We have direct perception of our bodies (and other bodies as well) as well as our mental acts and states. . . I suppose if we begin from observation of what we have identified as direct perception, we should say this, for it occurs through the senses, and perhaps also through “internal awareness.” Is direct perception such that it cannot be wrong? Well, it is not direct perception, but falls short of it, if it gets it wrong, but that does not mean we can give any account of what happens in direct perception that would provide the necessary guarantee. We can only give an account of a given (sort of) direct perception, which explains why it will not be wrong under the usual conditions in which it occurs. If we give any causal account at all how a direct perception occurs, we have to allow that the necessary series of events happens only when given certain background conditions (that’s the only way causation works), and that seems to imply an externalist account. That is, direct perception requires not just that the perceiver be a certain way and do certain things, but that the perceived be a certain way and do certain things, and that the environment in which perceiver and perceived occurs be of an appropriate sort in which all this will or can happen. So no explanation of direct perception (the instances of it we encounter in the world) can explain why it cannot possibly be wrong, but only why it is not wrong under the usual circumstances.
But philosophers sometimes suppose that there is a direct perception that cannot possibly go wrong. How do they argue for it? Not through observation. Rather they argue that if we are to get any perception right, there must be a direct perception that cannot be got wrong. Behind every seeming-to-be-the case, lies someone who interprets something he perceives more directly as an instance of what seems to be the case. Somewhere, at the bottom, there has to be an uninterpreted perception, a starting point. Else the whole process is unintelligible, and perception of any sort is impossible (which can be established as absurd through observation, for we do perceive perception). In any actual case of an interpreted perception (and these are commonplace, of course) we may grant this, but why suppose that the uninterpreted perception at the base of the series is infallible? Because, I suppose, that is the only way the interpreted perceptions can gain credence. But what if the uninterpreted perception at the base is simply right, as it happens, given the way in which it arises. We might be able to explain why it is right (in ‘physicalist’ terms), while allowing that it might not have been, if things had happened differently, and surely that would be good enough, as long as it is pretty likely that the favorable background conditions needed are in place. (In practice, we usually pick up immediately signs that they are not in place when they aren’t—we are aware how dark or noisy it is, that we are drunk, or whatever, and so we get cautious about our perceptual judgements. So if we seem to ourselves to have no reason to be suspicious, we probably have no reason to be suspicious.)
(There is perhaps another issue here with the argument for infallible direct perception. We can ask, as we track the interpretations of perceptions that lead to our perceptual judgment, whether at a given level, interpretation occurs or not. If we press it, perhaps it always does, since to make any judgment at all (“I am aware of a blue patch in my visual field,” whatever that means) some sort of reasoning from my present state, of which I am directly aware, must be made. Why suppose that such reasoning occurs? Well, if we set out to justify our perceptual judgment, we will talk about how the perception occurs, and then argue that, under ordinary circumstances, it cannot have occurred unless it is true. That seems to land us on some perceptual judgment that does not involve such reasoning at the base of the thing. Now, maybe we can be quite sure that a given judgment involves reasoning and interpretation, because we catch ourselves doing it, quite out loud, as it were, or at least we suddenly realize, only after staring a moment, what we are looking at, and we can only explain the moment’s delay by supposing that we had to figure something out. But, of course, that leads us to identify direct perceptions much too early in the process—I can just see that it is my wife, but surely that depends on learning what she looks like first, and on inference, and so it can’t be direct. There must be a hidden reasoning going on. That is what doesn’t follow. The learning might not have involved reasoning at all, but the formation of certain neural pathways (or whatever), the which happens through some biological process that came to be established through biological evolution in my species. Now, if I am to justify accepting this immediate perception, I must, perhaps, explain those pathways and how they arise, and why that sort of pathway formation leads to accurate perception rather than error. That, moreover, can only be done by presenting some analogue to reasoning (a physical process which is trustworthy because it is an analogue, all things being equal, to reasoning of a certain sort). But it does not follow that really that reasoning occurs outside the physical realm, as I suppose a good Platonist might hold. The physics and biology, even if it can only be understood as producing perception by being understood to produce something analogous or parallel to reasoning, can certainly produce perception without such reasoning actually occurring, or the analogy being an intentional one. To explain why such analogues occur in nature, so that it is reasonable to conceptualize perception in the way we do, and find persons who are perceptually aware in the world, we must resort to natural selection, and although we don’t have anything except a fuzzy outline of how it might have occurred through natural selection, that is, currently, the only clue we have to pursuing such an explanation. Postulating a rational being that is immaterial, and essentially rational without any supporting mechanism, without a functioning machine, is not providing an explanation at all. That we are able to speak of such things in the world does not mean that there must be such things, nor do we ordinarily suppose there are, but only that there are things more or less like that, as long as they find themselves operating under the right conditions. The essential rationality of any actually rational being we know of is rather an incomplete and rocky affair, the being in question often turning irrational on us if placed in the right environment. Only by pressing the privilege of ad hoc presumption to its limit can we make such a view work, and scientists who want explanations always set a limit to how much of the ad hoc is to be allowed.
No doubt I am missing something, but that will have to do for now.