Keynotes | K&Ts | GACs | Talks | Posters | Search

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

The Architecture of Cortical Timing: From Local Clustering to Global Hierarchies

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

Presenter: Sara Varetti

Cortical circuits exhibit spontaneous activity characterized by a hierarchy of timescales that reflects the anatomical hierarchy of cortical regions (Murray et al., 2014). Such long-ranging intrinsic dynamics, which extend beyond the duration of typical sensory stimuli, are thought to facilitate the integration of information over multiple temporal scales. Stern et al. (2023) have shown that clustered network architectures can intrinsically generate heterogeneous timescales and support long temporal fluctuations when the local clustering of neurons is substantial. We propose that variations in clustering strength across cortical areas contribute to the observed hierarchy of intrinsic timescales. To test this, we model a feedforward chain of neural networks, each representing a local circuitry, endowed with neural assemblies (clusters). Within each local network, the clustering affects the autocorrelation of the local population activity, generating local timescales that are also transmitted downstream as temporally correlated (colored) input noise. We examine how the intra-areal structured activity is propagated and interacts with the downstream population, and under which conditions these interactions can give rise to a gradient of intrinsic timescales across the chain. Our analysis reveals that a hierarchy of intra-clustering strengths, corresponding to increased clustering in strength and size, is essential to reproduce the cortical gradient of intrinsic timescales. In contrast, stronger feedforward coupling reduces the diversity of timescales. Increased local gain beyond the necessary level for sustaining activity, or added uncorrelated noise, shortens timescales and hence also hinders the development of hierarchical timescales. Finally, analyses of neuronal activity recordings show that deeper cortical layers display stronger functional clustering and longer intrinsic timescales, consistent with our model’s predictions. Together, these results provide a framework linking local recurrent architecture with inter-areal connectivity to the emergence of hierarchical cortical timescales, offering insight into how transient stimuli are integrated and transformed into long neural trajectories needed for cognitive functions.

Topic Area: Methods, Tools, Theory & Neural Coding