At one of the universities, there is a beautiful library building, with a row of fine pillars across the front, and the students pass this building every day and enjoy looking at it. It has long been a favorite experiment in the psychology classes at that university to have the students call up an image of the library, and to have them state how clear their image is, how complete and how vivid. Then they are asked to count the pillars from their image, and to tell what kind of capitals the pillars have, and whether the shafts are plain or fluted. But at this point the students begin to object. "We have never counted those pillars, and cannot be expected to know the number now." In fact, few of them give the correct number, and those who have reported clear and vivid images are little better off in this respect than those whose images are dim and vague. The image, then, does not give you facts that you did not observe in the presence of the object. The substitute stimulus, which now recalls the image, only recalls responses which you made when the real object was the stimulus. If you looked at the object simply to get its general appearance, the general appearance is all you can recall. If you noted the color of the object, you can probably recall the color. If you noted such details as the number of pillars, you can recall these details. But the substitute stimulus that now arouses the image is by no means the equivalent of the original peripheral stimulus in making possible a variety of new reactions. Its only linkage is with reactions actually made by you in response to the real object. The substitute stimulus, such as the name of a building, became linked with responses actually made by you, not with responses that you simply might have made, when the object was present. This important fact is closely related to the {373} unreliability of testimony that was mentioned before under the head of "unintentional memory". [Footnote: See pp. 346-348.] Facts recalled are facts previously observed. It is true, of course, that recalled facts can be compared and new facts be observed by the comparison. We may recall how John looks, and how James looks, and note the fact, not previously observed, that they look alike. A great deal can be inferred in this way by a person who is sitting in his room far from the objects thought about. But this noting of the relationships of different objects is a very different matter from observing what is there, in a single object or scene. What is there can only be observed when you are there. The Question of Non-Sensory Recall