Elliot Smith
@neurosmith.bsky.social
470 followers 640 following 110 posts
Human neuronal computations during cognition, seizures, brain stimulation, etc… @ Utah. Www.neurosmiths.org
Posts Media Videos Starter Packs
Reposted by Elliot Smith
markkho.bsky.social
I'm recruiting grad students!! 🎓

The CoDec Lab @ NYU (codec-lab.github.io) is looking for PhD students (Fall 2026) interested in computational approaches to social cognition & problem solving 🧠

Applications through Psych (tinyurl.com/nyucp) are due Dec 1. Reach out with Qs & please repost! 🙏
codec lab
codec-lab.github.io
Reposted by Elliot Smith
kevinjkircher.com
Sometimes I think about how from 1935-1975ish, Bell Labs produced an insane amount of revolutionary science and technology, including 11 Nobel Prizes, the transistor, UNIX, C, the laser, the solar cell, information theory, etc. The secret? Provide scientists with ample, steady, no-strings funding.
sites.stat.columbia.edu
neurosmith.bsky.social
Awesome! Congrats, Marcin! 👏👏🙌
Reposted by Elliot Smith
marcin-leszcz.bsky.social
1/3: What happens in the #brain when moving your eyes? Using closed-loop #EyeTracking with simultaneous intracranial #EEG in humans, we found that: *Saccades boost visual responses but reduce network connectivity. *Excitability changes and visual features are coded along the saccade–fixation cycle.
Reposted by Elliot Smith
mbeyeler.bsky.social
👁️🧠 New preprint: We demonstrate the first data-driven neural control framework for a visual cortical implant in a blind human!

TL;DR Deep learning lets us synthesize efficient stimulation patterns that reliably evoke percepts, outperforming conventional calibration.

www.biorxiv.org/content/10.1...
Diagram showing three ways to control brain activity with a visual prosthesis. The goal is to match a desired pattern of brain responses. One method uses a simple one-to-one mapping, another uses an inverse neural network, and a third uses gradient optimization. Each method produces a stimulation pattern, which is tested in both computer simulations and in the brain of a blind participant with an implant. The figure shows that the neural network and gradient methods reproduce the target brain activity more accurately than the simple mapping.
neurosmith.bsky.social
That’s beautiful 🥹 congrats!
neurosmith.bsky.social
This sounds super relevant. Which conference is that? Couldn’t find it with a cursory search
neurosmith.bsky.social
Looks awesome, congrats!
Reposted by Elliot Smith
zoechristensonwick.bsky.social
🚨New preprint alert🚨
We used closed-loop optogenetics to causally test the importance of inhibitory spike timing in network function and found that manipulating PV+ cell theta phase locking in the dentate gyrus can shift seizure susceptibility (both ways!)
Reposted by Elliot Smith
actlab.bsky.social
New preprint from the lab! 🧠
Led by Juliana Trach, w/ Sophia Ou

Using fMRI, we discovered evidence for time-sensitive reward prediction errors (RPEs) in the human cerebellum.

Builds on, and extends, recent work in both rodents and NHPs
figure showing cerebellar RPE responses
neurosmith.bsky.social
The whole human brain def doesn’t encode movement though, 💯
neurosmith.bsky.social
Different evolutionary and developmental factors. I’m a proponent for distributed reps and coding, but have def seen human neuronal reps specific for particular brain regions too.
neurosmith.bsky.social
Totally. I don’t mean to discount this obviously great work. I just mean that overtraining likely strengthens and broadens reps, but I don’t know and not sure one can really generalize. Also mouse and human brains are likely pretty different with different micro to macro anatomy and …
neurosmith.bsky.social
Way less variation across brain areas than in primates, no?
neurosmith.bsky.social
Agreed, but until we have widespread neuropixels in the human brain we have to rely on the innumerable evidence for functional specialization, ranging from cytoarchitecture (not relevant for mice) to lesion studies.
neurosmith.bsky.social
Here’s some high desert happiness for your TL:
A scorpion under UV light A slot canyon with a black dog A dog on the edge of a mesa The edge of an eroding mesa against a forest backdrop
Reposted by Elliot Smith
Check out @tifenpan.bsky.social 's just published paper! we demonstrate how to use RNNs to infer latent variables from cognitive models, even when standard methods don't work easily.
Reposted by Elliot Smith
marcelomattar.bsky.social
Thrilled to see our TinyRNN paper in @nature! We show how tiny RNNs predict choices of individual subjects accurately while staying fully interpretable. This approach can transform how we model cognitive processes in both healthy and disordered decisions. doi.org/10.1038/s415...
Discovering cognitive strategies with tiny recurrent neural networks - Nature
Modelling biological decision-making with tiny recurrent neural networks enables more accurate predictions of animal choices than classical cognitive models and offers insights into the underlying cog...
doi.org