A distinction in this respect should certainly be made between short-term and long-term memory. It has been suggested that the former offers certain analogies with the workings of a digital computer, where familiar pathways can be reactivated or re-excited or in such a way as to constitute a residual record of past activity. The latter, on the other hand, must be physically embodied in some way, since it can endure for almost a hundred years, surviving comas, shock treatments, and even (in the case of experimental rats) refrigeration. An accidental concussion will often take away the memory of the hours immediately preceding it. It therefore seems probable that short-term memory is the more readily interrupted, being embodied in connections which must continually re-excite themselves[*] to last. If they do prove durable, such memory chains are (somehow) transferred to long-term storage, perhaps embodied by a literal “growth” within the synapses.
Short-term memory requires excitement
Memory itself, or the information it encodes, takes the form of connections, not of molecules. The quest for the biochemists’ holy grail, the so-called “memory molecule,” is therefore a mistaken one. Some researchers have tried to pass on memories by injecting extracts from schooled brains, or by encouraging cannibalism in worms.[+] Brains can indeed be altered chemically by injection, but specific information transfer has never been unequivocally documented. Fantasies of cannibalism appear to be more psychologically than biologically coded.[#]
[The Home Encyclopaedia of Human Psychology, ed. G. O’Bannon, p. 455]
Long-term memories transferred by cannibalism