Speaker:
Zac Davis
Even under the most controlled stimulus conditions, an identical sensory stimulus will evoke different spiking responses each time it is presented within a cortical neuron's receptive field. Understanding the sources, nature, and consequences of neural response variability is critical to understanding how the brain processes sensory information. Spike variability is not due to noise in the spiking mechanism itself as neurons in vitro can emit nearly identical spike trains when stimulated with identical fluctuating currents. Rather, a major source of variability is fluctuations in synaptic activity arriving through the highly recurrent organization of the cortical networks in which each neuron is embedded. Using newly developed techniques to characterize the moment-to-moment dynamics of noisy multi-electrode data recorded from Area MT in common marmosets, we find the functional organization of cortical networks gives rise to intrinsic traveling waves (iTWs) of cortical activity. These waves regulate both the gain of stimulus-evoked spiking responses and the monkey's perceptual sensitivity during a visual detection task. In a large-scale spiking network model we find that iTWs naturally emerge from propagation delays in topographically organized recurrent connections similar to the horizontal fibers in cortex. The model predicts that iTWs are sparse, in the sense that only a small fraction of the neural population participates in any individual iTW. As a result, iTWs can occur without inducing correlated variability, which can impair sensory discrimination. The model also predicts, based on horizontal fiber connectivity principles, that iTWs fall into feature-selective motifs whose selectivity stems from the preference for connections among similarly tuned feature domains. Consistent with this prediction, we find wave motifs that modulate both neural activity and perceptual sensitivity in a feature-selective manner. Taken together, these findings suggest that traveling waves may reflect the inferential nature of sensory processing whereby intrinsic network activity shapes weak or ambiguous sensory input to construct our percept of the external world.