Poster Presentation

Poster F in Poster Session F: Thursday, August 6, 1:45 – 3:30 pm, Kimmel Center, Shorin & Rosenthal Rooms

Long-lasting working memory emerges from the dynamics of hierarchical neural assemblies

Sara Varetti¹, Marco Zenari², Merav Stern³; ¹International Higher School for Advanced Studies Trieste, ²Flatiron Institute, ³Rockefeller University

Presenter: Merav Stern

Working memory is essential to many cognitive functions and requires a persistent neural signal that can maintain information over time. Classical models associate memory persistence with fixed firing rates (e.g., Hopfield 1982, Amit and Brunel 1997, Rmasauer et al. 2020) in discrepancy with the dynamic nature of cortical activity (Romo et al. 1999, Stokes et al. 2013, Murray 2017, Buschman and Miller 2022, Pereira-Obilinovic et al. 2023). Alternative frameworks propose that working memory is maintained through trajectories of population activity (Laje and Buonomano 2013, Rajan et al. 2016, Schuecker et al. 2018). However, these models often lack clear mechanisms for information readout, and achieving long memory timescales requires training (Barak et al. 2013, Cueva et al. 2020, Masset et al. 2025). Here, we show that neural circuits with clustered connectivity, where neurons are organized into assemblies with strong intra-assembly connections, naturally generate long intrinsic timescales that support working memory without the need for parameter tuning. We demonstrate that neural assemblies can store transient inputs within their baseline population activity, which enables reliable decoding long after the input has ceased. This architecture allows individual neurons to remain dynamic and responsive, permitting persistent memory signals to coexist with complex, time-varying activity. We further examine how information propagates through circuits with progressively stronger clustering. We find that memory can be transferred across such a chain, remaining decodable in downstream circuits long after it is no longer accessible in the initially stimulated neurons. This architecture supports temporal filtering, robustness to noise, and sensitivity to new inputs, while preserving memory over long timescales. Because clustered assemblies are common in cortical organization, our results suggest that working memory is an inherent property of cortical circuits. Together, these findings provide a circuit-level mechanism for working memory that reconciles persistence, dynamic activity, and efficient readout within a biologically plausible architecture.

Topic Area: Memory, Learning & Knowledge Structures