Orientation-tuned surround suppression exhibits a unique laminar signature in human primary visual cortex

Abstract Spatial context modifies visual perception by enhancing novel and salient features over spatially redundant features in the underlying neural code of primary visual cortex (V1). Although multiple intracortical pathways contribute to contextual modulation, their specific contributions to different types of contextual modulation are not fully understood. Leveraging the distinct laminar connectivity patterns of feedforward, feedback, and lateral pathways, we used ultra-high-resolution fMRI (7T T 2 *-weighted, 0.6 mm isotropic resolution) to infer their relative contributions to contextual modulation in V1 by analyzing blood-oxygenation-level-dependent (BOLD) signal across cortical depth. Participants viewed sine-wave grating disks embedded in large surround gratings. Segmentation cues were introduced or removed by manipulating the relative phase and orientation of the surround gratings, yielding three contextual conditions and a surround-only condition to measure the effects of context in the absence of feedforward input. Our analysis isolated the effects of orientation-tuned surround suppression (OTSS) from orientation-independent border-induced modulation (BIM). The results show that BOLD laminar profiles differ by modulation type: OTSS was absent from deep layers, whereas BIM was more broadly distributed. We also find that voxels at all depths are driven by spatial context in the absence of feedforward input, which accords with the discovery of contextually-driven neural responses in mammalian V1. These laminar differences likely reflect different proportional contributions of feedback from higher-order visual areas and long-range lateral connections within V1. Our findings help to explicate the contributions of recurrent processing to visual contextual modulation and its impacts on laminar-dependent BOLD fMRI. Significance Statement Our sensory experience requires interpretation: the brain uses both hyper-local and large-scale scene cues to process sensory inputs. In primary visual cortex (V1), this contextual modulation arises from a mixture of intra- and inter-regional neural connections. Because these computations occur at different depths in the cortical gray matter, sub-millimeter resolution fMRI offers an opportunity to quantify the separate contributions of these neural pathways. Using a visual surround suppression paradigm, we measured a distinct depth profile for intra-regional contextual modulation, separating this orientation-tuned signal from other forms of border-induced modulation. This work validates an important new tool for understanding both typical and atypical neural network architectures in the human brain.