第32回日本神経科学大会

会期:2009.9.16-18
会場: 名古屋国際会議場

細かい奥行き弁別における視差信号の空間統合:数理モデルから生理学的モデルへ
Nonlinear spatial integration of disparity signals in fine depth discrimination: from a mathematical model to a physiological model
演題番号 : P2-i01 ポスター発表 発表日:2009.9.17

島本 一馬(Kazuma Shimamoto):1 土井 隆弘(Takahiro Doi):1 岡田 貴裕(Takahiro Okada):1 藤田 一郎(Ichiro Fujita):1 

1:大阪大院・生命機能・認知脳科学(Frontier Biosci, Osaka Univ, Toyonaka)


--------------------------------------------------------------------------------

Stereoscopic depth perception improves when a reference target is presented beside a test target. We previously estimated how depth perception weighs and integrates disparity signals from various locations in the two targets, and developed a mathematical model based on this estimated "spatial weight". This model assumed relative disparity between the test and reference targets was a priori. Here we present a physiology-oriented model that spatially integrates the disparity signals in V1 into signals supporting fine, relative depth perception. The new model nonlinearly combines outputs from model V1 neurons (binocular energy neurons), each of which detects absolute disparity at a different location. By selectively weighing the responses of the model V1 neurons, our new model could produce a psychophysical spatial weight similar to that of human subjects. As a result, we postulate that extrastriate cortices nonlinearly integrate spatially local signals from V1 neurons into relative-disparity signals in a manner similar to our model.