In a non-social control study, we were able to replicate the signatures of basis functions in both behaviour and dorsomedial prefrontal cortex , while eliminating socially specific signatures.

When we weakened the group framing in a big behavioural study, we found that participants indeed performed worse in the task, despite allegedly simpler task space and more training. This suggests people use the groupings as a scaffold for solving the task.

We found that participants represent combinatorial building blocks that aggregate over all four players in medial prefrontal cortex. This is useful because it prepares you for the specific decisions that are possible.

Players were grouped into two teams – self and partner against two opponents. This gave participants important constraints on decisions that were possible and that were not possible within the experiment. Participants could solve the task by remembering combinations of players - aka basis functions

We used a social group decision making task in which participants keep track of themselves and three other players – not knowing in advance who will be decision-relevant and who will not. This allowed us to see how people simplify information into a format that allows them to make good decisions.

New commentary with @yongling.bsky.social out in TINS! We discuss the intriguing new results from Ma et al. suggesting that the brain maintains multiple predictions how other people might act in order to adaptively respond to others during social exchanges. www.sciencedirect.com/science/arti....