Summary
International Symposium on Nonlinear Theory and Its Applications
2016
Session Number:A4L-D
Session:
Number:A4L-D-6
A Computational Model for Pitch Pattern Perception with the Echo State Network
Miwa Fukino, Yuichi Katori, Kazuyuki Aihara,
pp.-
Publication Date:2016/11/27
Online ISSN:2188-5079
Summary:
The predictive coding theory assumes that the sensory system of the cortex continuously predicts incoming stimuli and detects residual errors. The mismatch negativity (MMN) is a neural response to a deviance of learned regularity, and is regarded as an error signal in this theory. Here we report a preliminary study on a computational model of the auditory MMN using the Echo State Network which is one of the recurrence neural network models. We trained the network by an oddball task with two pitch patterns. The result shows that our model simulates a qualitatively similar waveforms with the MMN response to deviant pitches.