記憶モデルによる記憶実験

逆向抑制のコンピュータ・シミュレーション

抄録

The purpose of this study was to simulate the mechanism of human memory using a computer. For this purpose, a memory model was constructed based on the knowledge of neurophysiology and information processing techniques.
Computer engineers often use the word “memory” in information processing. However, the computer “memory” is really a “record”, and must be distinguished from human “memory”.
Then, what is the difference between memory and record? I made the following assumptions on this problem.
(1) In a record, if we record datum 2 in the same storage where datum 1 is stored, datum 1 will be completely extinguished. In memory, on the other hand, it would not be extinguished completely. That is, many data are stored overlappingly on the same memory address.
(2) Even if some parts of memory storage have been destroyed physically or the stored data have been deleted partially, human memory can still recover much of the original information.
Each datum is stored individually on the different memory address in the computer data filing. However, it appears that human memory is the nonaddress type filing which has the characteristics described in (1) and (2).
(3) In record, a datum is recalled correctly, but memorized data is not recalled perfectly; that is, ambiguity and errors are inevitable in human memory. A program was constructed so that the above characteristics of human memory could be partially realized.
For the purpose of simulation, an experiment on retroactive inhibition was run, using 2 nonsense syllable lists.
An example of the results are shown in Table 4.
The model simulated a neuron-circuit network, in which each syllable was coded at random using S neurons, and was given the corresponding electrical bit pattern.
The simulation was carried out with the following procedures.
First, N paired syllables (AB)_i were memorized, and common bit patterns were extracted from similar syllables.
Then, B_i was estimated from A_i, or the specific bit patterns, by the transmission of impulses through the neuron-synapse-circuits.
In these processes, unique bit patterns were recalled easier than many other similar patterns and retroactive inhibition was observed to occur.