Zebrafish capable of generating future state prediction error show improved active avoidance behavior in virtual...

Area of Science:

• Neuroscience
• Animal Behavior
• Decision Making

Background:

• Animals employ reward maximization and surprise minimization for decision-making.
• Neural mechanisms underlying these principles and their behavioral manifestations remain largely unknown.

Purpose of the Study:

• Investigate how reward value maximization and surprise minimization are represented in the brain and behavior.
• Examine neural ensembles in zebrafish dorsal pallium during active avoidance learning.

Main Methods:

• Utilized a closed-loop virtual reality system for training adult zebrafish.
• Analyzed neural activity in the dorsal pallium during active avoidance tasks.
• Correlated neural ensemble activation with behavioral escape efficiency.

Main Results:

• Identified neural ensembles in the dorsal pallium that assign rules to environmental cues (wall colors).
• Discovered a distinct neural ensemble activated by discrepancies between perceived and predicted scenery (prediction error).
• Zebrafish with the prediction error ensemble exhibited more efficient escape behavior than those with only rule-assigning ensembles.

Conclusions:

• Zebrafish utilize both reward maximization and surprise minimization principles in goal-directed behavior.
• The prediction error ensemble enhances escape efficiency by guiding actions to minimize discrepancies.
• Behavioral outcomes are influenced by the specific combination of cognitive principles employed by the animal.