A "natural belief" is a conviction that is so firmly embedded in us that no amount of skeptical reflection or philosophical reasoning can dislodge it. If there are any natural beliefs at all, then the conviction that we inhabit a material world is surely one of them. And some philosophers -- including Hume, Reid, Peirce (pictured here), Russell, and Moore -- have hoped to exploit this fact in a response to the skeptic. Note that it will not do simply to point out that our commitment to an external world is a natural belief. The skeptic will simply say in response: "That only shows that you cannot help believing in an external world. It does not show that you are justified in so believing, or that your belief is rational" He might liken our natural belief in an external world to the acrophobic's fear of heights. It may be unshakable; but that does not make it reasonable.

One way to run the argument is to insist on the following sort of principle as a normative rule governing the formation of our beliefs:

If this is right, then it is possible to be justified in believing that there is an external world without being able to argue for this proposition to the skeptic's satisfaction. One's justification would consist in the observation that the belief is natural, together with the claim that there is no positive reason to doubt it. (The mere conceivability of bizarre skeptical scenarios does not constitute positive reason to doubt.)

On this view, the difference between the acrophobic's fear of heights and our belief in an external world is precisely that the acrophobic has abundant evidence that high places are not always dangerous, and yet he persists in believing that they are in spite of this counterevidence; whereas we have no analogous evidence against our commitment to an external world.

The main challenges for this approach are to defend the norm (1), and then to say what exactly constitutes "positive reason to doubt" a proposition. Thus the skeptic may well ask, "Why isn't the fact that I cannot rule out the possibility that I am now dreaming a positive reason to doubt that I am presently awake?"

A closely related strategy can be found both in Reid and in Moore. It consists in the following line of thought.

The skeptic offers a very clever argument for a conclusion that is wildly at odds with ordinary common sense. That argument must have premises, and since it's a philosophical argument it probably has some philosophical premises: premises that ordinary people never consider, much less endorse, but which the philosopher endorses as a result of what he calls a "rational" insight. The skeptic takes his argument to establish that common sense is mistaken. But it is much more plausible and natural to take it as a refutation of the philosopher's philosophical assumptions. What the argument really shows is that philosophy and common sense are in conflict. By itself it does not show which to prefer. So it is open to us as defenders of common sense simply to say: "So much the worse for your philosophy! What you call an argument for a skeptical conclusion, I call a refutation of your philosophical assumptions!"

There is no doubt that this is a logically available strategy. Any valid argument can be regarded either as a proof its conclusion or as a refutation of one of its premises. However the strategy rings hollow unless its proponent can point to the dubious philosophical assumption that the skeptic is making. If he can't, then the suspicion arises that the skeptic has used no controversial philosophical principles, but has shown instead that common sense undermines itself. (That is certainly what Descartes' attempts to show in Meditation 1.) And if this is the skeptic's strategy, it is no easy matter to "turn the tables" in this way.

Russell concedes to the skeptic that we cannot be perfectly certain that an external world exists. But he maintains nonetheless that this is the only reasonable hypothesis. His idea is that among all of the various accounts that might be offered for the character of our experience (i.e., for the patterns we perceive in our sense data), the external world hypothesis is by far the simplest and therefore the most credible. We discussed this response at length in class, so I won't go through it here. The argument in schematic form runs as follows:

My sense data are thus-and-so.

The best explanation for this implies that I inhabit a world full of more or less permanent material objects that affect my senses in regular ways.

Therefore, by IBE, I inhabit a material world, etc.

Obviously the chief difficulty here -- apart from the general problem of defending IBE as a mode of inference -- is to defend the second premise. Russell cagily shifts in the course of his discussion from a general consideration of skeptical scenarios to a context in which the only rival to common sense is the solipsistic hypothesis. This is the view that the only things that exist are my mind and its sense data, which therefore have no external causes at all. But of course it is one thing to show that this hypothesis makes for a lousy explanation of the detailed character of my experience, quite another to show that, for example, the Evil Demon hypothesis will not do as an explanation. One interesting topic in this area would be a principled defense of the external world hypothesis against the Evil Demon hypothesis on broadly Russellian grounds.

G. E. Moore's celebrated "proof" of an external world sounds like a joke, but its intent is perfectly serious. After a great deal of stage-setting, all of which is interesting, Moore produces a very simple argument:

Here is one hand, here is another.

Therefore, at least two external objects exist.

He argues first that the argument is not literally circular: the conclusion is different from the premise, so you cannot object to it on the ground that it literally assumes what is to be shown.

Moore then points out that the premise clearly does imply the conclusion. An external object, according to Moore, is an object that can exist without our perceiving it. That, so far, is just a definition. Moreover we can agree that our concept of a hand applies only to objects that fit this description. If something looks rather hand shaped, but turns out to be the sort of thing that can only exist when someone is experiencing it, then we must conclude that it was not a hand at all, but rather a hand-image or something of the sort. (This implies that the images you confront in dreams, no matter how hand-like they may seem at the time, are not hands; so if only that sort of thing exists, then there are no external objects at all.) This is supposed to be a point about our concepts, and it is hard not to accept it. But this point guarantees that the argument is valid. As a matter of necessity, if the premise is true and there really are genuine hands, then there really are at least two external things.

Moore then argues -- and this is the tricky part -- that the only further requirement on a cogent argument is that the premise be known to be true. If this is right, then anyone who denies that Moore has in fact proved the existence of external things must deny that he knows the premise, "Here is one hand, here is another". Moore's only response to this challenge is as follows:

How absurd it would be to suggest that I did not know it, but only believed it, and perhaps that it was not the case! You might as well suggest that I do not know that I am now standing up and talking -- that perhaps after all I am not, and that it's not quite certain that I am! ("Proof", p. 145)

Moore suggests that anyone who is puzzled by these remarks must suppose that I cannot reasonably claim to know something unless I can prove it (i.e., give convincing arguments for it to someone who doubts it). The skeptic certainly does assume something along these lines. But Moore thinks that this is a mistake. He writes:

How am I to prove now that "Here's one hand, here's another". I do not believe that I can do it. In order to do it, I should need to prove ... as Descartes pointed out, that I am not now dreaming. But how can I prove that I am not? I have, no doubt, conclusive reasons for asserting that I am not now dreaming. I have conclusive evidence that I am awake; but that is a very different thing from being able to prove it. I could not tell you what all my evidence is; and I should require to do this at least, in order to give you a proof. (p.148)

Later on the same page Moore explicitly remarks: "I can know things which I cannot prove, and among things which I certainly did know, even if ... I could not prove them, were the premises of my two proofs."

It would be very interesting for someone to write a paper about this part of Moore's discussion. If knowledge and proof can come apart in this way, what is it to know something on the basis of evidence? What is it to "possess" evidence that one cannot cite? These are very difficult questions.

Bouwsma's reply to the Cartesian skeptic defies simple summary. (It is based on some ideas of Wittgenstein [see picture]; but his work is too difficult for us to discuss.) The rough idea is surely this. If the Evil Demon hypothesis is to make sense -- if it is to be an intelligible alternative to the real world hypothesis -- then we must be able to make sense of the notion of a perfectly undetectable illusion. We must be able to suppose that there is a real difference between the claim: "There is a rabbit in front of me" and the alternative claim: "Every conceivable test that I can perform, now and in the future, would support the claim that there is a rabbit in front of me". But Bouwsma doubts that there is any difference. Very crudely, his view might be this: we learn the words of our language by associating with them various test or verification procedures that we can apply in ordinary experience. Someone who understands the word "rabbit" as a word of English has learned to take certain experiences and test results as licensing the assertion of the statement "This is a rabbit". This conception of the meanings of words implies that it makes no sense to say: "This passes every test that I can imagine for counting as an F, but it is not an F". But this is just what we must assume in describing the Evil Demon hypothesis. For in that case the hypothesis is that the "objects" of our experience pass every conceivable test for being real flowers, people, rocks, and so on, and yet are not real instances of these kinds. Bouwsma's reply is that this possibility makes no sense, and so we don't need a substantive "scientific" argument of the sort Russell, e.g., seeks to provide in order to rule it out.