Luca Kämmer^1,2,3 (), Lisa M. Kroell³, Tomas Knapen^4,5,6, Martin Rolfs^3,7,8,9, Martin N. Hebart^1,2,10; ¹Vision and Computational Cognition Group, Max Planck Institute of Human Cognitive and Brain Sciences, ²Department of Medicine, Justus Liebig University Giessen, Giessen, Germany, ³Department of Psychology, Humboldt-Universität zu Berlin, Berlin, Germany, ⁴Spinoza Center for Neuroimaging, KNAW Netherlands, ⁵Computational Cognitive Neuroscience and Neuroimaging, Netherlands Institute for Neuroscience, ⁶Experimental and Applied Psychology, Vrije University Amsterdam, Netherlands, ⁷Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Germany, ⁸Exzellenzcluster Science of Intelligence, Technische Universität Berlin, Germany, ⁹Bernstein Center for Computational Neuroscience Berlin, Germany, ¹⁰Center for Mind, Brain and Behavior, Universities of Marburg, Giessen, and Darmstadt, Germany

Humans sample their visual environment with frequent, rapid eye movements to make best use of the high resolution of their fovea. Despite the constant shifts in visual input, perception remains remarkably stable. Foveal cortex has been proposed as a key component in maintaining this stability by receiving feedback from cortical areas that encode peripheral information during saccade preparation. Here we aimed to uncover (1) whether foveal feedback can be found in early visual areas, (2) whether the information encoded in this activation is stimulus-content specific, and (3) which neural regions might be involved in driving this feedback. To dissociate neural processes elicited by direct visual input from those related to foveal feedback, we designed a gaze-contingent fMRI study where the saccade target is removed before it reaches the fovea. As targets, we used natural images and independently manipulated object shape and category. Even though targets were never presented in the fovea, the results showed reliable decoding from foveal parts of early visual cortex, suggesting feedback of feature-specific information during saccade preparation. Breaking the analysis down by stimulus type revealed a drop in decoding across category but not across shape, demonstrating that the classifier was sensitive to shape, but not semantic stimulus information. Cross-decoding to a control condition with foveal stimulus presentation was above chance, indicating a similar neural representation of foveal feedback signals to direct stimulus presentation. An eccentricity-dependent analysis revealed a dip in decoding between foveal and peripheral processing, indicating that the results are unlikely to be explained by spillover from peripheral cortex. Exploring the regions involved in modulating foveal feedback, we found a specific involvement of the intraparietal sulcus. By characterizing the neural representations underlying the processing of peripheral information in foveal retinotopic areas, these findings underscore the importance of foveal processing in active visual perception.

Acknowledgements: This work was supported by a doctoral scholarship of the German Academic Scholarship Foundation ("Studienstiftung"), a Max Planck Research Group grant, the ERC Starting Grant COREDIM and the Hessian Ministry of Higher Education, Science, Research and Art.