Skip to main content
. 2019 Oct 30;10:4934. doi: 10.1038/s41467-019-12623-6

Fig. 5.

Fig. 5

Match between neural and DCNN face-spaces in the occipital and temporal clusters of face contacts. a The same analysis as presented in Fig. 2a for set 1, with the difference of computing the correlation coefficients separately, for an occipital cluster (occipital inferior gyrus) and for a temporal cluster (fusiform gyrus) of face contacts. Anatomical locations of contacts assigned to the two clusters are presented on the inflated common cortical surface (surface colored labels are identical to those in Fig. 1a). Gray contacts were not assigned to any cluster. Error bars denote bootstrap s.e.m. on image pairs. b Same as panel a for the face contacts and images included in set 2. Anatomical labels were taken from individual patients’ cortical surface reconstruction; therefore, the contacts do not always fall within the targeted gyri on the common surface. Note that the correlation profile remained highly similar for the two clusters, with no significant difference found in any of the layers (permutation test on contacts’ ROI label)