Correlated activity fluctuations in neocortex influence sensory responses and behavior. Neural correlations reflect anatomical connectivity and change dynamically with cognitive states, such as attention. Yet, how anatomical connectivity and cognitive states define the population structure of correlations is not known. We measured correlations in single cortical columns and found that the magnitude of correlations, their attentional modulation and dependence on lateral distance are predicted by On-Off dynamics, synchronous fluctuations in population activity across cortical layers. We developed a network model, in which spatial connectivity correlates the On-Off dynamics across nearby columns. We show that attentional inputs modulate the spatial extent of On-Off dynamics, resulting in spatially non-uniform changes in correlations. We confirm this prediction in our columnar recordings by showing that attentional modulation of correlations depends on lateral distance. Our results reveal how heterogeneous spatial patterns of correlations arise from the connectivity and network dynamics during attention.
bioRxiv Subject Collection: Neuroscience