Wednesday, 7 March 2012

McDowell's second Edgington Lecture, How perception yields knowledge

(The same caveat as for the first lecture applies to this truncated report.)

How perception yields knowledge

How does perception yield knowledge? In Charles Travis’ phrase, by placing the surroundings in view. But this lecture will take issue with aspects of Travis’ unpacking of that idea. Travis thinks that there is no content in experience itself.

Recognising a pig is more than just having it in view. One may lack the recognitional capacity or ability. So having a pig in view depends on cognitive capacities. Thus capacities can be additional to experience. This is a Travisian case. But Travis thinks that this is always the case. According to him, content is involved only when we make something of having the world in view.

Travis also thinks that if experience has content then we only encounter representations which serve as mere evidence for judgement.

But the structure of the case of recognising a pig need not be shared by all cases of perceptual experience. Cognitive capacities are not always additional to experience.

Consider the thought expressed by ‘I see that it is a pig’. Experience’s contribution in this case merely places the pig in view. The thing being a pig is not given in the experience. But unlike the presence of the pig, some things are given in experience: its size, colour and surface configuration. If experience places things in view, their visual features are present in the experience. That givenness consists in experience having content such as the presence of something pinkish which is at the disposal of the subject merely in having the experience. All one needs do is bring this – albeit with some loss of specificity – to explicit awareness. Some of what perception enables us to know is included in the experience. Content is needed for experience to make perceptible features visible.

Consider the argument from illusion as applied to the content view. If subjective character consists in perceptual content then if experience can make knowledge possible it cannot be by placing the environment in view. The worry is that experience itself can never provide conclusive warrant (ie such that the belief formed must be true) so experience itself can never show belief formed from it is knowledge.

This worry can be addressed in the following way. The way experiences have content is that they purport to reveal environmental realities. Of such an experience, it can be true that it does. If so, its content is its revealing. And if so, it has epistemological significance it does not share non-revealing cases.

That a belief is based on a non-revealing experience can still be rational. There can be a rational pattern with experiences with less than conclusive warrant. But these are cases where the subject at least seems to have conclusive warrant. So our grasp of less than conclusive warrant is derivative of revealing cases where experience warrants beliefs conclusively.

In rejecting a highest common factor model of warrant, one need not reject the idea that there is something common between ‘good’ and ‘bad’ disjuncts.

Perceptual belief may depend on experience plus recognitional capacities (eg. the pig case). But consider a case where the belief just registers environmental reality. One can know how one knows.

Perception is a fallible capacity. The broader capacity to have environmental reality present to one may break into sub-categories eg. the capacity to have something red and a rectangle present. But such subsidiary capacities have to have a kind of togetherness (a red rectangle is more than just redness and rectangularity present at the same time). A chromatic sensation is not enough.

But one can abstract to subsidiary capacities. Think of the fallible capacity to detect redness. One can have an experience as of redness, partly constituted by such a fallible capacity, and yet redness not be present. defective instances will still have an appropriate togetherness. It will still seem to the subject as though the experience reveals redness. When not defective, the same capacity can place redness in view. The content will be experiential / visual and be the same representational content in both disjuncts. But in non-defective cases the content is a case of revealing.

Two loose ends remain. First, consider the objection ‘One would only have a conclusive warrant if one knew that the experience was not defective’. If one has a capacity, one can know how one knows something. The capacity is to be knowingly presented with redness and know one has that experience. So capacity is fallible at both levels (knowing and knowing how one knows). But this does not imply that one cannot know how one knows when the capacity is not defective. Self-knowledge, like ground level knowledge, does not need to be infallible.

Second, since the defence of naive realism seems to need elaborate defence, can it really be called ‘naive’? Yes. The discussion of the two lectures has merely cleared away apparent difficulties to leave naive realism in place.

First lecture.