A plausible neural circuit for decision making and its formation based on reinforcement learning

Area of Science:

• Neuroscience
• Computational Biology
• Insect Behavior

Background:

• Animal behavior is governed by complex neural circuits.
• Understanding the neural basis of deterministic behaviors like insect phototaxis is crucial.
• Existing models often lack detailed consideration of biological neuron characteristics.

Purpose of the Study:

• To design a plausible neural circuit model for insect phototactic flight.
• To simulate information processing within the circuit and its control over flight behavior.
• To explore neural decision-making mechanisms using a reward-punishment feedback model.

Main Methods:

• Designed a neural circuit model incorporating firing characteristics of biological neurons (excitatory and inhibitory).
• Simulated circuit information processing using a distributed PC array.
• Drove a flying behavior simulation using real-time average firing rates of output neuron clusters.

Main Results:

• Developed a computational neural circuit that logically replicates phototactic behavior rules.
• Simulations demonstrated how output neuron firing rates control angular velocity for flight.
• Explored cooperative multi-neuron control and information encoding mechanisms.

Conclusions:

• The designed neural circuit successfully models insect phototaxis based on biological neuron principles.
• The model's generality allows for designing behavioral logic rules from general neural processing modes.
• This study establishes a link between microscopic neural activity and macroscopic animal behavior.