|Budget Amount *help
¥40,040,000 (Direct Cost: ¥30,800,000、Indirect Cost: ¥9,240,000)
Fiscal Year 2017: ¥11,050,000 (Direct Cost: ¥8,500,000、Indirect Cost: ¥2,550,000)
Fiscal Year 2016: ¥11,310,000 (Direct Cost: ¥8,700,000、Indirect Cost: ¥2,610,000)
Fiscal Year 2015: ¥9,750,000 (Direct Cost: ¥7,500,000、Indirect Cost: ¥2,250,000)
Fiscal Year 2014: ¥7,930,000 (Direct Cost: ¥6,100,000、Indirect Cost: ¥1,830,000)
|Outline of Final Research Achievements
How faces are decomposed in the brain? Faces may be decomposed into facial parts, such as eyes, nose and mouth, and/or elements representing relational information of facial parts. We addressed this question by approximating visual features encoded by face neurons as natural image fragments. Resulting features reproduced response preference to particular views, and combination of them made view invariant face identification possible. Visualizing sub-region of faces captured by these features revealed that our brain decomposed faces into local features and complex features each of which detects a spatial arrangement of the local features. However, these local features are not explicit representation of facial parts. They are edge and color contrast derived from facial parts, hairline, and face lines. We suggest that the deep layered ventral visual pathway can be approximated by a shallow network comprising layers of low level processing and of combining low level information.