MAPPING AND RECOGNITION OF EVENTS IN ASSOCIATIVE MEMORY

In the first part, the author sums up general features of his model of holographic associative memory, consisting of a short‐time component and the associative layer, bearer of permanent records, realized as changes of synaptic weights according to a suitable algorithm. In living systems, some means are necessary to safeguard the memory from overflooding with irrelevant information. It is shown that the selection of biologically important patterns and events may be based upon the system of pre‐determined unconditioned reflexes. One possible way of introducing this principle is by allowing the unconditioned stimulus to activate the mechanism of synaptic changes so that the preceding event, still kept in the short‐time memory, may be recorded permanently, while other patterns and events occurring in the inactivated state are remembered only after numerous repetitions.