Bayesian prediction and evaluation in the anterior cingulate cortex.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Computational Psychiatry

Background:

• The dorsal anterior cingulate cortex (dACC) is linked to diverse cognitive control functions, including conflict monitoring, error anticipation, and reward processing.
• Existing research presents varied interpretations of dACC's precise role in these functions.

Purpose of the Study:

• To quantitatively differentiate proposed functions of the dACC using a Bayesian ideal observer model.
• To investigate the neural basis of cognitive control and prediction error signaling in the dACC.

Main Methods:

• Utilized a Bayesian ideal observer model to predict stop-trial probabilities and response errors in a stop-signal task.
• Employed functional magnetic resonance imaging (fMRI) to measure dACC hemodynamic responses.
• Analyzed dACC activity in relation to absolute and signed prediction errors, conflict, and error likelihood.

Main Results:

• dACC hemodynamic response significantly encoded both absolute and signed prediction errors.
• After accounting for prediction errors, dACC showed no residual correlation with conflict or error likelihood.
• Findings align with some monkey neural recording studies but contrast with other neuroimaging research.

Conclusions:

• The dACC appears to report at least two distinct types of prediction errors, extending beyond reward-based learning.
• The dACC's role in cognitive control may be more specifically related to prediction error signaling than previously thought.
• This study refines our understanding of dACC function in decision-making and adaptive behavior.