Brain-wide representation of social knowledge

Abstract

Understanding how the human brain maps different dimensions of social conceptualizations remains a key unresolved issue. We performed a functional magnetic resonance imaging (MRI) study in which participants were exposed to audio definitions of personality traits and asked to simulate experiences associated with the concepts. Half of the concepts were affective (e.g. empathetic), and the other half were non-affective (e.g. intelligent). Orthogonally, half of the concepts were highly likable (e.g. sincere) and half were socially undesirable (e.g. liar). Behaviourally, we observed that the dimension of social desirability reflected the participant's subjective ratings better than affect. FMRI decoding results showed that both social desirability and affect could be decoded in local patterns of activity through distributed brain regions including the superior temporal, inferior frontal, precuneus and key nodes of the default mode network in posterior/anterior cingulate and ventromedial prefrontal cortex. Decoding accuracy was better for social desirability than affect. A representational similarity analysis further demonstrated that a deep language model significantly predicted brain activity associated with the concepts in bilateral regions of superior and anterior temporal lobes. The results demonstrate a brain-wide representation of social knowledge, involving default model network systems that support the multimodal simulation of social experience, with a further reliance on language-related preprocessing.

Keywords: abstract concepts; language models; searchlight decoding; social cognition.

Fig. 1.

Illustration of the experiment workflow with sub-figures labelled from A to C, (A) A total of 36 social concept definitions matched one of our four subcategories reflecting a combination of the affect and social desirability of the social knowledge. (B) Participants listened to the definition of a social concept and were asked to mentally simulate a person behaving the way described in the definition. (C) We acquired one anatomical and eight functional sequences in a single scanning session.

Fig. 2.

Distributions of ratings of social concepts. Participants read each concept definition and rated the extent to which the described behavior involved the emotions of oneself or others (affect; red) as well as whether such behavior was socially desirable (social desirability; gray) on a scale from 0 (very non-affective; very unlikable) to 100 (very affective; very likable).

Fig. 3.

NeuAverage whole-brain searchlight classification scores of the affect dimension (affective vs non-affective) of the social concepts. The heatmap levels represent the clusters where the ROC-AUC scores were statistically significant.

Fig. 4.

Neuroimaging results. Average whole-brain searchlight classification scores of social desirability. The heatmap levels represent the clusters where the ROC-AUC scores were statistically significant.

Fig. 5.

Neuroimaging results. Average difference in the whole-brain searchlight ROC-AUC scores where decoding social desirability was better than decoding affect. The heatmap levels represent clusters that were statistically significant.

Fig. 6.

Neuroimaging results. Average correlation coefficient maps and the corresponding corrected P value maps of the standard RSA and encoding-based RSA. (A) The correlation coefficients of the standard RSA that were greater than the empirical chance level. The average correlation maps were masked by the randomized P value cluster map that thresholded voxels with a significance level of 0.05. (B) The clusters where the average correlation coefficients of the encoding-based RSA were greater than the empirical chance level.