State-dependent modulation of dependencies in a laminar cortical circuit

Adaptive information processing is crucial for flexible brain function and spatial attention is a quintessential example of this process. It is crucial for recognizing and interacting with behaviorally relevant objects in a cluttered environment. How the deployment of spatial attention aids hierarchical computations of object recognition is unclear. Two key mechanisms have been proposed: First is an improvement in the efficacy of information transfer, supported by an attentional increase in inter-areal correlations along the hierarchy. A second proposal is an improvement in information capacity through a reduction of shared fluctuations, supported by attentional decrease in correlations within a visual area. Since pairwise correlations capture both directed and shared components of fluctuations, it is unclear if they support the proposed mechanisms. To test these proposals, it is crucial to estimate the attentional modulation of directed information across and shared information within the stages of visual hierarchy. Using network-based statistical modeling of laminar activity in visual area V4, we estimated the strength of laminar directed dependencies. We quantified their modulation across attention conditions (attend-in vs. attend-away) in a change detection task and found that they are indeed strengthened between the input and superficial layers. Using partial information decomposition framework, we estimated modulation of shared dependencies and found that they are reduced within laminar populations. Surprisingly, we found a strengthening of directed dependencies within the laminar populations, a finding not previously predicted. Crucially, this modulation pattern was also observed across behavioral outcomes (hit vs. miss) that are mediated by endogenous state fluctuation. By “decomposing” the modulation of dependency components and in combination with prior theoretical work, our results suggest a computational model - enhanced information flow between and improved information capacity within encoding stages - of optimal sensory states that are attained by either task demands or endogenous fluctuations in brain state.