Modeling the representation and transformation of material-related information in the visual cortex

2016 Fiscal Year Final Research Report

• Project/Area Number: 25330179
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Cognitive science
• Research Institution: National Institute for Physiological Sciences
• Principal Investigator: Goda Naokazu 生理学研究所, システム脳科学研究領域, 助教 (30373195)
• Project Period (FY): 2013-04-01 – 2017-03-31
• Keywords: 視覚 / 物体認知 / 質感 / テクスチャ / 多感覚 / fMRI / CNN

Outline of Final Research Achievements

Our brain can easily recognize the material of an object - what the object is made of - just by seeing it. This study is aimed at understanding the neural processing of the material-related information through the visual stream in the brain and developing a computational model that can mimic the processes. A series of functional brain imaging and psychological experiments revealed that (1) the higher ventral visual areas in monkeys represent visual and non-visual material properties of objects, and (2) the neural representation can be affected by simple visuo-haptic experience on the objects, suggesting that is formed through learning of the crossmodal association. Further analysis with the state-of-the-art computer vision techniques revealed that (3) deep convolutional neural network models can capture the representation of the visual and nonvisual material properties. These new findings greatly advance understanding of material information processing in the brain.

Free Research Field

認知神経科学