Distinct timescales of recurrent dynamics in frontal and visual cortices

Recurrent dynamics are thought to be central to computation in the cerebral cortex. Yet, how these dynamics differ across areas remains unknown. Here, we used two-photon microscopy to image Ca2+ activity from hundreds of neurons and while delivering brief, focal excitatory input with near-cellular- resolution optogenetics. Specifically, we compared how the primary visual (V1) and the secondary motor (M2) cortices distinctly sustain this input while mice ran spontaneously. Single-neuron responses to focal stimulation were much more prevalent in M2 than V1. Moreover, evoked network-level dynamics in M2 outlasted that in V1 by seconds, largely due to the consistent recruitment of seconds-long activity sequences in M2. Further, using principal components analysis, we found that this consistency depended on the initial state of the network at the time of focal photostimulation. Interestingly, in both areas, the pattern of single-neuron responses to stimulation depended on the spontaneous activity timescales of those neurons. Finally, using two-photon holography, we observed that increasing the number of simultaneously stimulated neurons led to an increasing fraction of early responding cells in V1, while M2 neurons continued responding with long-lasting sequences. Thus, layer 2/3 circuits in these two cortical regions exhibit distinct response properties to focal input, likely due to stronger recurrent interactions within M2. While this had been predicted by morphological and transcriptomic gradients, to our knowledge this is the first in-vivo, functional demonstration of qualitatively different recurrent dynamics across the cortex. Crucially, these differences likely shape how cortical regions are recruited during behavior, highlighting the specialized roles of visual and frontal areas in processing information. Our findings provide valuable insights into the organization of cortical networks and set the stage for developing more biologically realistic models of cortical dynamics.