There is, however, still a further refinement necessary. Whatever is included under our present definition is an image, but some things not included are also images. The sight of an object may bring with it a visual image of some other object frequently associated with it. This latter is called an image, not a sensation, because, though also visual, it is not appropriate to the stimulus in a certain sense: it would not appear in a photograph of the scene, or in a photograph of the retina. Thus we are forced to say: the sensation element in the reaction to a stimulus is that part which enables you to draw inferences as to the nature of the extra-cerebral event (if any) which was the stimulus;[9] the rest is images. Fortunately, images and sensations _usually_ differ in intrinsic quality; this makes it possible to get an approximate idea of the external world by using the usual intrinsic differences, and to correct it afterwards by means of the strict causal definition. But evidently the matter is difficult and complicated, depending upon physics and physiology, not upon pure psychology. This is the main thing to be realised about images. [9] _I.e._ the immediate stimulus, not the “physical object”. The above discussion has suggested a definition of the word “image”. We might have called an event an “image” when it is recognisably of the same kind as a “percept”, but does not have the stimulus which it would have if it were a percept. But if this definition is to be made satisfactory, it will be necessary to substitute a different word in place of “percept”. For example, in the percept of a visible object it would be usual to include certain associated tactual elements, but these must, from our point of view, count as images. It will be better to say, therefore, that an “image” is an occurrence recognisably visual (or auditory, etc., as the case may be), but not caused by a stimulus which is of the nature of light (or sound etc., as the case may be), or at any rate only indirectly so caused as a result of association. With this definition, I do not myself feel any doubt as to the existence of images. It is clear that they constitute most of the material of dreams and day-dreams, that they are utilised by composers in making music, that we employ them when we get out of a familiar room in the dark (though here the rats in mazes make a different explanation possible), and that they account for the shock of surprise we have when we take salt thinking it is sugar or (as happened to me recently) vinegar thinking it is coffee. The question of the causation of images--_i.e._ whether it is in the brain or in other parts of the body--is not one which it is necessary to our purposes to decide, which is fortunate, since, so far as I know, there is not at present any adequate evidence on the point. But the existence of images and their resemblance to perception is important, as we shall see in the next chapter.