The brain processes visual information across diverse spatial and temporal scales, but the dynamics that bridge these scales remain poorly understood. Here, we analyze Neuropixels recordings from mouse visual cortex during a visual change-detection task and uncover a nested spatiotemporal structure linking low-frequency θ, high-frequency γ, and neuronal spiking across cortical layers and areas. θ forms broad traveling waves that flexibly switch direction across a stimulus presentation. Following stimulus onset, top-down θ propagates from deep to superficial layers and from higher- to lower-order areas, but after stimulus offset, bottom-up θ travels in the reverse direction. These nonstationary θ waves predict behavioral responses to changes in stimulus content. In contrast, γ activity appears as brief, spatially localized packets that sharpen with stimulus onset. Extending conventional cross-frequency coupling analyses, we show that both γ packets and neuronal spiking are phase-locked to the traveling θ waves. This cross-scale coupling exhibits systematic gradients across layers and areas aligned with the visual hierarchy. Together, these findings reveal a spatiotemporal θ-γ neural code that multiplexes bottom-up and top-down information, linking cross-scale dynamics to adaptive hierarchical visual processing.
Harris et al. (Fri,) studied this question.