Space, Time, and Fear: Survival Computations along Defensive Circuits

Area of Science:

• Neuroscience
• Ethology
• Computational Biology

Background:

• Predator avoidance behaviors are crucial for survival.
• These behaviors operate across diverse spatiotemporal scales.

Purpose of the Study:

• To investigate the distinct computational mechanisms underlying threat avoidance at different scales.
• To explore the neural basis of reflexive versus planned defensive actions.

Main Methods:

• Naturalistic observations of animal behavior.
• Analysis of computational architectures (model-free vs. model-based).
• Examination of neural circuitries involved in defensive behaviors.

Main Results:

• Proximal threats elicit reflexive, myopic actions supported by model-free computations.
• Distal threats allow for complex processing via model-based operations like planning and replay.
• Model-free and model-based systems interact, with safety periods enabling model-based learning for future reactions.

Conclusions:

• Defensive behaviors are orchestrated by distinct, yet interacting, computational systems.
• Neural circuits supporting threat responses are distributed and specialized.
• Effective policy determination for survival relies on integrating these computational approaches.