Visual representations of faces, places, and objects are encoded within widely-distributed and partially individualized neural "fingerprints"

titleVisual representations of faces, places, and objects are encoded within widely-distributed and partially individualized neural "fingerprints"
start_date2025/12/15
schedule15h15
onlineno
location_infoRoom B10
summaryA long-standing view in cognitive neuroscience holds that neural representations of visually-perceived objects are encoded by category-specific modules within occipito-temporal regions that vary only slightly in their precise location across participants. Yet several recent studies deploying multivariate pattern classification suggest more widely distributed and graded neural representations that vary substantially in location and code across individuals. Such results may reflect spurious correlations that arise due stimulus associations that are idiosyncratic to a specific person or even to the immediate testing context, without playing any causal role in perception/representation. Alternatively, they may indicate that visual representations of objects depend on coalitions of neural populations distributed across cortex, in ways that vary substantially across individuals. To adjudicate these views, I will describe new work using whole-brain decoding to map how and where information about faces, places, and objects is encoded within individual participants. The results suggest that regions well outside canonical areas encode information about stimulus category, with substantial similarity across subjects in the classic face/place system, but highly variable locations and codes elsewhere in the brain. To assess the stability of this distributed code, we compare results within subject across sessions; and to assess its specificity to a given person, we compare results across subjects in the same session. Finally, to evaluate whether the distributed and individualized elements of the neural code play a causal role in perceptual decisions, we compare the effects of transcranial magnetic stimulation (TMS) on visual decisions about faces, applying the stimulation to the occipital face area (a common region shared across subjects) and to an out-of-system region bespoke to each participant based on their imaging analysis. The results provide strong evidence that visual representations depend on interactions within a widespread and partially individualized cortical network, with important implications for how we think about neuro-cognitive representation more generally.
responsiblesAllen