Summaries like this, in your inbox every morning.
Sign up free →What happened
Scientists at EPFL's NeuroAI Lab developed an AI model (a topographic neural network) that predicts where and how to stimulate higher-level visual regions of the brain to evoke perception of specific objects. Dutch researchers tested the model's predictions on two implanted monkeys, and the results were presented in April at the International Conference on Learning Representations. So far, the team has shown they can shape object perception by biasing how the brain represents a visual stimulus that is already present.
Why it matters
Current visual prosthetics are limited to simple shapes and light flashes because they target lower-level brain regions and face hardware constraints from needing multiple electrodes. Higher-level brain regions could theoretically support perception of complex objects like houses and cars, but researchers did not know exactly where and how to stimulate them—until now. This AI-guided approach could unlock a new generation of prosthetics that restore meaningful visual perception to people whose retina, optical nerve, or both cannot be repaired.
What to watch
The researchers' next step is to create object perception from scratch—stimulating the cortex while no visual stimulus is presented—rather than merely biasing existing visual input. The team also plans to investigate whether this modeling approach works for auditory stimulation in cochlear implants, supported by a grant from the Horton Health Foundation.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
5 minutes a day. The AI essentials.
200+ sources · Email / LINE / Slack
