Appearance and Reality

Last class, we talked about how material-object language arises because of certain facts about how our sense data are structured. That was the main thing we needed to explain. The reason, however, that that seemed to be in need of explanation is because of the phenomena that have been used as the basis of the argument from illusion: hallucinations, mirages, round coins looking elliptical, square towers that appear round from a distance, navy blue tuxes looking black, …. Thus, the explanation is not complete until we see how those phenomena fit.

The part about sense data giving rise to a belief in the existence of objects that in fact do not exist is easy: We gave general criteria last time for taking a material object to exist on the basis of sense data. The cases of objects that seem to exist but do not are cases in which the criteria at first seem to be fulfilled, but are subsequently found not to be.

The hard case is objects appearing to have properties that they do not. Sometimes a round coin looks round; sometimes elliptical. Why do we conclude that it is really round?

In such cases all of the relevant sense data are parts of collections of sense data that do belong to an object, they reverse appropriately, and so forth. In addition, all the sense data, as sense data, are on a par, there isn't anything internal to the sense data that marks them as giving real properties of the object to which they belong or merely apparent properties.

In general, all objects do appear to have the properties that they actually have under favorable circumstances—and they sometimes appear to have properties that they do not really have under unfavorable circumstances. (Think what it would be like for that to be false.) That's fine, but it doesn't answer our question without an account of what favorable circumstances are. A list of favorable circumstances isn't enough, since we need to know what it is about the items on the list that makes them favorable. In addition, favorable circumstances are not only different for different properties. Even for a single property they can be different for different objects: The best distance at which to view the shape of an object varies with the size of the object. Seeing the true color of an object always requires good light and properly functioning eyes, but it may require different backgrounds for different objects, for example. Not only isn't a list enough, you can't give the list. It seems hopeless to provide rules for all the cases, but some account of what count as favorable circumstances is required so that we can give the answer that there are no illusions under favorable circumstances, only under unfavorable circumstances, some content, and the account must be given entirely in terms of sense data. (Why? The whole project of the book is to explain why we take some appearances, that is, sense data, to be veridical and some delusive; how we arrive at a picture in which sticks that are "really" straight can appear to be bent. In other words, the project is to explain on the basis of sense data alone, that is, in the sense-datum language, the difference between veridical and delusive perception.)

Ayer's claim is that the properties an object appears to have under the circumstances in which the apparent properties provide the greatest predictive value are the "real" properties (Chris suggests that greatest value for using the objects would be better), and thus, that favorable circumstances are those that give us the sense data with the most predictive value.

Ayer's claim suggests (what was not part of the data on which we constructed the theory and is therefore more impressive), what is actually the case, that if certain peculiar kinds of sense data had more predictive value than the ones we usually take to be real, we might take those to be even better, more real than the usual ones. Ayer's example is measurement of the shape of an object. That has more predictive value than any mere inspection in which a ruler is not present. We may therefore, in certain cases, have special kinds of sense data that we take to tell us more about what is really the case than the usual ones. Another example would be doing an analysis to determine the chemical composition of something.

There is a mistake in Ayer's account: it just isn't true that there is a best distance from which to observe the shape of an object. You might need to get far away to see the overall shape (and of course you need to look from several positions, because the shape includes the shape of the back of the object), but you might need to look closer up to see some of the details of the shape. That isn't a serious problem for the view, it requires a minor modification: to determine the real shape, color, texture, whatever, of an object requires making not one observation from "the" favorable perspective, it requires making many observations (a collection) from a collection of perspectives that permit acquiring the maximally predictive information.

Often the maximally predictive information is not obtained by casual observation but by means of arcane scientific measurements. That provides some reason to say that what those measurements tell us is more real than what we see. Thus, expositors of science make claims like "the table is really mostly empty space, it is composed of protons, neutrons, and electrons." There is nothing wrong with that claim until it is presented as a rival to our ordinary claims: the claim may be perfectly correct, but it does not show that our ordinary assessment of the table as made of formica, chrome, and steel is incorrect or illusory—both are correct, and they are not rivals, merely different levels of detail of description of the same thing. For most purposes, most of the time, we only need to predict what we will sense in ordinary ways, and so we have a notion, for example, of the real shape, adequate to those purposes. When that doesn't suffice we may move to a different level of description. As Ayer recognizes (e.g., 271), "real" is relative to current interests and purposes, just like flat: flat for a soccer field is very different from flat for a mirror.

Now, some side issues: touch is less subject to illusions concerning shape than vision: round coins always feel round, though they sometimes may look elliptical. That means that the shape it feels like an object has is a better predictor of subsequent experiences than the shape an object looks like it has. Thus, some philosophers have taken the felt shapes to be the real shapes, and have accorded touch a kind of special status of genuineness. But what is really at issue is predictive value, and careful visual observation, and certainly measurement, may trump touch. Thus, the difference between vision and touch for ascertaining shape is one merely of degree, not of some fundamental philosophical divide. That is important: see the next paragraph.

Some of the things we sense can also be measured, and some of those fit into a highly predictive scientific theory of subsequent experience. Those things (shape, mass) have been called primary qualities, and the others given a secondary status as mere appearance that are not really qualities of the objects themselves (color). There is no such fundamental distinction (or at least the way it has been drawn fails). The observations with the best predictive value give us real qualities of material objects, even though some of those permit more precise predictions than others. In particular, touch is not, in any metaphysically fundamental way a primary sense and vision a secondary sense.

Scott asks how one and the same table can be both solid (from our ordinary perspective) and not solid at all at the level of subatomic particles. I answered that ordinary things can't penetrate the table, but subatomic particles can. He replied that this seems to make what is real thoroughly relative, but isn't the table real, and doesn't it have real properties, independently of our language and decriptions? How do we account for that reality? In this case, it is true of the table, independently of our description that it is impenetrable to large objects at low energies, and easily penetrable by subatomic particles at higher energies. Both things we said were real are real. To give a general answer that applies to both, we would need to drop the term "solidity" and use some more refined method of description (for example, an s matrix).

---

Next time, we start on Austin. Austin's criticisms of Ayer's project and of his solutions to problems are absolutely devastating. It is therefore worth pointing out that Austin only criticizes the parts of Ayer's project (and there are a lot of them) for which he has devastating criticisms. There is a lot he never mentions, and of that, there is a lot with which he agrees. For example, he agrees that attention to language is central to understanding most philosophical problems, and he agrees that talk of sense data is talk in a new language (though he doesn't like the term "language"), not an empirical matter, and there is a lot more.

-- ShaughanLavine - 28 March 2003 - 22 Mar 2007