For decades, neuroscientists have mapped the human visual cortex as a patchwork of specialized regions—one cluster that lights up for faces, another for places, another for body parts. But a provocative new study published in Nature Human Behaviour suggests these maps are not stamped by genetics alone. Instead, they are actively sculpted, moment by moment, by the very way our eyes dance across the world. The research reveals that the idiosyncratic gaze patterns each person uses to explore a scene are intimately woven into the brain’s category-selective architecture, offering a radical view of how active vision builds individuality into the brain.
The visual brain is famously home to regions like the fusiform face area (FFA), the parahippocampal place area (PPA), and the extrastriate body area (EBA). Each is defined by a sharp preference for one visual category over others. Classic models treat these patches as largely predetermined computational modules, with a fixed anatomical assignment that can be modified only across slow developmental or evolutionary timescales. Yet such models struggle to explain why the exact coordinates and selectivity strength of these areas vary so noticeably from one person to the next. The new work, led by Daniel Kollenda and colleagues, turns this puzzle on its head by asking a deceptively simple question: What if your brain’s face area is located exactly where your eyes most reliably land when a face appears?
To test this, the researchers combined high-resolution functional magnetic resonance imaging (fMRI) with precision eye-tracking while participants freely watched naturalistic movies and images. This setup allowed them to align the brain’s hemodynamic responses with the millisecond-accurate sequence of fixations, saccades, and microsaccades that constitute active vision. Crucially, they did not just correlate broad gaze measures with brain activity. They used population receptive field (pRF) models—tools that map each voxel’s spatial sensitivity—to warp the retinal input dynamically, computing the exact visual content that fell onto a participant’s fovea at every sampled moment. This created a personalized “retinotopic diary” of category exposure, specific to each individual’s looking habits.
The findings were striking. Within each observer, the spatial distribution of strongly category-tuned voxels did not simply fall into a generic group-averaged template. Instead, the neural selectivity landscape matched the statistical structure of that individual’s oculomotor habits. Participants who spent more time fixating on faces when faces appeared activated a more extensive and spatially distinct face-selective territory, with sharper tuning profiles, than those whose gaze drifted toward backgrounds or objects. A similar tight coupling held for place-selective cortex: the more an individual’s eyes sampled building edges, doors, or horizon lines, the more robust and precisely localized their PPA became. The brain’s category map, in other words, behaved less like a rigid blueprint and more like a statistical mirror of the observer’s perceptual sampling strategy.
Digging deeper, the team found that this link was not merely a byproduct of low-level visual features such as contrast or spatial frequency. Even after controlling for the local luminance and edge density landing on the retina during each fixation, the high-level categorical content—face versus house versus limb—accounted uniquely for the neural tuning strength. This suggests that semantic content itself, not just raw visual energy, is a driver of functional specialization, but only when actively selected for by the observer’s own gaze. Passive viewing conditions, in which participants were instructed to maintain central fixation, severely weakened this individual-level correlation, underscoring that visual sampling must be actively generated by the observer to sculpt selectivity.
The mechanism behind this phenomenon likely involves a form of Hebbian reinforcement at the network level. Fixating on a face sends a synchronized volley of bottom-up activity from early visual areas into face-preferring neurons, while simultaneously triggering top-down attentional signals that amplify that very circuit. Over time, neurons that consistently co-fire with foveated faces strengthen their mutual connections, sharpening the region’s category preference. Because each person foveates faces with subtly different biases—some fixate the eyes, others the mouth, still others dart between features—the resulting face area becomes a unique neural fingerprint. This implies that the brain’s stable functional architecture is, paradoxically, built upon the noisy, moment-to-moment decisions of the oculomotor system.
The research carries profound implications for understanding both typical cognition and clinical conditions. In disorders such as autism spectrum disorder, where face avoidance and atypical gaze patterns are well-documented, the new framework predicts that the altered oculomotor statistics could cascade into a differently organized fusiform gyrus, not as a primary deficit but as a dynamic adaptation to reduced foveal sampling of social stimuli. Conversely, perceptual learning therapies that train specific gaze strategies might directly remodel category-selective maps, opening avenues for noninvasive interventions. The work also challenges artificial intelligence researchers: convolutional neural networks trained on passively collected image datasets may miss a foundational principle of biological vision, namely that the training statistics of the human visual system are not random snapshots but are actively curated by the learner’s own motor output.
In a broader sense, the study reframes vision not as a camera passively receiving a world but as an embodied dialogue where each saccade writes a line of neural code. The individuality of our brain’s functional map may be less about the genes we inherit and more about the unique visual biography etched by a lifetime of looking. Active vision, once considered a mere delivery system for images, now emerges as a core architect of the mind’s internal categories.
Subject of Research: Active vision and category selectivity in the individual human brain
Article Title: Active vision is linked to category selectivity in the individual brain
Article References: Kollenda, D., Akbari, E., Broda, M.D. et al. Active vision is linked to category selectivity in the individual brain. Nat Hum Behav (2026). https://doi.org/10.1038/s41562-026-02494-5
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41562-026-02494-5
Keywords: active vision, category selectivity, fMRI, eye tracking, individual differences, visual cortex, retinotopy, population receptive fields, fusiform face area, parahippocampal place area
