Back

Compositionality of social gaze in the prefrontal-amygdala circuits

Qi, G.; Dal Monte, O.; Fan, S.; Chang, S. W. C.

2025-07-29 neuroscience
10.1101/2025.07.28.667161 bioRxiv
Show abstract

Social gaze underpins primate communication, yet the neural principles enabling its flexibility remain unknown. Each social gaze can be deconstructed into three primitives: gaze content, social state, and gaze duration. To reduce dimensionality and facilitate generalization, the brain needs to represent these primitives in an abstract format. Here we show that social gaze is governed by a compositional code built from these primitives in the brain. In male and female macaques (neural recordings in two males) engaged in real-life social gaze interaction, behavior analyses revealed that partner responses were determined by how primitives were combined, rather than by their independent sums, providing evidence for behavioral compositionality. The basolateral amygdala and the anterior cingulate gyrus represented content and state in an abstract format and orthogonally to one another, whereas the dorsomedial prefrontal and orbitofrontal cortices exhibited limited generalization. Linear mixed-selective neurons encoding both content and state in the basolateral amygdala and the anterior cingulate gyrus, but not in the other two areas, facilitated the abstraction underlying generalization. Moreover, distinct channels routed content and state information across prefrontal-amygdala circuits to minimize interference, which was mediated by linear mixed selectivity neurons. These findings identify a neural grammar for social gaze, revealing compositional computations as a principle of flexible social communication.

Published in Nature Communications (predicted rank #2) · training set

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.