I’ll start with Austin’s criticism of the Phenomenal Principle. Your quote is,
If, to take a rather different case, a church were cunningly camouflaged so that it looked like a barn, how could any serious question be raised about what we see when we look at it? We see, of course, a church that now looks like a barn. We do not see an immaterial barn, an immaterial church, or an immaterial anything else.
If we take Austin at his word here, he is wrong, both in the implication that we do not see a barn in any sense, and in the implication that any insistence that we do implies insistence that we therefore must see something immaterial.
It takes a brief detour into linguistics to show that this is so. We know from semantics that there are intensional verbs that create scopally ambiguous sentences. One such famous example, discussed by Quine, is ‘seek.’ Suppose I say, ‘I am seeking a unicorn.’
There are in fact two readings of this sentence, whose truth conditions can be rigorously separated and given different, precise, logical forms. One reading can be paraphrased:
(1) There is an x such that x is a unicorn and I am seeking x.
On this reading, of course, it is not possible to actually search for a unicorn, because there aren’t any, and so the sentence must be false. But there is another reading:
(2) I am seeking the following: an x such that x is a unicorn.
This is perfectly intelligible and perfectly possible. I can search for a unicorn even though unicorns don’t exist (in fact, we often search for things that don’t exist precisely in order to find out
whether they exist).
‘See’ is an intensional verb in this same way, which has a scopal ambiguity similar to the
de dicto versus
de re distinction ((1) is
de re; (2) is
de dicto). So, if I say, ‘I see a barn,’ this can mean either (3) or (4).
(3) There is an x such that x is a barn and I see x.
Now this reading is clearly false, since there is no barn: only a church that looks like one. But Austin, and philosophers of perception generally, are insensitive to the second reading.
(4) I see (my visual experience is consistent with) the following: there is an x such that x is a barn.
That this reading is possible can be seen from the fact that without, it, the game in which people look at the clouds and ask, ‘what do you see?’ and say, ‘I see a man,’ ‘I see a shoe,’ and so on, would make no sense. In such situations the
de re reading is always false: there is never an x such that x is a man and one sees x in the clouds, because
there are no men where the players are looking in the sky, only clouds. Yet we can make perfectly true statements such as ‘I see a man,’ pointing at a cloud, on the latter reading. This is not a deviant way of speaking at all. Note that these are
structural ambiguities that arise
systematically, and are in no way
ad hoc, for which semanticists offer accounts.
So, in correcting myself, after I find out that the purported barn is actually a church, there are actually two things I can say, depending which of the two readings is intended. On the one hand, I can say, ‘I was wrong. I didn’t see a barn at all; I actually saw a church.’ But on the other hand, I can say, ‘I saw a barn, but there turned out to be no such thing — just a church.’ Both these responses are perfectly intelligible, and we actually know the structural reasons
why they are. Now Austin, and most philosophers, not only do not give the second reading its fair due, but also seem to exclude its very existence. This is unfortunate, and I think it has to do with metaphysical presuppositions, and an insensitivity to linguistic analysis (unfortunately, quite common in ‘ordinary language philosophers’).
This is crucial for the sense datum dispute. To say that one saw a shrinking object is not to commit oneself to saying ‘there is an object such that it was shrinking and I saw it.’ In eagerly trying to reject the latter interpretation, one wrongly rejects the more crucial point made by the indirect realist, which is that ‘I see a shrinking table’ is perfectly intelligible whether or not there actually is a table that is actually shrinking. So on the
de dicto reading, one can in fact say, on one reading, ‘I saw something shrinking.’ You just have to make clear which meaning is intended — one way it is true, the other way false. The direct realist is wrong that there is no sense in which it is true.
What this means is that although the indirect realist may have a tendency to
hypostatize the shrinking table, i.e. to treat it on a
de re reading as if there really were a thing that was shrinking, this misstep nevertheless leaves the importance of their philosophical point untouched, which is that even linguistically, we are sensitive to the notion of a visual medium through which things are seen, and we can characterize that medium as distinct from the objects that medium purportedly reveals to us. Direct realists are wrong to the extent that they deny or downplay this (the ‘one step removed’ that you talk about). The visual medium, the experiential field, is something to be spoken abut in its own right, and it has properties that vary independent of the object. We cannot just skip over it and pretend the objects are ‘just there’ for us with no more to do. That is not a viable philosophical position, and if a direct realist is forced to claim it is, so much the worse for his position.
This, then, is the kernel of truth in indirect realism that the direct realist is insensitive to and wrong about. You phrase the debate in terms of direct versus indirect realism, and in so doing imply that other alternatives, in particular idealism and skepticism, are not worth considering; but leaning toward the skeptical view myself, I’m in a somewhat privileged position to speak about the faults of both types of realism without the bias that attends trying to defend one’s own favored position. Indirect realists correctly claim that perception is mediated in a far less trivial way than the direct realist is willing to grant; direct realists correctly claim that there is no sense in which there is a two-step perceptual process that takes one from one immediate object to another mediate one.
Next I’ll take the Heidegger quote.
We never...originally and really perceive a throng of sensations, e.g., tones and noises, in the appearance of things...; rather, we hear the storm whistling in the chimney, we hear the three-engine aeroplane, we hear the Mercedes in immediate distinction from the Volkswagen. Much closer to us than any sensations are the things themselves. We hear the door slam in the house, and never hear acoustic sensations or mere sounds.
If we take Heidegger at his word, what he says is false. ‘We never…?’ We certainly do. We are met with visual impressions all the time that we aren’t sure what to make of, and so we do not first see them as determinate objects with specific significances, but a mess that we’re not quite sure what to do with (‘What the hell am I looking at?’). One problem here is that philosophers (and this goes for your paper as well, though interestingly not Heidegger’s quote which unusually focuses on sound) are overly accustomed to speaking primarily of visual perception, which is unfair to the full range of our experiences: visual perceptions are unique in that they, far more than those in other sensory modalities, seem (in my opinion, give the
illusion) that they simply grasp objects the way they are without any further ado. Of course, even in vision, this is not only not always true, but is in fact
never true (see below on this).
But the case is far easier to see with sensory modalities classically considered more ‘subjective,’ such as smell. We so much more rarely smell things and immediately know what sort of specific thing that we are smelling that, if philosophers focused on cases of smell rather than vision, I think none of them would be tempted to say the sorts of wrong things that Heidegger does on this subject. Of course I may encounter an aroma as
the smell of jasmine or
the sell of chocolate chip cookies; but there are so many manifold smells that confront me just as weird, unidentifiable sensations, ones that I’m not sure how to interpret and will likely never smell again. These confront me not as the smell of particular objects, but rather as olfactory
impressions upon me: ones that are painful or pleasant, imbue me with certain sensory affections, and so on, but that I cannot pin down. And when I do, this process can take, not milliseconds as in the case of veridical vision, but often whole seconds, even minutes or hours. These sorts of cases destroy the illusion that I simply ‘smell things as they are;’ there is a laborious process of piecing together, or projecting,
what I smell. Here the direct realist is wrong, and the indirect realist correct.
In the case of vision, as I’ve implied above, the situation is actually the same: we never simply see things right away as ‘what they are’ without further ado. There is again a laborious process of interpretation, but one that we are so well attuned to that in many cases it takes only milliseconds, and we are not consciously aware of it happening. We can actually
measure how long this takes with modern physiological techniques, and in the lab we can purposely mess up the interpretive process that projects some object out of sensory impressions. If Heidegger were right,
there would be no such process to mess up in the first place, since we
first see the object, and only
then do we abstract to sensation. And yet even outside the lab, our attempts at inferring from some sensation to some projected object go wrong, not only with outright hallucinations, but also e.g. when looking at surfaces when we can’t tell whether they’re flat or cornered (or for that matter, in seeing rainbows — there is a sense in which a rainbow is an object that we see, and a sense in which it is not, and it arises due to a curiosity in our visual mechanisms). It should also be noted that our modern laboratory equipment was not the advent of this realization: Schopenhauer said as much before modern psychology was a real discipline, and he had many examples of such ‘messing up’ of the visual interpretive process that you could perform for yourself purposely on a child’s allowance. The ‘remove’ that the indirect realist speaks of is very real, and you can see it for yourself in the process of its
happening and its
breaking down. Not only that, but ontogenetically we must
learn how to see, and so the direct realist needlessly privileges adult humans with fully functioning visual capacities that have had years of practice at what they do, they who have forgotten how hard it was to unscramble the ‘blooming, buzzing confusion.’ It is almost as if someone literate thought that in seeing words, we just ‘see their meaning for what it is’ as a perceptual matter, and denied that there was really any process involved in constructing their meaning (note that for the literate person, seeing the meaning of a word, too, requires no conscious effort). Visual perception is much like reading. There is in this sense very much an ‘inference,’ though of course it is not always, not even usually, a conscious one or one of ratiocination.
In other words, when you say this:
If I think about my own perceptual experiences, it is obvious that I am immediately aware of the true sizes and shapes of objects (unless I have "snapped out" of normal perception).
You are speaking falsely. And your comment about only awareness being relevant does not salvage you, since this can be brought easily to awareness by messing up the inferential process, or even in the case of smells that take minutes or hours to recognize.
Now, as to your comments on the duck-rabbit:
This is the duck-rabbit, made famous by Wittgenstein in his Philosophical Investigations. Notice that we switch between exclusive ways of seeing it, now as a duck, and now as a rabbit. This is termed multistable perception, and it does not fit very well with the notion that we infer a duck or a rabbit from neutral stimuli. It is not because we analyze the picture, checking off its properties against a mental database, that we see it as a duck or as a rabbit. Rather, we immediately see it as a whole, excluding other ways of seeing it at any instant. In doing so we directly see what is significant or what has meaning, and exclude the rest. Perceiving as this or that is built right in to perception. This is a clue to a big problem with The Argument, namely its underlying assumption of a structure of perceptual experience with (a) a static, passive and purely sensory gathering of stimuli, combined with (b) rational deliberation on the representations that result from these sensations. The task of bringing out the problems with this view, and supplying a better picture of perception, will have to wait for a future article. I think we can at least say that twentieth century psychology and philosophy have given us good reason to think about perception differently.
This passage is to me bewildering, because I think that the duck-rabbit shows exactly the opposite of what you want it to show. In the case of the duck-rabbit, we’re faced with a situation in which it makes no sense to say that you just see the picture immediately as it is, because there
is no such way that it is (if by this we mean among the alternatives between a duck and a rabbit). In fact, in switching from one to the other, you can
see the inferential process in action. All that’s relevant to the switch is that process — the picture hasn’t changed relevantly. Yet if, as the direct realist claims,
what we see are simply things, this would be impossible. Second, I think you are wrong to say that we exclude other ways of seeing it: it is possible in working visually around the duck-rabbit image to hold it in limbo between its duck and rabbit gestalts, and the visual sensation in that case becomes very odd, and not at all in keeping with how the direct realist claims visual experience should look or function generally. The comment about the
indirect realist making perception too passive is also odd: it is usually in my experience the indirect realist who emphasizes the activity of perception, and the direct realist who wants to minimize one’s role in perceiving in order to maximize the role of the object: we simply see things
as they are, and thus there is little role for perception other than to just
open us up to that way. For the indirect realist, the task of perception is far more laborious: given a stew of impressions upon the senses, one has to
cobble together the object, not just
receive it as is. And in doing so, one of course can cobble together either a rabbit or a duck, and then switch at will between the two processes, changing the perception without changing any ‘object.’
How is that possible, if perception ‘immediately’ just sees what is there?
Now the kicker is,
everything is a duckrabbit. So I also think this is in error:
I do not think so, because such cases cannot be taken as the paradigm of perception.
They are the paradigm. The duck-rabbit in one sense exists at the periphery; but in another sense it does so precisely because it shows you in a visceral way what is
always going on. That we have perceptual mechanisms that tend to immediately prefer one interpretation does not in any way mean that
there is no interpretation, via precisely the medium that the indirect realist speaks of.
Finally, I want to talk about hallucination. This is important because while I believe that many of the direct realist claims above are empirically wrong, coming from false claims about linguistics or how perception works, this is where I think that the position generally shows itself to be
internally incoherent, that is, not tenable even according to its own claims.
The clue to unravelling this argument is the phrase "object of awareness" in the conclusion. Where does Robinson get it from? Well, he is explicit in taking the second premise to imply mental images or sense-data in both hallucination and perception. To bring out the mistake here it will be useful to look once again at the distinction I have already made between two ways of using the word "see". Here I am going to make use of Audre Jean Brokes' terms and label these uses as (1) perceptual visual experience, and (2) ostensible perceptual experience. (1) is seeing external things, and (2) is "seeing" mental images. Incidentally, I use scare quotes for the second usage because in a discussion about perception, seeing external things naturally ought to be the privileged usage, of which the second is derivative.
This move to the distinction between genuine and ostensible visual experience is one that eventually the direct realist always seems forced to make in response to the uncomfortable fact that hallucinations happen. Now, this puts the direct realist in a really foul position. The reason is that, the direct realist must simultaneously claim that (1) hallucination can, at least at some points, be phenomenologically indistinguishable from veridical perception (as you happily concede, and though some deny this, I like you don’t take that denial seriously in lieu of serious argument); (2) that therefore there can be phenomenologically indistinguishable states that still differ as to whether or not they are states of perception. What this means is that, if we add a few plausible assumptions, the direct realist cannot,
in principle, and according to his own claims, tell the difference between ostensible and genuine perception, ever. The reason is of course that there is no way to tell such a difference if the only distinguishing evidence one can have between the two is phenomenological, and
ex hypothesi the direct realist is forced to admit that no such evidence can possibly exist. It follows that the direct realist not only never knows when he is perceiving anything, but
cannot know; it is impossible. That is to say, no evidence that one could possibly have for perception is such that it distinguishes between veridical perception and a long, coherent, constructed dream. This spells a problem for the direct realist in two ways: (1) how does he know so much about perception, that he can give us a whole theory about it, when he has never experienced one case of it that he can in principle tell it apart from cases that are not perception? (2) What on Earth is even the relevance of his metaphysical thesis about the objectivity and perceiver-independence of objects, if all perceptual experience is equally coherent and behaves experientially the same way whether that status obtains or not?
Now, there are many answers to these worries, but I have never heard one that comes close to being adequate.
Thank you for writing this article and taking the time to read my criticisms.