Elementary Construction of the Material World

This, finally, is Ayer's answer to the main question to which the whole book has been devoted: How do we come to use a material object language, based on our sense data?

Had our sense data been structured differently, we might not have been able to employ a material object language. (It may, however, be a biological or physical necessity that our sense data is suitably structured.)

Whatever exactly that means, there are certain patterns to our sense data that play a central role in constituting material object language.

• Some sense data resemble others to varying degrees.
• There is substantial stability and reproducibility of similar sense data under suitable circumstances.
• Sense data are systematically repeatable.
• Some of the systematic repetition depends only on the movements of the observer.

Note that the early stages of "object conservation" shown to exist by Piaget represent exactly the kind of empirical discoveries about sense data that Ayer is sketching.

Suppose that when you moved through space certain sounds occurred in a systematic way similar to that in which visual sense data actually do occur, and suppose that other people reported the same thing. It would be convenient to stop focusing on the occurrent sensations of sound and instead to think and speak of them as located at various places independent of us. It might not be necessary to do that, but it would be extremely useful.

The point is that that is exactly what occurs with our visual and tactile sense data, and that is exactly why material object language is useful. We learn to associate certain patterns of voluntary kinaesthetic sensation with certain constancies in our visual and to a lesser extent other sense data. We build up from that a theory of how to navigate from one place to another, to reidentify an object, and so forth. Ultimately, we reach a level of abstraction at which objects have unseen parts (backs and interiors) and at which objects all stand in spatial relation to each other in a three-dimensional space.

We at some point learn some simple principles about laying out a world: if you keep going in a direction you keep getting to new places and that you get back to where you started by reversing your motion in the sense of going straight in the opposite direction, and in the sense of reproducing the sequence of sense data.

The account just given answers the question, even in the sense of making clear how to go further in providing details. However, as Ayer argued in the last section, we shouldn't expect there to be further details that are universally applicable to all experience of material objects, but rather complicated differences in individual cases. Thus, the account gives the basics of as complete an answer as one could hope for to the general problem of constituting a material object language.

Ayer gives an account of why we take certain visual appearances to be associated with certain locations, and hence to be appearances of objects at those locations but do not do something parallel for sound. That is, even if there is a characteristic sound at a location, we do not take the sound to be part of an object. (We may take it to, for example, be made by the object, but that is not the same as taking it to be part of the object. To say that something is orange and to say that is it noisy are not claims of the same sort.) At MIT there was a study of how to make maps of the campus for blind students. Blind students use navigational canes, and so the maps put, not, for example, colors, but sounds at various locations (echo of a concrete overpass above, sound damped by nearby grass, ….) That looks good for Ayer, not bad: he never thought of examples like that, but he abstractly says why visual data get tied to objects when sound doesn't, and, for the blind person, his criteria apply.