2025-01-29 コロンビア大学
Image credit: m-sato@Tii
<関連情報>
- https://www.engineering.columbia.edu/about/news/harnessing-chaos-how-brain-turns-randomness-robust-memory
- https://www.pnas.org/doi/10.1073/pnas.2316745122
ランダムノイズはロバストワーキングメモリ計算に重要な低速異種シナプスダイナミクスを促進する Random noise promotes slow heterogeneous synaptic dynamics important for robust working memory computation
Nuttida Rungratsameetaweemana, Robert Kim, Thiparat Chotibut, and Terrence J. Sejnowski
Proceedings of the National Academy of Sciences Published:January 16, 2025
DOI:https://doi.org/10.1073/pnas.2316745122
Significance
Our study addresses a fundamental challenge in the field of neural network modeling, offering critical insights into the training of recurrent neural networks (RNNs) that simulate cognitive processes. Specifically, we demonstrate that by introducing random noise during training, we not only expedite the learning process but also establish robust models capable of maintaining information necessary for working memory tasks. Further analyses revealed that the introduction of noise selectively increased the synaptic decay time constants of inhibitory units, leading to a sustained stimulus-specific activity crucial for memory maintenance. Our findings not only shed light on the optimization of RNN training methods but also hold profound implications for understanding how higher cortical areas evolved to compensate for inherent noise to maintain information.
Abstract
Recurrent neural networks (RNNs) based on model neurons that communicate via continuous signals have been widely used to study how cortical neural circuits perform cognitive tasks. Training such networks to perform tasks that require information maintenance over a brief period (i.e., working memory tasks) remains a challenge. Inspired by the robust information maintenance observed in higher cortical areas such as the prefrontal cortex, despite substantial inherent noise, we investigated the effects of random noise on RNNs across different cognitive functions, including working memory. Our findings reveal that random noise not only speeds up training but also enhances the stability and performance of RNNs on working memory tasks. Importantly, this robust working memory performance induced by random noise during training is attributed to an increase in synaptic decay time constants of inhibitory units, resulting in slower decay of stimulus-specific activity critical for memory maintenance. Our study reveals the critical role of noise in shaping neural dynamics and cognitive functions, suggesting that inherent variability may be a fundamental feature driving the specialization of inhibitory neurons to support stable information processing in higher cortical regions.
