Two Skeptical Arguments
I’ve been claiming that there are some really powerful skeptical arguments (on the show and in response to Ken’s previous post). I have also been claiming that one aspect of their force is that they do not depend on setting the standards for knowledge very high. Here are two such arguments.
1. Hume’s argument.
The first is inspired by David Hume. The argument begins with the assumption that our beliefs about the external world are at least partly based on how things appear. For example, I believe that I am presently seated at my desk at least partly because that is the way things visually appear to me. But that can’t be the whole story, the argument continues. I must also be assuming, at least implicitly, that the way things appear is a good indication of the way things really are. If I were not relying on that assumption, Hume argues, then the fact that things appear to me a certain way would not be a reason to think that they are that way. But now how am I to justify this assumption about the reliability of appearances? How can I know that the way things appear is a good indication of the way things really are? According to Hume, there is no way to justify that assumption. For example, suppose I were to rely on appearances, reasoning that, as far as I can tell, the way things appear to me appear to be a reliable indication of the way things really are. This, of course, would be to argue in a circle, taking for granted the very thing at issue. Here is Hume’s argument put more formally.
(H)
1. All my beliefs about the external world depend for their evidence on both a) the way things appear to me, and b) an assumption that the way things appear to me is a reliable indication of the way things really are.
2. But the assumption in question can’t be justified.
Therefore,
3. All my beliefs about the external world depend for their evidence on an unjustifiable assumption. (1, 2)
4. Beliefs that depend for their evidence on an unjustifiable assumption do not count as knowledge.
Therefore,
5. None of my beliefs about the external world count as knowledge. I don’t know anything about the external world. (3,4)
Clearly, a linchpin of this argument is premise (2): that an assumption regarding the reliability of appearances cannot be justified. In support of premise (2), Hume considers various possibilities for justifying the assumption in question. One consideration that Hume emphasizes is that the assumption is itself a contingent claim about the external world. That is, the assumption claims that sensory appearances are, as a matter of contingent fact, related to the way things are in a particular way. This suggests that the assumption can be justified, if at all, only in the way that contingent claims about the external world are justified in general—i.e. by relying on the way things appear! But this, of course, would be to argue in a circle, taking for granted the very thing at issue. Here again is the reasoning in support of (2).
(H2)
1. All my beliefs about the external world depend for their evidence on both a) the way things appear to me, and b) an assumption that the way things appear to me is a reliable indication of the way things really are.
2. The assumption in question is itself a belief about the external world.
Therefore,
3. The assumption depends on itself for its evidence. (1, 2)
4. Beliefs that that depend on themselves for their evidence can’t be justified.
Therefore,
5. The assumption in question can’t be justified. (3, 4)
A natural thought is that the assumption that appearances are a reliable guide to reality can be justified in some other way, perhaps by some sort of a priori reflection that proceeds independently of appearances. But Hume thinks that this line of reasoning is a dead end. This is because the assumption in question makes a contingent claim about the way things are—it is a matter of contingent fact, and not a matter of necessity, that appearances do or do not reflect the way things really are. But that sort of fact cannot be known through a priori reflection. In short, a priori reflection gives us knowledge of necessary truths rather than contingent truths.
2. Descartes’s argument.
The second skeptical argument is inspired by Descartes’s Meditation One, and in particular by Barry Stroud’s reading of that meditation. To understand the argument, consider the claim that one sees a goldfinch in the garden, based on one’s observation that the bird is of a particular size and color, and with a tail of a particular shape. Suppose now that a friend challenges one’s claim to know, pointing out that woodpeckers also are of that size and color, and also have tails with that shape. As Stroud points out, this seems to be a legitimate challenge to one’s claim to know that the bird is a goldfinch. More generally, if one’s evidence for one’s belief that the bird is a goldfinch is consistent with the possibility that it is in fact a woodpecker, then one does not know on the basis of that evidence that it is a goldfinch. Based on this sort of reasoning, the skeptic proposes the following plausible principle:
1. A person knows that p on the basis of evidence E, only if E rules out alternative possibilities to p.
Further support for this sort of principle comes from reflection on scientific enquiry. Suppose that there are several competing hypotheses for explaining some phenomenon, and suppose that these various hypotheses are “live” in the sense that current evidence does not rule them out as possibilities. It would seem that one cannot know that one of the hypotheses is true until further evidence rules out the remaining ones. Again, principle (1) above looks plausible.
The second step in the skeptical argument is to point out that there are various possibilities that are inconsistent with what we claim to know about the external world. For example, it is possible that things appear to me visually just as they do now, but that I am actually lying in my bed asleep rather than sitting at my desk awake. It is possible that things appear to Descartes’s just as they do, but that he is actually the victim of an evil demon, a disembodied spirit who only dreams that he inhabits a material world and is presently seated by the fire. To be clear, it is no part of the skeptical argument that such alternative possibilities are true, or even that they are somewhat likely. The point is only that they are possibilities, and so undermine our knowledge if our evidence does not rule them out.
The third step in the skeptical argument is to claim that our evidence does not in fact rule these possibilities out. The gist of the present claim is something like this: These possibilities are consistent with all the evidence that we have or could have at our disposal. Even if, practically speaking, we don’t usually give such possibilities a thought, upon reflection we have no evidence available to us that counts against them, and in favor of our preferred beliefs.
If we put these three claims together we have the materials for a powerful skeptical argument. Here is the argument stated more formally.
(D)
1. A person knows that p on the basis of evidence E, only if E rules out alternative possibilities to p. (Principle 1 from above.)
2. It is a possibility that I am not sitting at my desk awake, but merely dreaming that I am.
Therefore,
3. I know that I am sitting at my desk only if my evidence rules out the possibility that I am merely dreaming. (1, 2)
4. But my evidence does not rule out this possibility.
Therefore,
5. I do not know that I am sitting at my desk. (3, 4)
And of course the skeptical argument is supposed to generalize. That is, it is supposed to apply to beliefs about the external world in general. We therefore have:
6. The same line of reasoning can be brought to bear against any belief about the external world.
Therefore,
7. No one knows anything about the external world. (5, 6)
One way to understand the notion of “ruling out” a possibility is as follows: A body of evidence E rules out a possibility q if and only if E supports not-q in a non-circular way. Here we can understand support as a semantic notion: Evidence E supports propositions p, in the relevant sense, just in case E entails p or E makes p probable. Putting these ideas together, we get the following interpretation of premise (4) of argument (D).
4a. My evidence for my belief that I am sitting at my desk neither entails nor makes probable (in a non-circular way) the proposition that I am not dreaming.
Why might one accept premise (4a)? One reason for accepting (4a) is the considerations put forward by Hume’s argument above. That is, one might think that my evidence for believing that I am sitting at my desk is the way things appear to me, together with my assumption that the way things appear to me is a reliable indication of the way things are. But as Hume’s reasoning shows, there is no non-circular way to justify the assumption in question, and therefore no good evidence for either that assumption or further beliefs that are based on it. In particular, my evidence cannot entail or even make probable (in a non-circular way) the proposition that I am not dreaming. Insofar as this is the reasoning behind (4a), argument (D) is parasitic on argument (H).
There is, however, another way to understand the notion of evidence ruling out alternative possibilities. On this understanding, a body of evidence E rules out alternative possibilities to p just in case E discriminates the state of affairs represented by p from alternative states of affairs. For example, hearing my wife coming in the door from work, my auditory experience rules out the possibility that it is my children coming home from school or a burglar coming in through a window. In effect, I have the capacity to “tell the difference,” so to speak, and this is what allows me to know that it is my wife who has just come in the house. On this understanding of “ruling out”, it does seem plausible that my evidence must rule out alternative possibilities in order to ground knowledge. For example, how could I know that my wife has just come home, on the basis of hearing her come through the door, if I could not discriminate that state of affairs from my daughter’s coming through the door? Moreover, premise (4) of argument (D) becomes plausible on this understanding of “ruling out.” We now have
4b. My evidence does not discriminate my sitting at my desk from my merely dreaming that I am sitting at my desk.
One might think that this claim is obviously right. To be clear– I assume that the skeptical argument must be wrong somewhere. My point here is that it isn’t obvious where, or that the argument is invoking some very high standard for knowledge.