Fig. 4: Experimental verification of the model-predicted preference map.

a Test stimuli for the four example domains and their corresponding average activation patterns. The first column shows the central 4×4 degrees of each stimulus. The second and third columns show the measured and model-predicted average activation patterns, respectively. R denotes the correlation coefficient between the measured and the predicted pattern. The scale bar denotes 2 mm. b The blue bars in the histogram show the distribution of correlation coefficients between the measured and the model-predicted activation patterns. As a control, the red bars in the histogram display the correlation coefficients between measured activation patterns of all pairs of different stimulus categories. Arrows indicate means of the respective distributions. Source data are provided as a Source Data file. c Population response matrices (z-scored, color scale lower left) to the test stimulus set for all classified cortical pixels (Fig. 3c). Cortical pixels were sorted from top to bottom based on their responses to the test stimuli for their respective category. d Measured stimulus preferences across the cortical surface. For each cortical pixel, we averaged its response to the test stimuli for each domain and identified the one with the highest response as its preferred category. The color of the cortical pixel represents its preferred category, the black contour outlines the model-predicted domains, and the hatched area represents the unclustered regions (grey in Fig. 3c, d).