INFORMATION STORAGE AND ORGANIZATION IN THE BRAI N 1 F. ROSENBLATT Cornell Aeronautical Laboratory
If we are eventually to understand and the stored pattern. According to
the capability of higher organisms for this hypothesis, if one understood the perceptual recognition, generalization, code or "wiring diagram" of the nerv recall, and thinking, we must first ous system, one should, in principle, have answers to three f undamental be able to discover exactly what an questions: organism remembers by reconstruct 1. How is information about the ing the original sensory patterns from physical world sensed, or detected, by the "memory traces" which they have the biological system ? lef t, much as we might develop a 2. In what form is information photographic negative, or translate stored, or remembered ? the pattern of electrical charges in the 3. How does information contained "memory" of a digital computer. in storage, or in memory, influence This hypothesis is appealing in its recognition and behavior ? simplicity and ready intelligibility, and a large family of theoretical brain The first of these questions is in the models has been developed around province of sensory physiology, and is the idea of a coded, representational the only one for which appreciable mem ory (2, 3, 9, 14). The u nderstanding has been achieved. alternative ap proach, which stems This article will be concerned pri from the tradi tion of British marily with the second and third empiricism, hazards the guess that the questions, which are still subject to a images of stimuli may never really vast amount of speculation, and where be recorded at all, and that the the few relevant facts currently sup central nervous system simply acts plied by neurophysiology have not yet as an intricate switching network, been integrated into an acceptable where retention takes the form of theory. new connections, or pathways, With regard to the second question, between centers of activity. In many two alternative positions have been of the more recent developments of maintained. The first suggests that this position (Hebb's "cell assembly," and Hull's cortical anticipatory goal 11
storage of sensory information is in
the form of coded representations or response," for example) the "re sponses" which are associated to images, with some sort of one-to-one stimuli may be entirely contained mapping between the sensory within the CNS itself . In this case stimulus the response represents an "idea" 1 The development of this theory has been rather than an action. . The impor carried out at the Cornell Aeronautical Lab tant feature of this approach is that oratory, Inc., under the sponsorship of the there is never any simple mapping of Office of Naval Research, Contract Nonr- 2381(00). This article is primarily-an adap the stimulus into memory, according tation of material 'reported in Ref. 15, which to some code which would permit its constitutes the first full report on the program. later reconstruction. Whatever in- 386