Reverse engineering the control law for schooling in zebrafish using virtual reality

Area of Science:

• Animal behavior
• Robotics
• Neuroscience

Background:

• Understanding collective behavior in biological systems is crucial.
• Quantifying social interactions is key to identifying social deficits.
• Zebrafish (Danio rerio) are a model organism for studying social behavior.

Purpose of the Study:

• To reverse-engineer the sensory-motor control of social response during zebrafish schooling.
• To test models of social interactions in situ using virtual reality.
• To evaluate the potential of an evolved control law for autonomous systems.

Main Methods:

• Utilized virtual reality technology with virtual robot fish to study juvenile zebrafish.
• Developed a networked system allowing real fish to interact in a shared virtual environment.
• Applied a Turing test and scalability test to validate the derived control law.

Main Results:

• Identified single- and multitarget-oriented pursuit as a key feature of social response.
• Demonstrated that zebrafish behavior is accurately modeled by a proportional derivative control law (BioPD).
• Showed that the BioPD law is robust to incomplete sensory input and based on egocentric positional information.

Conclusions:

• The evolved control law 'BioPD' fully accounts for key features of zebrafish social pursuit behavior.
• The BioPD law offers a simple yet effective model for social interactions.
• This evolved control law demonstrates potential for optimizing pursuit performance in autonomous vehicles across various domains.