TY - JOUR
T1 - Visual cortical processing—From image to object representation
AU - von der Heydt, Rüdiger
N1 - Funding Information:
I wish to thank Ernst Niebur for complementing my neurophysiology with computational neuroscience; Fangtu T. Qiu for creating a powerful and versatile system for visual stimulus generation, behavioral control, and recording; and Ofelia Garalde who as an animal lab technician contributed a lot to the experimental success of the 14 neurophysiological studies reviewed here.
Publisher Copyright:
Copyright © 2023 von der Heydt.
PY - 2023
Y1 - 2023
N2 - Image understanding is often conceived as a hierarchical process with many levels, where complexity and invariance of object representation gradually increase with level in the hierarchy. In contrast, neurophysiological studies have shown that figure-ground organization and border ownership coding, which imply understanding of the object structure of an image, occur at levels as low as V1 and V2 of the visual cortex. This cannot be the result of back-projections from object recognition centers because border-ownership signals appear well-before shape selective responses emerge in inferotemporal cortex. Ultra-fast border-ownership signals have been found not only for simple figure displays, but also for complex natural scenes. In this paper I review neurophysiological evidence for the hypothesis that the brain uses dedicated grouping mechanisms early on to link elementary features to larger entities we might call “proto-objects”, a process that is pre-attentive and does not rely on object recognition. The proto-object structures enable the system to individuate objects and provide permanence, to track moving objects and cope with the displacements caused by eye movements, and to select one object out of many and scrutinize the selected object. I sketch a novel experimental paradigm for identifying grouping circuits, describe a first application targeting area V4, which yielded negative results, and suggest targets for future applications of this paradigm.
AB - Image understanding is often conceived as a hierarchical process with many levels, where complexity and invariance of object representation gradually increase with level in the hierarchy. In contrast, neurophysiological studies have shown that figure-ground organization and border ownership coding, which imply understanding of the object structure of an image, occur at levels as low as V1 and V2 of the visual cortex. This cannot be the result of back-projections from object recognition centers because border-ownership signals appear well-before shape selective responses emerge in inferotemporal cortex. Ultra-fast border-ownership signals have been found not only for simple figure displays, but also for complex natural scenes. In this paper I review neurophysiological evidence for the hypothesis that the brain uses dedicated grouping mechanisms early on to link elementary features to larger entities we might call “proto-objects”, a process that is pre-attentive and does not rely on object recognition. The proto-object structures enable the system to individuate objects and provide permanence, to track moving objects and cope with the displacements caused by eye movements, and to select one object out of many and scrutinize the selected object. I sketch a novel experimental paradigm for identifying grouping circuits, describe a first application targeting area V4, which yielded negative results, and suggest targets for future applications of this paradigm.
KW - computational model
KW - figure ground organization
KW - neural mechanism
KW - object individuation
KW - object permanence
KW - selective attention
KW - spiking synchrony
KW - visual cortex
UR - http://www.scopus.com/inward/record.url?scp=85164529636&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85164529636&partnerID=8YFLogxK
U2 - 10.3389/fcomp.2023.1136987
DO - 10.3389/fcomp.2023.1136987
M3 - Review article
AN - SCOPUS:85164529636
SN - 2624-9898
VL - 5
JO - Frontiers in Computer Science
JF - Frontiers in Computer Science
M1 - 1136987
ER -