1. Our understanding of the localization, formation, and arousal of memory traces, of the process of abstraction and generalization and of the psychophysical relations of all these processes is still very unsatisfactory in spite of thousands of relevant investigations. The author comments several findings of different researchers, gives a summary of his and his coworkers' relevant investigations and discusses briefly some possible hypothetical conceptions. 2. Up to now the extension of nervous structures which are involved in the formation of engrams is not sufficiently known. It is also not clear whether or not brain regions of sensations and regions of corresponding mental images are identical (Fig. 1). During learning processes excitations in relatively extensive brain regions occur (Fig. 2). In mammals the brainstem and the midbrain and especially the hippocampus formation can be involved. These statements strengthen the opinion that the formation of a simple memory trace takes place in a network which comprises very many neurones.