Social structure learning in human anterior insula.

Area of Science:

• Cognitive Neuroscience
• Social Psychology
• Computational Modeling

Background:

• Understanding how humans form social groups and represent group boundaries is crucial, yet mechanisms beyond explicit labels and dyadic similarity remain unclear.
• Existing models often focus on interpersonal similarity, neglecting the role of inferring underlying social structures.
• Representing social group boundaries is fundamental to human social cognition and coalition formation.

Purpose of the Study:

• To investigate the neural basis of social group boundary learning.
• To differentiate the brain's processing of social information derived from interpersonal similarity versus inferred latent social structures.
• To determine which neural signals better predict social group choice behavior.

Main Methods:

• Developed a computational model of latent structure learning to predict social group representations.
• Utilized functional neuroimaging (fMRI) to analyze brain activity during social group tasks.
• Employed model-based analysis to correlate neural signals with dyadic similarity and latent structure learning estimates.

Main Results:

• Medial prefrontal cortex/pregenual anterior cingulate (pgACC) activity correlated with allyship estimates based on dyadic similarity.
• Right anterior insula (rAI) activity correlated with allyship estimates derived from latent social group structure.
• Neural variability in the rAI significantly predicted behavioral choices regarding ally selection, unlike pgACC variability.

Conclusions:

• The brain utilizes distinct neural pathways for processing social group information based on direct similarity and inferred structures.
• Latent structure learning, processed in the rAI, plays a critical role in predicting social group affiliation and behavior.
• These findings illuminate the psychological and neural mechanisms underlying the fundamental 'us' versus 'them' distinction.