Kurtkaya, B., Dinc, F., Yuksekgonul, M., Blanco-Pozo, M., Cirakman, E., Schnitzer, M., Yemez, Y., Tanaka, H., Yuan, P., & Miolane, N.
Abstract
Short-term memory is essential for cognitive processing, yet our understanding of its neural mechanisms remains unclear. A key focus in neuroscience has been the study of sequential activity patterns, where neurons fire one after another within large networks, to explain how information is maintained. While recurrent connections were shown to drive sequential dynamics, a mechanistic understanding of this process still remains unknown. In this work, we first introduce two unique mechanisms that can subserve short-term memory: slow-point manifolds generating direct sequences or limit cycles providing temporally localized approximations. Then, through analytical models, we identify fundamental properties that govern the selection of these mechanisms, \textit{i.e.}, we derive theoretical scaling laws for critical learning rates as a function of the delay period length, beyond which no learning is possible. We empirically verify these observations by training and evaluating more than 35,000 recurrent neural networks (RNNs) that we will publicly release upon publication. Overall, our work provides new insights into short-term memory mechanisms and proposes experimentally testable predictions for systems neuroscience.
Citation
Kurtkaya, B., Dinc, F., Yuksekgonul, M., Blanco-Pozo, M., Cirakman, E., Schnitzer, M., Yemez, Y., Tanaka, H., Yuan, P., & Miolane, N. (2025). Dynamical phases of short-term memory mechanisms in RNNs. ICML
BibTeX
@misc{kurtkaya2025dynamicalphasesshorttermmemory, title={Dynamical phases of short-term memory mechanisms in RNNs}, author={Bariscan Kurtkaya and Fatih Dinc and Mert Yuksekgonul and Marta Blanco-Pozo and Ege Cirakman and Mark Schnitzer and Yucel Yemez and Hidenori Tanaka and Peng Yuan and Nina Miolane}, year={2025} }
