Skip to main content
. 2020 Apr 17;23(5):101060. doi: 10.1016/j.isci.2020.101060

Figure 2.

Figure 2

Neural Decoding Analysis to Predict Emotion Scores

(A) Decoding accuracy for individual categories. Dots indicate accuracies obtained from individual cortical regions of the HCP parcellation and subcortical regions (n = 370, five subjects averaged; see Figures S1A and S1B for results of representative cortical regions and subcortical regions). Red/pink squares indicate accuracies by ensemble decoders that aggregate predictions from multiple brain regions (error bars, 95% confidence intervals [C.I.] across subjects). Brain regions with significant accuracy from all five subjects are colored (r > 0.095, permutation test, p < 0.01, Bonferroni correction by the number of brain regions). Color and shape of each dot indicate locations of individual brain regions as (C).

(B) Decoding accuracy for individual dimensions. Conventions are the same as in (A).

(C) Color map of individual brain regions. Several brain regions are outlined and labeled (abbreviations: VC, visual cortex; TPJ, temporo-parietal junction; IPL, inferior parietal lobule; PC, precuneus; STS, superior temporal sulcus; TE, temporal area; MTC, medial temporal cortex; DLPFC/DMPFC/VMPFC, dorsolateral/dorsomedial/ventromedial prefrontal cortex; ACC, anterior cingulate cortex; and OFC, orbitofrontal cortex).

(D) Cortical surface maps of decoding accuracies for individual categories (five subjects averaged). Conventions of the marker points for the subcortical regions are the same as in (C).

(E) Cortical surface maps of decoding accuracies for individual dimensions (five subjects averaged). Conventions are the same as in (D).

(F) Correlations of region-wise decoding accuracy patterns between paired subjects. In congruent conditions, each dot indicates a mean correlation between the same emotion pairs averaged across all pairs of subjects. In incongruent conditions, each dot indicates a mean correlation between different emotion pairs averaged across all pairs of subjects and emotions.

(G) Emotion identification accuracy based on inter-individual similarities of region-wise decoding accuracy patterns (error bars, 95% C.I. across pairs of subjects [n = 20]).