A neural circuit model forming semantic memory from episodic memory using STDP

The role of memory is important for human high-order functions such as thought, motion, and recognition. From many kinds of memory, this paper focuses on semantic memory and episodic memory which are declarative. It is thought that semantic memory is formed from common part of episodic memories. Though it is clarified which part of the brain is related to semantic memory, the neural circuit has not been clarified, especially concerning STDP which is minute learning regulation in time. We propose a neural circuit model forming semantic memory from episodic memory using STDP. By changing connections and input patterns in the model, we have examined the appropriate conditions by which semantic memory can be formed. The results show that three input routes, recurrent input, and appropriate propagation delay are necessary for forming semantic memory. Moreover, we have experimented forming a semantic memory after forming another semantic memory. The result shows that our model is able to form a semantic network. © 2007 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 90(2): 54–63, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20332