Reinforcement learning signals in the anterior cingulate cortex code for others' false beliefs

Area of Science:

• Cognitive Neuroscience
• Neuroeconomics
• Social Cognition

Background:

• Understanding others' mental states, including false beliefs, is crucial for social interaction.
• Reinforcement Learning (RL) models first-person decision-making using prediction errors (PEs).
• The Anterior Cingulate Cortex (ACC) is implicated in coding PEs during RL.

Purpose of the Study:

• To investigate whether the ACC signals PEs associated with others' false beliefs.
• To explore the neural mechanisms underlying the understanding of others' decision outcomes.

Main Methods:

• Functional magnetic resonance imaging (fMRI) was used to scan subjects.
• Subjects monitored third-person decisions and computer-generated responses.
• Trial outcomes were manipulated to create unexpected results, generating PEs.

Main Results:

• Gyral ACC activity correlated with unexpectedly positive outcomes of third-person decisions.
• Sulcal ACC activity was observed for unexpectedly positive outcomes in both third-person and computer trials.
• These findings suggest a specialized role for the gyral ACC in processing others' PEs.

Conclusions:

• The ACC appears to code PEs, extending beyond self-relevant learning.
• A distinct subregion, the gyral ACC, may be specialized for processing prediction errors related to others' beliefs.
• This research sheds light on the neural basis of understanding false beliefs in others.