Summary
A bioRxiv preprint compares a simple model of the primary visual cortex with phosphenes described and drawn by one cortical-implant recipient. The model reproduced a variety of reported shapes and suggests that some electrodes may activate multiple locations.
A bioRxiv preprint posted on October 20, 2026, reports that a simple model of the primary visual cortex (V1) reproduced a range of phosphene shapes described and drawn by one recipient of a cortical implant. The researchers also found that, for some electrodes, the model predicted stimulation at multiple locations; they suggest that activation of nerve-cell axons could be one explanation.
Phosphenes are visual sensations, such as spots or patches of light, produced without light entering the eye. Cortical visual prostheses use electrodes to stimulate the brain’s visual cortex in an effort to evoke such sensations. Reports from people with these implants indicate that stimulation by small depth electrodes does not always produce a small, distinct spot: percepts can include larger or irregular shapes.
How the model accounts for different shapes
The researchers compared the model’s predicted phosphene patterns with drawings and descriptions from a single implant recipient. They report that the model captured the variety of percepts this person described, rather than predicting only small, dot-like spots.
One possible implication is that an electrode’s effect may not map neatly to a single point in the perceived visual field. For a subset of electrodes, the model suggested activation at multiple locations. The authors propose that stimulation of axons—the extensions that carry signals from nerve cells—could contribute to this pattern. This is a proposed explanation, rather than a demonstrated account of how those percepts arose.
The comparison offers a way to connect electrode stimulation with the shapes a person reports seeing. That matters for interpreting results from cortical prostheses: the percept produced by stimulation may be more varied than a simple dot, and a model may help represent that variation.
The work is a model-based comparison with reports from one implant recipient, not a population-level assessment. The finding therefore concerns how well this simple V1 model captured this individual’s reported percepts.