Fig. 7: In silico tests reveal a V4 functional map of feature dispersity.

a Example of the distribution of critical features within receptive fields (RFs). Left: the preferred images and their corresponding heatmaps of a face-preferring cortical pixel; right: those of a cortical pixel preferring grid textures. The contour outlines the 2 standard deviations (2-std) of the Gaussian envelope of the cortical pixels’ RF. b Illustration of the content removal test. The first row shows the regions containing the K pixels with the highest heatmap values; the second row shows images with only the content in these regions preserved; the third row shows images with these regions’ content removed. See content removal approach in Supplementary Fig. 11 and Methods. The ratio of K pixels to the RF area is indicated at the top. c Model-predicted responses of the two pixels in (a) to Top K removed and Top K preserved stimuli for different K values. Solid lines represent the average response to the top 25 preferred images for each pixel, with the shaded stripes indicating standard errors (salmon: face-preferring, green: grid texture-preferring). We define feature dispersity (FD) as the ratio between the K values and the RF area at the intersect of the Top K removed and Top K preserved curves. Source data are provided as a Source Data file. d Feature dispersity maps for monkey C and monkey B. e Relationship between feature dispersity and frequency selectivity index across domains in monkey C. We averaged the indices of cortical pixels within each domain to obtain the domain index. Colors denote domain categories. Source data are provided as a Source Data file. f Relationship between feature dispersity and color selectivity index across domains in monkey C. Source data are provided as a Source Data file. For better visualization, (e, f) Show the RF-cropped preferred images of the example domains, where the cropped square region encompasses the receptive field (2-std) of the domain. See all the domains’ RF-cropped preferred images in Supplementary Fig. 13 and Supplementary Data 4, 5.