Abstract
During our daily visual experience, our eyes constantly receive complex information from the environment. This information is characterized by spatiotemporal regularities, with predictable distributions of visual features across the visual field and across time. To create our unitary experience of reality, the brain needs to integrate these inputs in an efficient way. Here, we tested whether such integration processes are mediated by oscillatory neural codes. Specifically, we hypothesized that the integration of spatiotemporally regular information is governed by low-frequency oscillations in the alpha band that were previously linked to top-down modulations of sensory processing. In an EEG experiment, participants viewed short video clips (3s) depicting everyday situations, which were shown through circular apertures in the right and left visual fields. Videos were presented (1) through the right aperture only, (2) through the left aperture only, (3) through both apertures in a spatiotemporally congruent way, with the two apertures showing parts of the same video, or (4) through both apertures but in a spatiotemporally incongruent way, with the apertures showing parts of different videos. To quantify oscillatory activity, we first computed trial-wise EEG powerspectra during the video presentation. We then used multivariate classification analysis to decode the different videos in each condition from multi-electrode patterns of oscillatory power in three discrete frequency bands: alpha (8-12Hz), beta (13-30Hz) and gamma (31-70Hz). When videos were only presented in one hemifield, or when two inconsistent videos were presented, we could decode the videos from activity in the gamma range, indexing differences in feedforward visual processing. By contrast, we found that spatiotemporally consistent videos were primarily decodable from alpha activity, confirming our hypothesis that alpha oscillations mediate the dynamic integration of natural information into seamless visual experiences. Together, our results highlight differential oscillatory signatures for independent versus integrative processing of natural inputs.