AS we saw in Chapter XXV., the cognitive value of perception—i.e. its capacity for giving rise to inferences which are often valid—is a product of two factors, one depending upon the human mind and body, the other purely physical. The factor which depends upon the human mind and body is that which is concerned with "mnemic" phenomena. These occur wherever there is life, and to some slight extent in "dead" matter; but the higher the type of life the more notable they become. It is, however, the physical factor in perception that I wish to consider in this chapter, as it appears when separated from the mnemic factor. That is to say, I want to emphasize the fact that a percept is one of a system of correlated events, all structurally similar or semi-similar, and that the physical world, so far as known, consists of such events. My main purpose in dwelling upon this topic is to make it clear that percepts fit easily and naturally into their place in the physical world, and are not to be regarded as something quite different from the processes with which physics is concerned.

Let us revert to our earlier illustration of the dictaphone and camera which record a conversation with its accompanying action, and are found to agree with the recollections of eyewitnesses. When we considered this coincidence in a previous chapter, we were concerned with fundamental doubts; now we will assume the four-dimensional manifold of physics and the justification (in principle) of the inference from perceived to unperceived events. Assuming this, what can we infer as to the relation between (a) the sounds heard by the listener, (b) the events just outside his ear when he hears, (c) the events at the dictaphone at the same time, (d) the dictaphone record, (e) the sounds heard by the man when he listens to the dictaphone?

The similarity between (a) and (e) is fundamental, and is known by a comparison of a percept with a memory. Thus the problem of the relation between perception and memory is involved; but as this problem is psychological, I will only say that the inference from a recollection (which occurs now) to what is recollected (which occurred at a former time) appears to me to be essentially similar to the inferences in physics, and to warrant only a belief in identity (or close similarity) of structure between the recollection and the event recollected. The grounds for the trustworthiness of memory seem to be of the same kind as those for the trustworthiness of perception. But I shall take all this for granted, since our theme is physics, not psychology. I shall therefore assume that (a) and (e) can be known to be similar in structure, in the sense explained in Chapter XXIV.

We have thus a chain of processes, (a) at one end and (e) at the other; the end-processes are similar in the technical sense, and we assume that the intermediate processes are also similar, both to each other and to the end-processes. Let us consider this in somewhat more detail. The relation of (a) and (b) is that of percept and stimulus—i.e. a relation of effect to cause. The effect is a complex process; we assume that recognizably different percepts must have different stimuli; therefore the cause must be a complex process, at least semi-similar to the effect. We may take it as similar, not merely semi-similar, by ignoring those respects, if any, in which the structure of the cause is more complex than the structure of the effect. A similar argument will enable us to treat (d) and (e) as similar. Since (a) and (e) are similar, it follows that (b) and (d) are similar. We cannot attribute this similarity to chance, since it is found to exist whenever the necessary conditions have been fulfilled. Hence we infer that (c) must also be similar to the other processes. Since the dictaphone may be placed anywhere in the neighbourhood of the speakers, we infer that throughout a region surrounding them there are physical events similar in structure to the aural percepts of the listener. For light, the same thing follows from photographs. Consequently a percept, considered physically, is not very different from other physical events. We may suppose, if we choose, that it differs from them in intrinsic quality, and we know that it differs causally, since it gives rise to memories and inferences. Even these, however, are not so different from certain physical processes as they seem at first sight.

Memory is shown by the capacity for producing events similar in structure to certain previous events, when the right stimulus is applied. We are not always remembering everything that we can remember; we remember things when we are asked about them, or when something occurs which recalls them by association. The dictaphone "remembers" in this sense. It is true that it cannot "infer": it will not answer a question which it has never heard answered. But physiological inference, which is causally the basis of all other inference, is not very unlike other physical processes, and may quite possibly proceed according to the laws of physics. However, I do not wish to pursue these psychological topics; it is only perception and its non-mental analogues that I wish to consider.

We have to suppose that a great many events are taking place everywhere, since both light and sound can be recorded by instruments and observed by percipients. Our visual field is very complex, and the physical stimulus must have at least equal complexity: if this were not the case, we could not see a number of objects at once, nor could a photographic plate photograph them. Physics, however, simplifies all this by taking the stimulus to a sensation to be a periodic process, not a static event. Our perception of colour, for example, does not seem to be a periodic process analogous to a light-wave; in this respect, the apparent structure of a visual percept differs from that which physics assumes in the external cause. A few words must be said cm this topic, in order to make clear its relation to our general theory of similarity of structure.

First: in a transaction such as the passage from stimulus to percept, we cannot expect complete similarity of structure: at most we can expect as much as we find in purely physical transactions. There is a great deal of difference between a light-wave and a quantum change in an atom, yet they are related as effect to cause. What we know about the atom we know in virtue of the light-waves which make us see things; unless differences in light-waves corresponded to differences in atoms, light-waves would not be vehicles of information about atoms. Now when light-waves reach the eye, they have effects upon the matter of the eye, which reverse the previous process from quantum changes to light-waves. It is possible, in view of such theories as we considered in Chapter XIII., that the relation between what happens in the atom and what happens in the eye is more direct than the above account would suggest, but it would not be prudent to assume that this is the case until the theory of light quanta has become more adequate. We cannot, therefore, assume any very close relation between the physical process in the eye and the physical process in the atom from which the light comes. And a fortiori we cannot assume a very close relation between the percept and the process in the radiating atom. Yet it is only in so far as such a relation exists that vision can be accepted as a source of physical knowledge; in so far as the correspondence fails, vision ceases to be trustworthy.

Secondly: there is no reason why the degree of correspondence between stimulus and percept which is required should not exist between a periodic process and a static occurrence. So long as different processes give rise to different percepts, the requisites in the way of correspondence are satisfied. There is therefore no theoretic difficulty in the view that the stimulus to a sensation of red is a vibration, while the sensation of red itself has not this character, but is a steady state capable of continuing for a short finite time.

Thirdly: we do not really know that our percept of a colour does not have the rhythmic character of the stimulus. We know something about percepts, but not all about them. We all know that if an object is made to rotate rapidly, for instance on a top, we can see it rotating if it does not go too fast, but when it passes a certain speed we see only a continuous band. This is to be expected in view of the existence of akoluthic sensations. But it by no means follows that there is not a flicker in the percept, although we cannot perceive a flicker. Exactly the same thing applies to light and sound generally, and to the apparent continuity of motion in the cinema. We cannot know, unless in virtue of some elaborate argument, whether our percepts are static or rhythmical, nor yet whether their physical stimuli are continuous or discrete. Such knowledge is rendered impossible by the fact that we can only assume semi-similarity, not full similarity, between percept and stimulus.

There is therefore no difficulty in the accepted theory that the stimuli to our most important percepts are rapid periodic processes. On the other hand, there is a great advantage in this theory, in that it simplifies the physical world which has to be assumed as the cause of our perceptions. A physical system, conceived merely as a set of material units in space-time, is capable of an indefinite variety of rhythmic movements. Some physical structures are resonant for one period, some for another. Thus our sense-organs can select one sort of movement as the stimulus to which they will respond, and reject all the rest. In fact, it may be said that the essential characteristic of a sense-organ is sensitiveness to one sort of stimulus, which, in the case of the eye or the ear, must be a periodic movement. In this the sense-organs do not differ from lifeless instruments, such as photographic plates and gramophones. Such instruments have something closely analogous to perception, when we leave out of account the mental consequences which we observe in ourselves as a result of perception. And in a certain extended sense we may say that every body which behaves in a characteristic manner when a certain stimulus is present, and only then, has a "perception" of that stimulus. We can infer the stimulus from the behaviour of such a body just as well as from our own percepts—sometimes better, as in the case of a very sensitive photographic plate.

The outcome of the discussion we have been conducting in Part II. has been to justify the ordinary scientific attitude, and to minimize the gulf which seems at first sight to exist between perception and physics. We have seen that the inference from percepts to unperceived physical events, though it cannot be made mathematically cogent, is quite as good as any inductive inference can hope to be. And we have found that there is no ground in philosophy for supposing the physical world to be very different from what physics asserts it to be. But we have found it necessary to emphasize the extremely abstract character of physical knowledge, and the fact that physics leaves open all kinds of possibilities as to the intrinsic character of the world to which its equations apply. There is nothing in physics to prove that the physical world is radically different in character from the mental world. I do not myself believe that the philosophical arguments for the view that all reality must be mental are valid. But I also do not believe that any valid arguments against this view are to be derived from physics. The only legitimate attitude about the physical world seems to be one of complete agnosticism as regards all but its mathematical properties. However, something can be done in the way of constructing possible physical worlds which fulfil the equations of physics and yet resemble rather more closely the world of perception than does the world ordinarily presented in physics. Such constructions have the merit of making the inference from perception to physics seem more reliable, since they save us from the necessity of assuming anything radically different from what we know. From this point of view, they have a certain interest, and I shall partially develop them, at least as regards space-time, in Part III. But they must not be confounded with scientific knowledge: they are hypotheses which may hereafter prove fruitful, and which have already a certain imaginative value. But they are not to be regarded as necessitated by any recognized principle of scientific inference.