Cross-scale dynamics in the working regime of the visual cortex

Abstract

The visual cortex, bombarded with input from a complex environment, must rapidly detect small changes while maintaining stable, persistent representations. From rapid action potentials to slow traveling waves, dynamical structures allow the brain to process information efficiently, and coupling these structures across scales enhances computational flexibility. Mounting evidence also suggests the brain operates near a critical point, where activity spans all scales and information capacity is maximized.

How do the cross-scale dynamics of the brain balance the efficiency of scale-specific structure with the flexibility of cross-scale interactions? We approach this question using new analytical tools, theoretical frameworks, biophysical models, and large-scale electrophysiological data from the mouse visual cortex. First, in Chapter 2, we reveal how cross-scale interactions between broad traveling waves, local oscillatory bursts, and single-neuron spiking support flexible hierarchical computation. Second, in Chapter 3, we develop a practical metric for detecting criticality in noisy, poorly sampled neural data that suggests higher areas of the visual hierarchy are positioned closer to criticality. Third, in 4, we ask what circuit-level regime reconciles structured dynamics with scale-free stochasticity. We find that both the mouse visual cortex and a biophysical circuit model exhibit non-Gaussian superdiffusion, subdiffusive long-range memory, and collective oscillations, which we capture using an empirical mean-field model based on the formalism of fractional calculus. Thus, across three chapters, we elucidate how the visual cortex leverages cross-scale spatiotemporal dynamics alongside anomalous scale-free stochasticity to enable flexible processing of complex visual information.