Summary: Neurons located in the deep layers of the cerebral cortex are the first to signal which side of a visual border belongs to an object and which side is background.
Source: Salk Institute
In the classic Rubin’s vase illusion, the same image can be seen either as a decorative vase or as two faces in profile. Which interpretation dominates depends on whether the brain treats the central shape as foreground or background.
Researchers led by Professor John Reynolds and Senior Postdoctoral Fellow Tom Franken have advanced our understanding of how the brain assigns “border ownership” — that is, how it determines which side of a contour belongs to an object and which side belongs to the background.
Published in the journal eLife on November 30, 2021, this study clarifies how different cortical layers communicate to create a coherent visual representation of the world.
“How the brain organizes and generates an internal model of the outside world remains one of the major open questions in neuroscience,” says Reynolds, holder of the Fiona and Sanjay Jha Chair in Neuroscience. “Our findings offer key insights into the neural computations that assign borders to objects, which could help explain perceptual disturbances in conditions such as schizophrenia.”
When you look at a scene, individual cortical neurons receive information from very small regions of the visual field. Neurons responding at an object’s edge receive limited local cues and therefore lack immediate context about which side is foreground. Previous studies, however, identified a class of neurons that very rapidly encode border ownership — a cue crucial for depth perception and for distinguishing objects from background features like shadows or texture.
Two broad mechanisms had been proposed for how border ownership might be computed. The feedforward hypothesis suggests that as visual signals progress through successive, deeper cortical areas, each stage performs additional computations until a stable scene representation emerges. The feedback hypothesis proposes that deeper, downstream areas process global context and then send information back to upstream neurons to resolve ambiguous local signals.
To test these competing ideas, Reynolds and Franken recorded neuronal activity with electrodes across cortical layers while animals viewed a simple stimulus: a square presented on a uniform background. The team identified neurons whose receptive fields covered a small segment of the square’s edge, then measured the timing and strength of border ownership signals across different cortical depths.
The critical result was that the earliest border ownership signals appeared in neurons located in the deep layers of the cortex. “This timing profile supports a major role for feedback pathways in determining border ownership,” says Franken, a physician-scientist supported by a K99 Pathway to Independence Award from the National Institutes of Health. “Feedback connections both arrive at and originate from deep-layer neurons, consistent with our observations.”
The researchers also found that neurons organized in vertical columns across layers tended to share the same border ownership preference. Some cortical columns consistently favored scenes where the object lay on the left side of a border, while other columns preferred the object on the right. This columnar organization indicates that feedback signals may be structured and topographically organized, pointing to a systematic architecture for contextual processing in cortex.
These findings have broader implications for understanding perception and its disorders. “Mapping how ensembles of neurons interact to form internal representations of the external world helps us identify where those processes may go awry,” Franken says. “Disruptions in feedforward-feedback loops could underlie the distorted perceptions, hallucinations, or delusions observed in psychiatric illnesses such as schizophrenia.”
Follow-up experiments planned by the team will probe how the feedback signals that originate in deep layers actually influence border processing and how those influences integrate with feedforward inputs to produce stable percepts.
Funding: This work was supported by the George E. Hewitt Foundation for Medical Research, a NARSAD Young Investigator Grant from the Brain & Behavior Research Foundation, and the National Eye Institute of the National Institutes of Health.
About this visual neuroscience research news
Author: Press Office
Source: Salk Institute
Contact: Press Office – Salk Institute
Image: The image is in the public domain
Original Research: The findings are reported in the journal eLife