Abstract
Physiologically, it is believed that past, present, and pre-future memory is stored in the parietal and temporal lobes, hippocampus, and frontal lobes respectively. The hippocampal networks consist of three types of synapses that form the circuit. The spatial-signal which serves as the input to the hippocampus is transmitted through a synapse in the dentate girus (DG) to the CA3 then through another synapse to the CA1. The hippocampusCA3 is characterized by a distinct biological neural network which has a recurrent network. This circuitry compiles past memory into the present. On this subject, Dr. Tonegawa (Massachusetts Institute of Technology) et al have reported that when they used genetic techniques to knock out feedback in the hippocampal CA3 neural network, an extremely large number of cues were required to accomplish one action. According to this , it can be believed that the hippocampal CA3 forms a context of time sequence, and the CA1 maps the spatiotemporal context into its synaptic weight space. For the CA3 →CA1 network, Tsuda(1996 and 1998) proposed a computational model of chaos-driven contracting systems in which the unstable network(chaos-driven network, CA3) forms a context of events via chaotic itinerary and stable network(contracting dynamics, CA1) encodes its information as Cantor coding. In the CA1 Tsukada, et al. (1996,2005) proposed a spatiotemporal learning rule in mapping of spatiotemporal information onto CA1 synaptic weight space. Short-term memory is formed in this way. [Jpn J Physiol 55 Suppl:S9 (2005)]
