Spatiotemporal Dynamics Underlying Object Completion in Human Ventral Visual Cortex

Abstract

Summary Natural vision often involves recognizing objects from partial information. Recognition of objects from parts presents a significant challenge for theories of vision because it requires spatial integration and extrapolation from prior knowledge. Here we recorded intracranial field potentials of 113 visually selective electrodes from epilepsy patients in response to whole and partial objects. Responses along the ventral visual stream, particularly the Inferior Occipital and Fusiform Gyri, remained selective despite showing only 9–25% of the object areas. However, these visually selective signals emerged ~100 ms later for partial versus whole objects. These processing delays were particularly pronounced in higher visual areas within the ventral stream. This latency difference persisted when controlling for changes in contrast, signal amplitude, and the strength of selectivity. These results argue against a purely feed-forward explanation of recognition from partial information, and provide spatiotemporal constraints on theories of object recognition that involve recurrent processing.

Beyond the feedforward sweep: feedback computations in the visual cortex

“…These behavioral measurements are consistent with the latencies reported in neurophysiological recordings during pattern completion. The latency of neurophysiological signals in areas V4 and the inferior temporal (IT) cortex in response to heavily occluded objects is delayed by about 50 ms with respect to the responses of the same circuits to the fully visible objects 84,85 . These behavioral and neurophysiological observations are further corroborated by computational models: state‐of‐the‐art bottom‐up models struggle during recognition of heavily occluded objects, unless they are extensively trained with those specific occluded objects 86,87 …”

“…During natural visual conditions, many objects are partially visible either because they are occluded by other objects in front of them or because of poor illumination or because of unusual viewing angles. Despite such challenging visual conditions, primate visual recognition is quite robust even when up to 90% of the object is occluded, even in the absence of contextual cues, and even when subjects have minimal prior experience with the object in question 84 …”

“…Despite such challenging visual conditions, primate visual recognition is quite robust even when up to 90% of the object is occluded, even in the absence of contextual cues, and even when subjects have minimal prior experience with the object in question. 84 Behavioral, neurophysiological, and computational evidence suggest that purely bottom-up computations are generally insufficient to perform pattern completion of heavily occluded objects. At the behavioral level, recognition of heavily occluded objects takes longer than the recognition of the whole object counterparts.…”

Stimulus features coded by single neurons of a macaque body category selective patch

“…9-13). Although this technique has been used in neuroimaging [functional MRI (fMRI), EEG, magnetoencephalography (MEG), and electrocorticography] studies (11,12,14), it has rarely been exploited in single-unit studies (15,16), and this only for face stimuli.…”

Healthy aging delays the neural processing of face features relevant for behavior by 40 ms

“…MI scaled , therefore, reflects a measure of MI adjusted for a systematic upward bias in the information estimate that might arise due to limited data sampling, especially if the numbers of trials in the two age groups are systematically different. It also converts MI to be the effect size for a log-likelihood test of independence (Sokal & Rohlf, 2012). All group-difference analyses were performed using the scaled MI values.…”