Separating Probability and Reversal Learning in a Novel Probabilistic Reversal Learning Task for Mice.

Area of Science:

• Neuroscience
• Cognitive Science
• Animal Behavior

Background:

• The exploration/exploitation tradeoff is crucial for learning and decision-making.
• Existing probabilistic reversal learning (PRL) tasks may exceed mice's cognitive abilities.
• A novel task is needed to study this tradeoff in mice.

Purpose of the Study:

• To develop a new probabilistic learning task for mice.
• To investigate the exploration/exploitation tradeoff in rodent decision-making.
• To model mouse choice behavior using reinforcement learning.

Main Methods:

• Developed a two-lever operant chamber task with varying saccharin reward probabilities.
• Reversed reward contingencies when mice reached 80% preference for the high-reward lever.
• Applied reinforcement learning (RL) models to analyze animal choice behavior.

Main Results:

• Mice successfully learned and performed near-optimally with 80%/20% reward probabilities.
• Some mice showed lever preference even with near-equal probabilities (60%/40%).
• RL models effectively captured mouse decision-making, including learning rates and noise.

Conclusions:

• The new task enables effective study of exploration/exploitation in mice.
• This approach advances the neuroscience of decision-making and learning in rodents.
• Reinforcement learning models provide valuable insights into mouse behavior.