Partial information transfer from peripheral visual streams to foveal visual streams may be mediated through local primary visual circuits

Visual object recognition is driven through the what pathway, a hierarchy of visual areas processing features of increasing complexity and abstractness. The primary visual cortex (V1), this pathway's origin, exhibits retinotopic organization: neurons respond to stimuli in specific visual field regions. A neuron responding to a central stimulus will not respond to a peripheral one, and vice versa. However, despite this organization, task-relevant feedback about peripheral stimuli can be decoded in unstimulated foveal cortex, and disrupting this feedback impairs discrimination behavior. The information encoded by this feedback remains unclear, as prior studies used computer-generated objects ill-suited to dissociate different representation types. To address this knowledge gap, we investigated the nature of information encoded in periphery-to-fovea feedback using real-world stimuli. Participants performed a same/different discrimination task on peripherally displayed images of vehicles and faces. Using fMRI multivariate decoding, we found that both peripheral and foveal V1 could decode images separated by low-level perceptual models (vehicles) but not those separated by semantic models (faces). This suggests the feedback primarily carries low-level perceptual information. In contrast, higher visual areas resolved semantically distinct images. A functional connectivity analysis revealed foveal V1 connections to both peripheral V1 and later-stage visual areas. These findings indicate that while both early and late visual areas may contribute to information transfer from peripheral to foveal processing streams, higher-to-lower area transfer may involve information loss.