PH29A Cartesian and later scepticism

Pojman, ch. 2.  Pojman (ed.) Part II.

The Stanford Encyclopedia has a fairly sophisticated discussion of Descartes' epistemology. For another sketch of matters relating to Descartes, see Wolf's lecture outline or a lecture on scepticism more generally at Washington University in St Louis.  An interesting discussion of mediaeval aspects of Descartes can be found in a paper at Macquarrie. For texts of Descartes' Meditations you can check the following: one traditional English translation of the Meditations; the original Latin text; or the authorised French version. All three languages are available at this site.

There are a number of other papers that might be helpful, though they are not pitched at an introductory level: 

Descartes' situation - realises he has picked up a lot of falsehoods, so needs to make a clean break.  Start again and avoid any possible error.  There are three main steps in his first meditation:

  1. reject perception because it can be in error
  2. might I be dreaming?
  3. malicious demon

The appeal to dreaming is typical of direct argument against the possibility of knowledge.  How can you know if you cannot rule out the possibility that you might be dreaming?  At least two variant suggestions: you might be dreaming now, you might always be dreaming (or in a state that bears the same relation to reality as ordinary dreams do to what we take as awake).  While these can be taken as alternative hypotheses, I call this a direct attack since the main line is that you do not know p if it there is any epistemic possibility that not-p.  

The demon is typical of sceptical hypotheses. Note modern variant: a brain in a vat.

In Descartes these manoeuvres are methodological. They are used to clear away confusions but in the hope of being overcome.

"The Sceptic" is portrayed as holding scepticism as a substantive matter: we do not know anything.  It is important that the sceptic is using know like the rest of us; the argument is that on our own terms, using standards we actually endorse, we are mistaken in thinking we know the things we claim to know. 

Problem is that Descartes' positive programme doesn't work, nor has anyone else's clearly done so. Possible responses:

  1. ignore problem
  2. admit defeat but make it look acceptable — Hume, some moderns
  3. attempt a direct answer to sceptical arguments: they make sense but can be rebutted
  4. attempt an indirect answer: they don't really make sense or undermine themselves.

Not always clear how to classify a response — e.g. Nozick says I don't know that I'm not a brain in a vat, but so what? I do know that I'm living in Barbados. (And I know that if I'm living in Barbados I'm not a brain in a vat.)  So we have a partial concession to the sceptic together with an attempt to reject his most damning conclusions — but that requires us to abandon what seems like a very plausible principle (that if you know p and you know that p entails q then you know q).

There are at least two distinct ways of ignoring scepticism.  One is to acknowledge a problem but think that one has other things to do than to answer it.  More interesting is the view that confronting scepticism is a waste of time.  (Herman J. Saatkamp, Jr. has a chapter of a book commenting on Rorty's general position which includes something like this: he gives the following quotation from Rorty: "One of the disagreements between Haack and myself is that she thinks that epistemology is a natural and obvious topic of reflection, whereas I think that it survives nowadays only because some philosophy professors still think it important to take epistemic scepticism seriously — a spiritual exercise I find profitless.")

Hume's line. Sceptical positions arise out of reason but cannot be controverted by it. Human nature makes us ignore what this use of pure reason tells us. Popper said Hume is mainly right; let us relinquish any concern for knowledge of that sort, the best we can do is.... Michael Williams argues that scepticism is not, as Hume thought, "natural" but rather incorporates a body of contentious epistemology. We might therefore be able to keep it at bay by rejecting this theoretical baggage.

Direct rebuttals: Moore, Malcolm, Wittgenstein.  (Traditional theories of knowledge.)

Moore put up his hands as proof that there are two hands, and thus that there are objects in the external world.  Malcolm and Wittgenstein appeal to certain things that we take for granted, that "stand firm" and on whose basis we conduct our life.  These are not derived from any evidence; we would think someone mad who doubted them.  But while we may operate in these ways, it is difficult to see what is wrong with addressing the sceptic's challenge when it is raised.  Maybe I don't get p as an inference from some other evidence, but when you challenge p, can I construe my situation so that there is positive support for p? Michael Williams' line here would be that no doubt we can on a piecemeal basis, but that we ought not to accept a challenge that requires us to justify the totality of our putative knowledge of the world in one go. 

Some traditional theories of knowledge have been designed to avoid sceptical possibilities.  With what success, we shall see later.

Indirect responses: 

  1. Carnapian verificationalist approach, in which answers are possible within a framework but no answer between frameworks. Carnap claimed we choose whether to use a "thing-language"; if we do, we can answer questions about whether this thing in front of us has certain properties or not, and these claims entail that there are such things, but we cannot make sense of a global question about whether we are right to adopt the framework of "thing language" and so whether there are any things at all, understood as a traditional epistemologist wishes to construe it. (There is an article by Weisberg on the web about Carnap's approach to scepticism.) More recently anti-realist considerations about meaning (elaborated in particular by Dummett) have been invoked to similarly cast doubt on the intelligibility of global scepticism. Sceptical hypotheses are designed to be "verification-transcendent" — they preserve everything we currently have to go on, but they suppose that the reality we are confronting is radically different from what we take it to be. Anti-realism rejects the possibility that we can understand claims that are in this way verification-transcendent. While the arguments get very complex, it shares with earlier forms of verificationalism the problem that prima facie we do appear to understand some verification-transcendent claims, such as sceptical hypotheses, and this appearance is somewhat more plausible than the outré theory of meaning being offered to us.
  2. Transcendental arguments appeal to conditions for the possibility of our experience or our concepts being the way they are. If something is the only way we could have got what we have, then we cannot seriously question it. So if the only way we could have experiences that appear to us to be of objects in an external world is by there being such a world, then it is idle for a sceptic to allow us the appearance but to question the reality it purports to reveal. The main problems here are that it is often difficult to establish that something is the only way we could have what we have and that the conclusion is usually a matter of what we have to take to be the case — but then the sceptic remains to be convinced that it really is the case. (There is an article on the web by Thomas Grundmann and Catrin Misselhorn, Transcendental Arguments and Realism (.pdf format) in this general area.)

Dreaming

The crucial claim here is that there is no mark by which we can tell from the inside whether we are dreaming or awake. Whatever you might invoke (that p), it is intelligible that you might be dreaming that p. (Bernard Williams [Appendix 3 to Descartes, Penguin, 1978] attributes this argument to Margaret Macdonald's 1953 paper, but an instance of it is used by Hobbes in his reply to Descartes' sixth meditation: "I want to know whether it is certain that a man who dreams of wondering whether he is dreaming or not may not dream that his dream fits together with a long series of past events" — the criterion Descartes had offered for telling waking from dreaming.) 

Bernard Williams comments that being able to tell whether S seems to entail both (a) one can tell that S when S and (b) one can tell that not-S when not-S. We may think that (a) and (b) must always go together, but his resolution of the apparent oddity of the crucial claim here is that they don't necessarily. We can find cases where (b) holds but (a) doesn't. I can tell I'm not dead when I'm not dead, but I can't tell I'm dead when I'm dead. In place of dead he also offers in a dreamless sleep or (idealised) anoxia (lack of oxygen that leads to overconfidence, including discounting the relevance of evidential symptoms of anoxia). Williams goes on to review different sorts of reason why a person cannot tell that S when S (the person doesn't exist, he cannot do anything, he cannot tell anything, he cannot rationally tell anything) and contrasts it with what he takes Descartes to have thought of dreaming, viz. a situation in which a person can rationally decide but his experience is such as to lead to the wrong conclusion (illusions, forgeries). Of these he queries whether the person really cannot tell that S when S — we do "see through" apparently bent sticks and forgeries. These situations are complex in a way dreaming is not in that there is something that arouses a doubt and there is something one can do to settle it.

He goes on to say that Descartes is wrong to treat dreaming like illusions; it goes rather with cases in which a person cannot tell anything. Here the fact that, when S, a person cannot tell anything shows nothing about her ability to tell not-S when not-S. So we cannot tell we are dreaming when we are dreaming, though we can have false beliefs, but we may, and can, tell we are awake when we are awake.

Two further comments after the class.  One is that awake/dreaming is not quite S/not-S; Williams himself offers in a dreamless sleep and dead, both of which a further contrasts and not equivalent to either of our starting points.  (I don't think this vitiates the argument, but it is worth being clear about it.)  The second is a response to the point made in the class that one can realise one is dreaming when one is dreaming.  On reflection I think Williams' response would be that this is a case of some other mental activity accompanying dreaming; one is half-awake and realises that a sequence of experiences belong to a dream (and one may then decide to get back into them).  I think he wants to insist that whatever thought or belief occurs to you within a dream, it cannot be regarded as telling you anything, even if it turns out to be true.  You may then dream that your are dreaming but that isn't a way of telling that you are dreaming.  It is as unreliable as dreaming that you are in the seventh heaven.  This is how he thinks we conceive of dreaming, as an actvity that cannot tell us anything.  (I must admit that many  people do think dreams tell them things.  Williams would have to say they are mistaken; they haven't got a proper conception of dreaming.  Maybe Descartes was one of them, according to Williams' diagnosis of his conception of dreaming as akin to perceptual illusion.)

Other minds

Pojman, ch. 14; Pojman (ed.), Part IX.

It is a characteristic of Descartes attempted response to sceptical doubt that he should bequeath a specific sceptical problem about the existence of other minds. I can defend the truth of "I am". "I exist", whenever I utter it or conceive it in my mind, against the strongest sceptical hypothesis (Meditation 2). But in thus holding on to my own existence as a thinking thing I have put in question that of all others, since all I have to go on in their case is their observed behaviour. On what grounds can I justify the further supposition that their behaviour is accompanied by what I find in my own case, conscious sensations and thoughts?

Pojman reviews the main answers that have been offered:

  1. arguments from analogy

    A common criticism is that the sample size is minimal.  I'm not sure this is fatal: when we have the right background assumptions a sample of one  can give us reliable answers.  Those sorts of assumption might not be permissible if the argument were trying to tell us where we get our everyday beliefs from, but if it is a matter of showing how those beliefs are defensible perhaps they are alright.  I think this might answer the Plantinga argument that Pojman employs: by analogy I could infer that the only pains are my pains.  It seems that this argument supposes the analogical argument simply to be of the form this A is B, that A is B, ... so all A are B.  My point is that reliable uses of a sample of one involve a much more structured set of assumptions.  We need, of course, to delve much deeper into what those assumptions amount to.  Michael Williams suggests that they include the case of something being a natural kind: one DNA molecule tells us a lot about DNA whereas one chair may not reveal much about chairs.  But even here, it is only when we focus on properties essential to the kind that we can be so confident — that this molecule comes from a cat doesn't permit us to assume that all DNAmolecules come from cats.

  2. behaviourist construals of mental life

    Obviously if having a mind is somehow equated with behaving in certain observable ways, then there is no special problem about observing the minds of other human beings.  The problem is with the equation.  Crude equations break down over our ability to pretend, more sophisticated ones don't quite give us enough to settle the sceptical doubt: we may need observable behaviour to learn the language for mental experiences, but does that prove that your total experience of pain, say, is anything like mine?

  3. brute assumption intelligible on Darwinian lines

    The issue here may be the same as my comment above on Wittgenstein: we may operate with non-inferential beliefs about other minds, but when challenged, can we substantiate them?  Natural selection aims at fecundity, but what's that got to do with truth? 

Return to home page.

Ed Brandon, last revised 26 September 2001.