Accumulation of rapid and small synaptic increase as a basis for implicit memory
To understand the neural basis of implicit memory, a cortical neural network was modeled and simulated. As a cognitive process that relies on implicit memory, we employed “priming”, in which the identification of a stimulus is facilitated as a consequence of prior exposure to it. The network was trained to learn a visual scene that contains multiple objects each of which is composed of features with different sensory modalities. After the training, limit-cycle attractors corresponding to the learned objects are formed in the dynamic system of the network. Each limit-cycle attractor contains point attractors corresponding to the features of an individual object. In the priming test, the network is first stimulated (primed) with a cue feature that belongs to one of the objects. After the stimulation, we let the network identify one of its associate feature stimuli that belong to the same object. The identification of the associate stimulus is greatly enhanced if the cue stimulus is presented before the identification process, thus the network is primed. We demonstrate that the neural basis of implicit memory arises from the stabilization of relevant attractors, which is established by the rapid and small increase in the strength of synaptic connections during priming period. Repetitive trials of priming are stored as experience, in which synaptic accumulation is essential for the storage of the experience.