Temporal regularities shape perceptual decisions and striatal dopamine signals

Area of Science:

• Neuroscience
• Computational Neuroscience
• Animal Behavior

Background:

• Perceptual decisions are influenced by sensory evidence but also by past choices and outcomes.
• Choice history biases may represent adaptive strategies for exploiting temporal regularities in natural environments.
• The capacity for observers to adapt these biases to varying temporal regularities remains largely unexplored.

Purpose of the Study:

• To investigate whether and how observers adapt their perceptual choice history biases to different temporal regularities.
• To elucidate the computational principles and neural correlates underlying adaptive choice history biases.

Main Methods:

• Behavioral experiments with male mice exposed to visual stimuli with varying temporal regularities.
• Analysis of choice history biases, distinguishing between fast and slow components.
• Modeling using a normative reinforcement learning algorithm with multi-trial belief states.
• In vivo monitoring of dorsal striatal dopamine activity.

Main Results:

• Male mice demonstrated adaptation of perceptual choice history biases to different temporal regularities over hundreds of trials.
• A slow adaptive bias evolved over days, coexisting with a fast, non-adaptive bias.
• A reinforcement learning model accurately captured both fast and slow history effects.
• Dorsal striatal dopamine levels correlated with the model's predictions and observed behavior, indicating reward prediction signaling.

Conclusions:

• Perceptual choice history biases are adaptive and can be modified by temporal regularities in the environment.
• A normative reinforcement learning framework effectively explains the computational basis of these adaptive biases.
• Dorsal striatal dopamine plays a crucial role in signaling reward predictions associated with adaptive choice history biases.