Rules warp feature encoding in decision-making circuits.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Humans can follow arbitrary stimulus-response rules, guiding behavior through simple policies.
• Rule identity is encoded in decision-making circuits, but its computational impact on choices is less understood.
• Rules may simplify decision-making by reducing the need to process irrelevant information, potentially lowering metabolic demands.

Purpose of the Study:

• To investigate whether the brain leverages the computational simplicity offered by rules.
• To examine how rules influence neural computations during decision-making.
• To determine if following rules reduces the energetic cost of cognitive processes.

Main Methods:

• Recorded neural activity from orbitofrontal cortex, ventral striatum, and dorsal striatum in macaques.
• Utilized a rule-based decision-making task where rules were modeled as latent causes of choices.
• Contrasted neural activity during rule-based decisions with residual choices unexplained by simple rules.

Main Results:

• Following rules decreased the energetic cost associated with decision-making.
• Neural representations expanded dimensions relevant to the current rule and compressed irrelevant ones.
• Rule-based decisions were distinguished from non-rule-based choices through computational modeling.

Conclusions:

• The brain utilizes rules to reduce decision-making costs by optimizing neural representations.
• Distributed representational warping in decision-making circuits facilitates rule-based behavior.
• Rules enhance cognitive efficiency by selectively processing relevant information and minimizing computational load.