Neural circuits encode prior knowledge of temporal statistics

Area of Science:

• Neuroscience
• Computational Neuroscience
• Bayesian Inference

Background:

• The brain infers external states despite sensory uncertainty, relying on prior knowledge from environmental statistics.
• Bayesian inference theories explain this reliance, supported by behavioral and neuroscience data.
• Direct evidence for neural encoding of prior knowledge and environmental statistics is limited.

Purpose of the Study:

• To investigate if cerebellar circuits learn and encode prior probability distributions.
• To determine if Purkinje cells are involved in predictive motor behaviors reflecting learned statistics.
• To explore the neural mechanisms underlying Bayesian inference in the brain.

Main Methods:

• Eyeblink conditioning in mice to study temporal variable learning.
• Recording Purkinje cell simple and complex spike activity.
• Computational modeling of cerebellar plasticity mechanisms.

Main Results:

• Cerebellar circuits learn prior probability distributions of temporal variables.
• Purkinje cell signaling encodes these learned representations.
• Purkinje cells elicit predictive eyeblink responses reflecting stimulus statistics.

Conclusions:

• Cerebellar Purkinje cells acquire prior knowledge shaped by environmental statistics.
• The cerebellum may be crucial for learning and internalizing event probabilities.
• Findings advance understanding of neural computations implementing Bayesian inference.