Long-term memory performance optimization via Neural network-based curve fitting in Drosophila

Authors

Lu, Y.-C., Chen, C.-Y., Yen, L.-H., Yang, C.-L., Liu, Y.-D., Chen, W.-J., Feng, K.-L., Wu, M.-C., Chiang, A.-S., Yao, D.-J., Ho, C.-M., Chiou, S.-H., Chu, L.-A.

Abstract

Long-term memory (LTM) formation typically requires extensive training. While operant conditioning is expected to produce stronger LTM than classical conditioning due to active learning, laser-based social conditioning in Drosophila yielded an unexpected discrepancy: operant paradigms produced higher short-term memory (STM) but rapid LTM decay, whereas classical paradigms maintained higher LTM. To resolve this, we applied the AI Complex Systems Response (AI-CSR) framework, which reconstructs high-dimensional learning landscapes to predict globally optimal training conditions. AI-CSR optimization doubled operant LTM scores, yielding the strongest 24-hour social memory reported in flies and confirming the superiority of active learning previously obscured by standard protocols. Conversely, AI-CSR halved classical conditioning training time without altering LTM performance. Finally, single-cell RNA sequencing revealed expanded neuronal recruitment marked by distinct gene activation and inhibition profiles. Together, these findings link circuit-level reorganization with the molecular programs underlying efficient LTM, demonstrating how AI-guided optimization can uncover latent learning capacities in biological systems.

Preprint server: bioRxiv
The authors list and abstract were imported from bioRxiv on 26 Jun 2026.

Stats

1-terrible, 9-excellent. How would you rate this preprint? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.