Striatal arbitration between choice strategies guides few-shot adaptation

Area of Science:

• Neuroscience
• Animal Behavior
• Computational Neuroscience

Background:

• Animals exhibit rapid action changes in dynamic environments, necessitating effective adaptation strategies.
• Context-switching challenges require flexible behavioral adjustments based on environmental feedback.

Purpose of the Study:

• To investigate distinct action-selection strategies employed by animals following expected versus unexpected outcomes.
• To develop and validate a computational model explaining behavioral adaptation in response to outcome expectation.
• To elucidate the neural mechanisms, particularly within the striatum, underlying these adaptive strategies.

Main Methods:

• Designed novel behavioral measures to differentiate trial-by-trial dynamics after expected and unexpected events.
• Utilized reversal learning tasks across diverse rodent species and task complexities.
• Developed a computational model incorporating outcome-contingent decision variables (support-stay and conflict-shift bias).
• Analyzed electrophysiological data to identify neural correlates of the model's variables.
• Performed targeted inactivation of striatal pathways (direct and indirect) to assess their causal roles.

Main Results:

• Conventional learning models failed to capture choice behavior following unexpected outcomes.
• The proposed model, with distinct decision variables based on outcome expectation, successfully replicated observed behavior.
• Striatal neurons were found to encode key variables of the proposed computational model.
• Inactivation of the striatal direct and indirect pathways selectively disrupted the influence of past outcomes on action selection after unexpected events.

Conclusions:

• Unexpected outcomes trigger unique action-selection strategies distinct from those following expected outcomes.
• The striatum plays a critical role in arbitrating between different adaptive strategies for rapid learning (few-shot adaptation).
• The findings highlight specific roles for striatal direct and indirect pathways in modulating behavior based on outcome history and expectation.