The middle sections of McDowell’s ‘Perceptual Experience: Both Relational and Contentful’

I’ve been reading McDowell’s recent paper ‘Perceptual Experience: Both Relational and Contentful’ [McDowell 2013]. Although I think I understand the broad outlines of the picture presented – which is not to say that there are no deeply puzzling aspects – I have been confused by the choreography in the middle. McDowell talks about an argument with a first and second premise. But he also talks about an argument which can be read in either a modus ponens or a modus tollens way. I’ve been puzzled by what he’s talking about. But I now think I have an interpretation. Crucially, there are two distinct arguments.

The argument that can be given either a modus ponens or modus tollens spin has the following elements. There is a premise which comprises a simple claim and a more complex conditional premise. There’s then a simple conclusion.

Presented as modus ponens, the simple premise is that:

• P1: the epistemic significance of experiences consists in their having content in the way they do.

The second, conditional premise is that:

• P2: If the epistemic significance of an experience consists in its having content in the way it does, then it cannot be by bringing environmental realities into view that experiences enable their subjects to know things about the environment.

Or more naturally:

• P2: If the epistemic significance of an experience consists in its having content in the way it does, then it cannot be that experiences enable their subjects to know things about the environment by bringing environmental realities into view.

The conclusion is that

• C: It cannot be by bringing environmental realities into view that experiences enable their subjects to know things about the environment.

Or more naturally,

• C: It cannot be  that experiences enable their subjects to know things about the environment by bringing environmental realities into view.

This modus ponens version leads to a highest common factor view of experience. The best experience does is to provide a defeasible warrant rather than bringing reality itself into view with conclusive warrant.

Read the other way round as a modus tollens argument, and holding the conditional constant, the conclusion is that it is not true that the epistemic significance of experiences consists in their having content in the way they do. One such view is Travis’ in which experiences play no content-bearing role. They are simply routes to the world for subsequent judgement. They enable the world to be in view for subsequent conceptual judgement. Here’s the description of that view:

We would be immune to the puzzlement those questions express if we accepted a relational conception of the sort that refuses to conceive the epistemic significance of experience in terms of content. On this view, experiences contribute to one’s having knowledge about one’s environment, but only by placing one’s surroundings in view. If one knows something about one’s environment through perception, that is because one has made something of things an experience places in view for one. One has brought them under concepts, exercising cognitive capacities that are extra to having the things in view. If someone has a knowledge-constituting warrant for a belief about how things are in her environment, the warrant depends on the epistemic credentials of the capacities she has exercised in applying concepts to things her experience anyway places in view for her. So the puzzlement does not arise. There is no application for the conception that causes it; there is no such thing as an experience that itself provides its subject with a conclusive warrant, or indeed any warrant, for believing something about her environment. [ibid: 149]

I don’t think that there’s a very strong argument against this position in this paper. We know that McDowell continues to think that it falls prey to the Myth of the Given and we know that Travis continues to think that it doesn’t. If I am at all on track with Travis his key idea is that it is an entirely acceptable view of the exercise of reason that it mediates between the conceptual (which exemplifies generality, understandings, occasionalism etc) and the non-conceptual (that which instances general concepts) so we do not need accept McDowell’s acceptance that there is almost a truth in the Davidsonian idea that the only thing that can serve as a reason for a belief is another belief. (Almost a truth because belief-like experiences can play just this role too. That’s the point of Mind and World’s transcendental empiricism.) But I will park this disagreement here. McDowell aims to show, at least, that it should receive no further support from the forced choice discussed here. He can head off a reason for thinking it obligatory. (That said, this does not seem to be the motive Travis himself has for his position which - mainly? - concerns, instead, the distinction between the conceptual, which admits of understandings, and the non-conceptual and the argument that making experience itself part of the conceptual is incoherent. For some attempts to get clear on that see this.)

Now one possible way to reject the conclusion that provides such a further motivation for Travis’ account would be to adopt the content view and attempt to head off the worry caused by the fact that contents can be misleading. Sadly it won’t do for reasons which have to do with the distinction drawn in ‘Knowledge and the internal’ between mathematics or logic and empirical knowledge.

We might be tempted by a thought on these lines: if one takes an experience to make knowledge available when it does not, that reflects some flaw in one’s cognitive conduct—haste, inattention, or whatever. That way, we could suppose an experience can conclusively warrant a belief about the environment, consistently with the requirement that one’s self-consciousness in enjoying it must put one in a position to know it has that epistemic significance. It is just that to avail oneself of that possibility of knowing the epistemic significance of an experience, one needs to ensure that one’s cognitive conduct, in exercising one’s capacity to know the epistemic significance of one’s experiences, is flawless. The idea would be that if one exercised that capacity without such flaws in cognitive conduct as haste or inattention, one would not be at risk of taking an experience to enable one to know something about the environment when it does not.
But the trouble with this idea is that it is not credible. Taking an experience to make knowledge available when it does not can be blameless. [ibid: 150]

So going back, both the MP/MT arguments depend on a conditional for which there is a distinct argument ie one which isn't the same as these MP and MT arguments. It happens earlier in the paper on p146 where the second premise is that experiences can appear to make knowledge about the environment available when they do not. The more complicated first premise is in this passage:

The epistemic significance of an experience must be available to the subject in enjoying the experience. It must reside in some aspect of the experience’s subjective character. So on the content conception, an experience’s having content as it does must be constitutive of its subjective character, in so far as its subjective character is relevant to its putting its subject in touch with her surroundings, or at least seeming to. [ibid: 146]

Now I think that McDowell accepts both of these. But they can seem, wrongly, to lead to the conditional which drives the MP/MT forced choice. So having set out that forced choice he has to explain why these two premises don't lead to the conclusion of this other argument: the conditional which drives the forced choice in the MP/MT argument.

Here the key thought is that a possibility has gone missing in the conventional view:

The premise that has been giving us trouble is that one can take an experience to make knowledge available when it does not. That does not show that when an experience makes knowledge available, it is not by bringing a suitable environmental reality into view for one. The point is just that the capacity to be in such positions is fallible.
We take in stride the fact that a capacity to know through perception such things as that there is something red and rectangular in front of one is fallible. We must, if we are going to suppose we ever have such knowledge. What makes it look as if the capacity cannot work in the way I have described— yielding knowledge whose title to count as such depends on experience making environmental realities present to us and thereby giving us conclusive warrant for beliefs about the environment—is the thought that knowing the epistemic significance of an experience that warrants a bit of knowledge we have through that fallible capacity would have to be an act of another capacity: a separate capacity to know such things, which would have to be in principle infallible. And the argument’s second premise is that there is no such infallible capacity. But the argument fails, because the capacity by which one knows the epistemic significance of one’s experiences is not another capacity, but just the capacity whose fallibility we anyway have to take in stride. [ibid: 151]

So at this point, to cut a longer story very short, the disjunctivism of McDowell’s account in ‘Criteria, defeasibility’, ‘Knowledge and the internal’, ‘Perception as a capacity for knowledge’ and others heads off the worry that the fallibility of observation impacts on the warrant of seeing when, in fact, all does go well. This includes the idea that the fallibility of the faculty of observation as a whole does not rule out the idea that ‘when all goes well’ one’s experience itself provides warrant sufficient for the truth of the belief; that the same faculty enables one to know how and that one knows; and the fact that, had one been in the bad disjunct one would not have known that has no negative implication for the possibility that if one is in the good disjunct one can know that.

Still, this paper makes something clear which hadn't been to me, at least, before. The content of experience can be the same in veridical and misleading cases, between the good and bad disjuncts. The word ‘content’ is used in this way. Content is a highest common factor. But there’s more to experience than its content, there is the way an experience has the content it does. This marks the key difference between being a case of seeing, or a case of having the world in view, and merely enjoying/suffering an appearance.

One further interesting claim is that visual experiences share a de se character. Seeings are de re but the broader class of visual experiences are de se:

Visual experiences as such, whether seeings or not, have content that enables those that are seeings to place their subjects in such relations to objects. But that is provided for, not by a de re character that belongs to the contents of seeings, but by a de se character that belongs to the contents of visual experiences as such. When I formulate an aspect of the content of an experience with the words “(There is) something red and rectangular in front of me”, “in front of me” is schematic for a much more specific placement, in relation to me, of an apparent red rectangular thing. That is enough to secure that if the experience is a seeing, it enables me to refer, with a visually grounded demonstrative, to the red rectangular thing I see. [ibid: 155-6]

What puzzles me here is why the de re character of a reference I chose to make of my experience - the sort of thing Travis starts his story with - belongs* to the content of the seeing and not the judgement made on its basis.
(*It is hard to know how to put this. It isn't that the content is de re articulated since the content of the good and the bad disjuncts are the same: de se. But the way the content of a genuine seeing comes about is de re.)

McDowell, J. (2013) ‘Perceptual Experience: Both Relational and Contentful’ European Journal of Philosophy 21: 144-57