –, –.

Moreover, contrary to Pryor’s diagnosis of why the proof fails as a response to scepticism, when each of the scepticisms is taken as I have argued, as posing a higher-order challenge concerning the rational redeemability of certain warrants, neither needs to assume that we have no warrant for the premise that here is a hand and so no warrant to transmit to the conclusion that there is an external world. We may indeed have warrant, yet be unable to redeem it, exactly as both forms of scepticism in effect argue. So when taken as a response to Cartesian scepticism, Moore’s proof is simply irrelevant; taken as a response to Humean scepticism, it is epistemically circular. In either case it is a failure. Yet, I have also suggested, a satisfying account of our immediate reaction to it of irritation and dismay is available if the Cartesian reading is what we tacitly understand the issue to involve. I think, first, that this is sociologically plausible. Cartesian doubt, with its dreams and demons, was famously focused on knowledge of the external world, while Humean scepticism was developed primarily in connection with induction – indeed, the extension of the template to apply to scepticisms about the material world, other minds and the past may even be as recent as Wright’s work. Secondly, if this is granted, then the key point in dissolving the paradox of Moore’s proof is that the kind of question-begging it exemplifies when read as an anti-Cartesian argument is indeed obvious, consisting in its failure even to acknowledge the specific sceptical challenge. The more intricate kind of question-begging involved if it is read as an anti- Humean argument is less obvious: spotting this requires both appreciating the level of the sceptical challenge, that it concerns the right to claim warrant for belief in an external material world, and registering that using Moore’s inference to conclude that we do indeed have the right is, arguably, stymied by the kind of epistemic circularity I have discussed. The latter is quite a subtle matter. Small surprise that it took about forty-five years of reflection on the proof to draw it into focus! Finally and connectedly, it is only if we take Moore’s proof as an attempted response to Cartesian scepticism that we can explain the kind of reaction we display on first encounter with it: we perceive immediately not just that the proof is somehow defective, perhaps because it exemplifies the intricate form of epistemic circularity exposed by Wright’s reading of it, but that it is totally beside the point: it simply fails even to address the kind of sceptical challenge it is meant to answer.16 Università di Modena e Reggio Emilia

16 I would like to thank Sebastiano Moruzzi, Crispin Wright and Elia Zardini of the Arché Research Centre at St Andrews for helpful comments on previous versions of this paper, as well as an anonymous referee.