Flexible route planning and rapid structure learning by mice in complex environments.

Area of Science:

• Neuroscience
• Computational Biology
• Animal Behavior

Background:

• Action selection relies on predictive environmental models, crucial for behavior but poorly understood at circuit/algorithmic levels.
• Spatial navigation involves multiple systems (habits, vector-navigation, route planning), necessitating assays to dissociate these for studying world models.

Purpose of the Study:

• To develop and optimize a computational behavioral assay for quantifying flexible navigation based on environmental structure knowledge.
• To enable precise quantification of brain-behavior relationships in spatial navigation.

Main Methods:

• Mice navigated complex mazes with randomized start/goal locations across trials.
• Developed a computationally optimized behavioral assay to generate large datasets of non-repetitive navigation trajectories.
• Focused on quantifying efficient path selection and rapid learning of maze structure.

Main Results:

• Mice demonstrated efficient navigation, consistently favoring shortest paths to goals.
• Mice exhibited rapid learning, acquiring knowledge of maze structure from initial sessions.
• The assay generated thousands of non-repetitive, goal-directed trajectories for analysis.

Conclusions:

• The developed assay effectively quantifies flexible navigation using environmental structure knowledge.
• The findings provide a valuable tool for investigating how world models support adaptive behaviors.
• This research advances understanding of the neural basis of spatial cognition and action selection.