Lightnews — Scholar-powered news

Reposted by The vOICe vision BCI 🧠🇪🇺

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 11h

To ChatGPT: How does LLM image description work? How does that complement low-level (pixel-level) scene description for the blind through visual-to-auditory sensory substitution? chatgpt.com/share/68e689... #blindness #blind #a11y #AI #LLM scene description

ChatGPT - LLM image description process

Shared via ChatGPT

chatgpt.com

1 2 1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 8h

(YouTube) The science of imagination: Away from the here and now (Part 1) www.youtube.com/watch?v=aUdc... via @champalimaudr.bsky.social

7. The Science of Imagination: Away from the Here and Now (Part 1)

YouTube video by Champalimaud Foundation

www.youtube.com

Reposted by The vOICe vision BCI 🧠🇪🇺

Ella Striem-Amit @striemamit.bsky.social · 10h

(please repost) If you're looking for a #neuroscience PhD program - and interested in human brain plasticity and reorganization (neuroimaging in people born with blindness, deafness or without hands), my lab is accepting students this cycle. Email me!

Home Page - Interdisciplinary Program in Neuroscience

The Georgetown Difference We are invested in providing a transformative experience through holistic training, accessible resources, and personalized career strategies to help you reach your aspiration...

neuroscience.georgetown.edu

12 12

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 11h

Thank you! I was just prodding ChatGPT on related concepts in providing complementary pixel-level (low-level) and semantic (high-level) vision to the totally blind bsky.app/profile/seei...

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 11h

Grok's thoughts on this at x.com/grok/status/...

Grok on X: "@seeingwithsound LLM image description uses vision encoders (like ViT) to turn pixels into embeddings, bridged to language models for semantic captions. It complements pixel-level sensory substitution (e.g., The vOICe) by adding high-level context—like "a dog under a table"—to raw soundscapes" / X

@seeingwithsound LLM image description uses vision encoders (like ViT) to turn pixels into embeddings, bridged to language models for semantic captions. It complements pixel-level sensory substitution (e.g., The vOICe) by adding high-level context—like "a dog under a table"—to raw soundscapes

x.com

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 11h

To ChatGPT: How does LLM image description work? How does that complement low-level (pixel-level) scene description for the blind through visual-to-auditory sensory substitution? chatgpt.com/share/68e689... #blindness #blind #a11y #AI #LLM scene description

ChatGPT - LLM image description process

Shared via ChatGPT

chatgpt.com

1 2 1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 11h

I love that analogy! Do you perhaps have (or know of) a paper about it?

1

Reposted by The vOICe vision BCI 🧠🇪🇺

andrew haun @andrewhaun.bsky.social · Aug 30

These LLM AIs are all suffering from a sort of synthetic Anton's syndrome - they have absolutely no capacity for spatial experience (or simulation of it) and yet are unaware of it - they have all the "high level" stuff you get from seeing, so why not behave as though it's *all* really there?

Tomer Ullman @tomerullman.bsky.social · Aug 28

oh, nice write up in The Register about Illusion Illusions:

www.theregister.com/2025/08/19/v...

original paper here:

arxiv.org/abs/2412.18613

2 2 8

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 15h

"For individuals with profound visual impairments, the ability to perceive and recognize objects such as faces, tools, or scenes would open the possibility of richer and possibly more useful visual perception." #blindness

2

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 15h

More information in the thread bsky.app/profile/hann...

Hannes Mehrer @hannesmehrer.bsky.social · 1d

🧠 New preprint: we show that model-guided microstimulation can steer monkey visual behavior.

Paper: arxiv.org/abs/2510.03684

🧵

1 1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 15h

Model-guided microstimulation steers primate visual behavior arxiv.org/abs/2510.03684 by @hannesmehrer.bsky.social @mschrimpf.bsky.social et al.; in humans compare higher-level visual cortex stimulation with verbal (object-level semantic, auditory) AI scene description? #BCI #NeuroTech

1 1 2

Reposted by The vOICe vision BCI 🧠🇪🇺

Kaiser Lab @dkaiserlab.bsky.social · 18h

From line drawings to scene perception — our new review argues for moving beyond experimenter-driven manipulations toward participant-driven approaches to reveal what’s in our internal models of the visual world. 👁️✍️🛋
royalsocietypublishing.org/doi/10.1098/...

Characterizing internal models of the visual environment | Proceedings of the Royal Society B: Biological Sciences

Despite the complexity of real-world environments, natural vision is seamlessly efficient. To explain this efficiency, researchers often use predictive processing frameworks, in which perceptual effic...

royalsocietypublishing.org

1 6 30

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 19h

"The observed transition from hub-dominated to distributed architecture challenges traditional views of fixed sensory processing hierarchies while supporting emerging theories of dynamic cortical reorganization."

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 20h

Network topological reorganization mechanisms of primary visual cortex under multimodal stimulation (in mice) www.frontiersin.org/journals/neu... #crossmodal #integration #plasticity #V1 #neuroscience

Frontiers | Network topological reorganization mechanisms of primary visual cortex under multimodal stimulation

IntroductionThe functional connectivity topology of the primary visual cortex (V1) shapes sensory processing and cross-modal integration, yet how different s...

www.frontiersin.org

1 2 4

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

Bloomberg: Engineered protein that responds to light could help restore sight www.bloomberg.com/news/newslet... Archived at archive.ph/46Olz #optogenetics

Engineered Protein That Responds to Light Could Help Restore Sight

Hi, it’s Ike in Boston. I’m spending most of my time these days thinking about brain implants. More on that below, but first ...

www.bloomberg.com

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

"The lack of transfer of learning from video game training to object recognition suggests that gamification-based rehabilitation for sight recovery technologies may have limited utility and may be most effective when targeted on learning specific visual tasks."

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

Perceptual learning of prosthetic vision using video game training jov.arvojournals.org/article.aspx... "A key limitation shared by both electronic and optogenetic sight recovery technologies is that they cause simultaneous rather than complementary firing within on- and off-center cells"

Perceptual learning of prosthetic vision using video game training | JOV | ARVO Journals

jov.arvojournals.org

1

Reposted by The vOICe vision BCI 🧠🇪🇺

Katrin Franke @katrinfranke.bsky.social · 1d

Our review is out 🎉

We discuss how neural computations in the retina shape behavior, continuing the top-down tradition of Lettvin et al.'s classic bug-detector neurons

Check it out 👀 arevie.ws/46HMEjd @annualreviews.bsky.social w/ coauthors @serenariccitelli.bsky.social @annaintegrated.bsky.social

5 18

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

PDF at pmc.ncbi.nlm.nih.gov/articles/PMC...

pmc.ncbi.nlm.nih.gov

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

"None of the 10 patients who had recent follow-up appointments were currently using the device", "Discontinuation reasons included performance issues, cognitive load, and physical discomfort, among others."

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

Long-term outcomes and patient experiences with the Argus II retinal prosthesis system journals.sagepub.com/doi/10.1177/... "Argus II provided potential benefits but also presented significant challenges related to device performance, cognitive demands, and long-term support"; retinal implant

Long-Term Outcomes and Patient Experiences With the Argus II Retinal Prosthesis System - Jonathan D. Shirian, Alex Yuan, Aleksandra Rachitskaya, 2025

Purpose: To evaluate the long-term outcomes and experiences of patients with the Argus II retinal prosthesis in individuals with advanced-stage retinitis pigmen...

journals.sagepub.com

1 1 1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 1d

Raster scanning can improve task performance in simulated prosthetic vision ieeexplore.ieee.org/document/111... "Our findings underscore the importance of sequentially presenting visual information"; #BCI #NeuroTech #blindness

A) Simultaneous stimulation. B) Dynamic stimulation. C) A demonstration of presentation methods at different refresh rates, without persistence. D) We used MRI scans from real subjects combined with plausible implants planning to derive a more realistic phosphene map. E) Tasks we used to evaluate simulated prosthetic vision.

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 2d

Thanks! I'm subscribed to WIRED since this year, and will keep an eye on NPR.

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 2d

Thank you for your take on this. Of course I read BBC and The Guardian too, but it would still be good to have a trusted insider news agency in the US if possible under the current US administration, especially since I'm based in Europe. Occasional failures forgiven.

1

The vOICe vision BCI 🧠🇪🇺 @seeingwithsound.bsky.social · 2d

How is NYTimes doing these days? I subscribed during Trump I, but do not use it much nor track its current trustworthiness/independence.

1