Lateral phase differences in a population model of the visual cortex are sufficient for the development of rhythmic spatial sampling

Abstract

When attending to many spatially distributed visual stimuli, attention is reweighted rhythmically at 4-8 Hz. The probability of detection depends on the phase at which a stimulus is deployed relative to this intrinsic rhythm. The reweighting oscillations can be observed both electrophysiologically and behaviorally, and appear to be regulated by the pulvinar. Based on these findings, we considered the computational consequences of allowing feedback to shape the distribution of inhibitory oscillations from the thalamus, as measured by a local field potential (LFP) phases in the 8 Hz low alpha-band, across laterally-connected regions of the visual cortex. We constructed a population activity model with lateral and feedforward connections. In agreement with prior models, we found that the sign of the lateral phase difference in the inhibitory low-frequency oscillations regulated the direction of communication between the laterally-connected regions. Furthermore, the phase difference induced periodicity in the dynamics of a downstream winner-takes-all attractor network such that periodic switching between states was observed. We finally simulated a simple spatial attention task. We found rhythmic 8 Hz sampling between two regions when a lateral phase difference was present--an effect that disappeared when the lateral phase difference was zero. These findings are in agreement with spatial attention literature and suggest that lateral phase differences are essential for manifesting communicational asymmetries in laterally-connected visual cortices. Our model predicts that population-specific phase differences are critical for sampling the spatial extent of stimuli. Author summaryWe conducted a computational study of the effects of lateral phase differences in a simulated model of the visual cortex. Lateral phase differences are defined to be when the phase of an intrinsic low-frequency inhibitory oscillation varies consistently across populations in the same cortical area. For example, our model was intended to capture the dynamics of a retinotopic cortex where feedback from the frotoparietal areas via the pulvinar nucleus assigned laterally-connected regions of the visual cortex different phases. We found that the sign of the phase differences influenced the direction of lateral communication. Furthermore, the phase differences introduced rhythmicity in the downstream areas, thus allowing us to simulate rhythmic spatial selection of stimuli. Prior to the current study, the influence of inter-areal phase differences in feedforward models had been well characterized. Our model provides new insights into the dynamics of population-specific lateral phase differences and predicts that the development of phase differences across the visual cortex are critical for the allocation of attention in space.