A Mechanistic Model for Reward Prediction and Extinction Learning in the Fruit Fly.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Insect Models

Background:

• Extinction learning is crucial for updating maladaptive memories, with clinical implications for therapies.
• Insect models, particularly *Drosophila melanogaster*, offer insights into memory formation and update mechanisms.
• Recent findings suggest parallel, opposing memory traces in the *Drosophila* mushroom body (MB) after memory extinction.

Purpose of the Study:

• To propose a minimalistic circuit model of the *Drosophila* MB.
• To simulate classical appetitive and aversive conditioning and memory extinction.
• To integrate existing anatomical data and explore circuit motifs.

Main Methods:

• Developed a circuit model with plastic synaptic connections between Kenyon cells (KCs) and MB output neurons (MBONs).
• Incorporated mutually inhibiting appetitive and aversive learning pathways.
• Modeled recurrent plasticity modulation via MBON projections to dopaminergic neurons (DANs) for reward prediction.

Main Results:

• The model successfully reproduced experimental results from learning and extinction protocols.
• Simulations of neuronal output blockade confirmed experimental findings and generated testable predictions.
• The model demonstrates rapid learning, with a step-like increase in odor value after single-trial conditioning.

Conclusions:

• The proposed model supports classical conditioning and extinction, consistent with *Drosophila* MB function.
• It elucidates the coexistence and interaction of parallel memory traces.
• The model facilitates understanding of single-trial learning and provides a framework for future research.