Williams and epistemological realism

As we saw in the first session, Descartes’ second and third methods of doubt explicitly (and the first method implicitly seems to although, on reflection, does not) work by articulating a ringer for everyday empirical beliefs and perceptions which we cannot distinguish from the everyday cases. Our experiences would seem the same whether we were awake and experiencing the external world or, either, dreaming or subject to a daemon / a brain in a vat. Once they are in play, it is difficult to rebut the sceptical argument because no feature within our experience seems capable of determining whether the pre-sceptical conception or the ringer applies. But it may be possible to block this argument before it gets started. That, at least, is what Michael Williams attempts. He challenges the ground rules of the sceptical argument. This he offers the following outline of the dreaming scepticism argument:

if we are to know anything about the world, we must sometimes know that we are not dreaming; but we can never know that we are not dreaming; therefore we never know anything about the world.

The first premiss seems hard to deny. And as for the second, if we are ever to know that we are not dreaming, there must be a test that we can, at least in some circumstances, apply to determine whether we are dreaming or not. But now suppose that there were such a test—any test, not just one conforming to foundationalist preconceptions— it will be of use only if we know that we have really applied it and have not just dreamed that we have applied it. That is, it will be of use only if we already have some way of determining that we are not dreaming, which leads us into a regress... [Williams 1988: 436]

But he goes on to offer the following diagnosis of how the argument works: crucially the hidden premiss which is a form of foundationalism.

We can begin with a look at the first premiss, which can be read two ways. On one reading, it is a truism: at least, something I have no wish to dispute. We can concede that the claim that we sometimes have knowledge of the world logically implies that we sometimes know we are not dreaming. Like other logical points, this is epistemologically neutral. It is entirely compatible with our holding that, since we often do know things about the world, we often know that we are not dreaming: we know that we are not dreaming in virtue of what we know about the world, and in this sense there will be tests for whether or not we are dreaming, though not necessarily any single procedure that applies in all situations.

But there is another way to read the first premiss: this is to take it to require that, if we are to know anything about the world, we must be capable of knowing that we are not dreaming: that is, of knowing this in some way that is independent of all knowledge of the world. On this reading, premiss one certainly promises to be useful to the sceptic, but only because it introduces a general and intrinsic dependence of knowledge of the world on whatever knowledge we can have whether or not we know we are dreaming. So premiss one is either trivial and useless or useful but just another way of insinuating a foundationalist constraint on knowledge of the world. Only by oscillating between the two readings can we sustain the illusion of deducing scepticism from a triviality.

Now, it will be said that scepticism is not meant to follow from premiss one alone, but only from premiss one in conjunction with the claim that there could not be a test, foundationalist or otherwise, for determining whether or not we are dreaming. But unless we read premiss one in the second way, the way that presupposes foundationalism, the argument for premiss two will fail. If the dependence of knowledge of the world on knowledge that we are not dreaming is understood in the first, innocuous way, we have no reason for conceding that there could be no test for determining whether or not we are not dreaming. All the argument for premiss two shows, then, is that there is no way of knowing that one is not dreaming that is independent of all knowledge of the world. But this conclusion poses no threat to knowledge of the world unless it is presupposed that such knowledge, by its very nature, stands in need of grounding in some more primitive stratum of knowledge. The argument for premiss two shows that there can be no purely experiential test for determining whether or not we are dreaming, and the epistemological significance of this conclusion derives entirely from the thought that knowledge of the world naturally requires some kind of grounding in experience. Once again, the dreaming argument shows that foundationalist ambitions are likely to be disappointed, but gives no independent reason for entertaining them in the first place. [Williams 1988: 437 italics added]

Elsewhere he summarises this move in in slightly different language (which may be clearer):

In effect, what the argument for [scepticism] ... really shows is that there is no way of knowing that we are not dreaming that is independent of all knowledge of the world: there is no purely experiential test by which to exclude the dream possibility. But this conclusion poses no threat to knowledge of the world unless we have already been given reason to think that such knowledge, by its very nature, always requires grounding in some more primitive stratum of knowledge. The argument for there being no test for determining whether or not we are dreaming turns out to be another way of making the point that knowledge of the world cannot be given a ground in experiential knowledge, which is not a step on the road to scepticism unless it has been established that knowledge of the world stands or falls with the possibility of giving it such a grounding. Once again we have an argument that shows that foundationalist ambitions are likely to be disappointed, but gives no particular reason for entertaining them in the first place. [Williams 1996: 87]

So Williams aims to show that the argument for dreaming scepticism can be blocked by showing that it depends on an 'unnatural' assumption that knowledge of the world depends on a substratum of knowledge of experience. Only if the latter can be used to show that there is no general problem with the former are we justified in our everyday beliefs and that is just what scepticism goes on to question (via ringers such as dreaming and the brain in the vat). But Williams argues that that is just an assumption. If it leads to scepticism then so much the worse for that assumption.

He calls the assumption that drives the sceptical argument ‘epistemological realism’. It is the idea that knowledge of the world is a natural kind, a uniform totality, which can be questioned or justified as a whole. That idea is certainly present in Descartes' discussion but it seems, there, to be merely a convenient way of doing the sceptical job quickly. Williams thinks it is more significant than that and actually underpins the sceptical argument.

Is its diagnosis sufficient to dissolve the sceptical threat?

Essential reading

- Williams, M. (1988) 'Epistemological realism and the basis of scepticism' Mind 97 (This is an earlier version of Williams' thoughts and you may find it clearer.)
- Williams, M. (2010) 'Epistemological realism' in Sosa et al (eds) Epistemology: an anthology, Oxford Blackwell

Try to decide what is the important part of the reading and what is less so. The key idea to get clear on is what he means by 'epistemological realism' and what sort of thing he aims to undermine. The first half aims to show how scepticism is driven by the quest to support or undermine knowledge as a general kind, all at once. The second half begins to explain how Williams supports a kind of contextualist view of justification.