All We Need Is Sense Data

Last class, we discussed the traditional, Cartesian argument that we sometimes see sense data. Austin now turns to Ayer's discussion of the traditional argument that we never see anything but sense data. Ayer calls the two parts combined the argument from illusion. Austin calls the first argument alone the argument from illusion.

Last time we saw that it is at the very least not clear what we are seeing that is appropriately described as sense data in any case, as the duck-rabbit, and the slippage between looks, appearances, and how things seem show.

But the argument from illusion, and Ayer’s sense data language require that we be convinced that it is at least coherent in the most ordinary circumstances to think that we are seeing sense data—not "I see my hand in front of my face," but "I see my-hand-like sense data that are presented as being in front of my face." It may well be plausible that we see something like, something in the vicinity of, sense data in the case of hallucinations, afterimages, and perhaps some other cases, but the second part of the argument from illusion is intended to show that all perceptions are perceptions of sense data.

The argument has three parts: first, that everything we see is in some sense always of the same type, second, that what we see is always indistinguishable from sense data, and, third, that indistinguishable things are likely to be (or must be) of the same type. Austin attacks each of those parts.

1. Everything We See Is Always of the Same Type

There is no argument presented that everything we see is always of the same type. That conclusion is smuggled in, suggested by the question, "What kind of thing is the object of our perceptions?" It is suggested most subtly by the terminology of perceptions, which implicitly claims that there is some kind of thing called a perception. It is also suggested by the consideration of what kind of thing we perceive, as if we should expect that there is some single kind of thing we perceive. It is cemented in place by the claim that we ordinarily take ourselves to perceive material objects, as if there is a single kind of thing we ordinarily perceive usefully called a material object.

So, the argument from illusion, and the suggestion that there is a coherent notion of a sense datum language from the beginning rest on an unsupported presumption.

2. What We See Is Always Indistinguishable from Sense Data

The second part of the argument suggests that what we ordinarily see is ordinarily indistinguishable from what we see in the standard cases in which we see only sense data. Thus, for example, a stick in water is supposed to be indistinguishable in appearance from a bent stick, a mirage from an oasis, a dream from reality, and so forth. That, of course, is absolutely not true, and the fact that it even seems plausible in the course of the argument from illusion shows just how far down the garden path the terminology of the argument has taken us before the so-called data are presented. Austin's point is not that such a confusion might never happen. He is prepared to grant that it might. The point rather is that Ayer is trying to convince us that indistinguishability is the normal case in order to convince us that we normally can be said to perceive only sense data.

Austin asks how a professional tea taster would react if I were to try to persuade her that Twinings and Constant Comment actually taste the same. At best, it might be conceded that my sense of taste is so lousy that I could never tell the difference, but not that the tastes are the same.

3. Qualitatively Indistinguishable Things Are of the Same Kind

Special effects rely entirely on the fact that movies of aliens would be qualitatively indistinguishable from movies of people in suitable costume suitably lit, and so forth. It is an important fact that lots of things of different kinds are, in fact, qualitatively indistinguishable, for example, artificial flowers from flowers and, Austin's example, pieces of soap that look and smell just like a lemon from lemons. The principle on which the third part of the argument relies isn’t even faintly plausible. What about the case of continuous series? Once again, there is a familiar fact here, which does not ordinarily lead us to conclude that, the real color of an object, for example, is of the same kind as its color in poor lighting.

Ayer's View

Ayer rejects that argument from illusion and the view that all we see is sense data, but he accepts a related conclusion: that, for the purposes of understanding "the foundations of empirical knowledge" it is a good policy to "adopt" "sense datum language."

"Ayer … accepts … all the really important blunders on which the Argument from Illusion rests." 55

In Chapter 6, Austin considers Ayer’s actual position, not the argument from illusion, but the argument for the utility of sense datum language, and he has a devastating criticism of Ayer’s argument.

Ayer argues that material object language and sense datum language are "empirically equivalent," that there can be no evidence that one is superior to the other, and hence the dispute whether we "really see" material objects or sense data is an empty one. (around p. 25 of Foundations of Empirical Knowledge).

What is Austin’s criticism? On p. 60: Sometimes, at least, according to Ayer, there is real disagreement about the nature of empirical facts. What kind of disagreement can that be? (Given that disagreement whether coins change shape and whether the cigarette you hand me is the one I take are not matters of empirical disagreement for Ayer, its hard to imagine what could be.) "Ayer’s answer is quite clear—they are facts about sense data."

Ayer warns against the very mistake of which Austin accuses him:

For we can say to the naïve realist: "You say that perceiving a material thing can be analysed into sensing a sense-datum and knowing that it is part of the surface of a material thing. But is it not significant to say that the sense-datum by means of which %$A$% is perceiving a coin is round, and that the sense-datum by means of which %$B$% is perceiving the same coin is elliptical? And would it not be self-contradictory to say that a part of the surface of a coin was both elliptical and round?"

As an argument ad hominem this may be decisive. But all that it proves is that the naïve realist's thesis is inconsistent with the conventions of the sense-datum language; so that if we refute him by interpreting his thesis in terms of sense-data we are begging the question against him. 47–48

Here is an example of the Ayer presuming that all the facts are facts about sense data:

No matter which of [the theories] we adopted, we should be able to describe our perceptions, whatever their nature …. But if the relation of these three theories to the relevant phenomena is precisely the same, then, as theories, they are not distinguishable from one another. 53

What’s wrong with taking the empirical facts to be facts about sense data? Ayer is trying to show us that there is a legitimate sense datum language, just as good as our ordinary "material object language." But he does so just assuming at the outset, without argument, that the only facts that matter, the only ones we must account for, the only reports we aren’t free to misuse in various peculiar philosophical ways, are those about sense data. He is assuming almost exactly what he is supposedly showing.

Sense datum language doesn’t look empirically adequate at all, if we take the basic facts to which we must answer to include those about material objects, for sense datum language, for example, can’t express the fact that a coin is round. Ayer has given us no argument at all for adopting sense datum language. He relies on the argument from illusion to introduce sense data. He assumes that all empirical facts are facts about sense data, and then he shows that we can describe all the facts in sense datum language. That isn’t so much an argument as a restatement of his implicit assumption.

Present considerations do not prove that you couldn’t justify interest in some project about how our experience of the world leads us to picture or describe the world in the way we do. It just shows that Ayer hasn’t made a case for such a project and that it is probably not possible to separate out a pure notion of our experience different from our actual way of locating ourselves in the world (that is, a notion like sense data) to play a central role in such a project.

One clear symptom, one that Austin emphasizes, of Ayer’s disdain for any empirical facts not related to sense data is Ayer’s willingness to distort the way in which we use material object language, accompanied by his strict standards for how we are to use sense datum language.

Austin’s tradition has had several important kinds of impact on philosophy. One, which is usually attributed to Quine, is the abandonment of, in Quine’s term, "first philosophy," which has been replaced to a large extent by the attitude that philosophy is on a par with the rest of our knowledge, not prior to or a foundation of the rest of our knowledge. That is usually today accompanied by some form of “naturalism”: the idea that science and our knowledge of the natural world is knowledge par excellence, and that it is what we should start with, not conscious experience or anything like that. That is Quine, not Austin, but Austin’s criticism of first philosophy played a real role in bringing that about. Austin seemed to want to take our ordinary views as embedded in our language as the central form of knowledge, not science. He hasn’t won on that, but philosophy of language has come to be seen as a central part of philosophy in part because of his influence.

-- ShaughanLavine - 9 Apr 2003 - 11 Apr 2003 - 03 Apr 2007