Mice infer probabilistic models for timing.

Area of Science:

• Neuroscience
• Animal Behavior
• Cognitive Science

Background:

• Animals learn reward timing, but traditional models assume fixed uncertainty.
• Natural environments present variable reward schedules, challenging fixed-uncertainty models.
• It remains unclear if animals can infer probabilistic models for reward timing.

Purpose of the Study:

• To investigate if mice can infer probabilistic models of reward timing distributions.
• To determine if mice adjust behavior based on the statistical properties (e.g., standard deviation) of reward delays.
• To assess the flexibility of mice in using prior information for reward timing inference.

Main Methods:

• Developed a novel behavioral paradigm requiring knowledge of reward delay distributions.
• Trained mice in this paradigm to assess their ability to infer timing statistics.
• Measured behavioral adjustments in response to varying standard deviations of reward delays.

Main Results:

• Mice accurately adjusted their behavior to the standard deviation of the reward delay distribution.
• Mice demonstrated flexible inference, adapting prior information use to task demands.
• Behavioral adjustments indicated an inference of the underlying reward timing probabilities.

Conclusions:

• Mice can infer probabilistic models for reward timing, going beyond fixed uncertainty.
• This ability allows for flexible adaptation to dynamic and complex natural environments.
• Findings suggest a sophisticated mechanism for temporal inference in animals.