Lightnews — Scholar-powered news

Eghbal Hosseini

@eghbal-hosseini.bsky.social

200 followers 320 following 17 posts

PhD in computational neuroscience; Postdoc at MIT; working with @evfedorenko.bsky.social eghbalhosseini.github.io

Posts Media Videos Starter Packs

Pinned

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

Why do diverse ANNs resemble brain representations? Check out our new paper with Colton Casto, @nogazs.bsky.social , Colin Conwell, Mark Richardson, & @evfedorenko.bsky.social on “Universality of representation in biological and artificial neural networks.” 🧠🤖
tinyurl.com/yckndmjt

Universality of representation in biological and artificial neural networks

Many artificial neural networks (ANNs) trained with ecologically plausible objectives on naturalistic data align with behavior and neural representations in biological systems. Here, we show that this...

tinyurl.com

2 28 110

Reposted by Eghbal Hosseini

Reza Shadmehr @rezashadmehr.bsky.social · May 22

Neuronal computation in the cerebellum via a vector calculus.

Work of Mohammad Amin Fakharian, Alden Shoup, Paul Hage, and Hisham Elseweifi

www.science.org/doi/10.1126/...

A vector calculus for neural computation in the cerebellum

Null space theory predicts that neurons generate spikes not only to produce behavior but also to prevent the undesirable effect of other neurons on behavior. In this work, we show that this competitiv...

www.science.org

5 35 110

Reposted by Eghbal Hosseini

CogCompNeuro @cogcompneuro.bsky.social · May 19

A detailed schedule for CCN2025 is available on our website now, including the specific keynotes / GACs / community events taking place:
2025.ccneuro.org/schedule-of-...

The early bird registration deadline is coming up this Friday (23rd May)!

8 13

Reposted by Eghbal Hosseini

Colton Casto @coltoncasto.bsky.social · Apr 21

New paper! 🧠 **The cerebellar components of the human language network**

with: @hsmall.bsky.social @moshepoliak.bsky.social @gretatuckute.bsky.social @benlipkin.bsky.social @awolna.bsky.social @aniladmello.bsky.social and @evfedorenko.bsky.social

www.biorxiv.org/content/10.1...

1/n 🧵

The cerebellar components of the human language network

The cerebellum's capacity for neural computation is arguably unmatched. Yet despite evidence of cerebellar contributions to cognition, including language, its precise role remains debated. Here, we sy...

www.biorxiv.org

2 20 49

Reposted by Eghbal Hosseini

David Lipshutz @lipshutz.bsky.social · Apr 9

📣 Grad students and postdocs in computational and theoretical neuroscience: please consider applying for the 2025 Flatiron Institute Junior Theoretical Neuroscience Workshop! All expenses are covered. Apply by April 14. jtnworkshop2025.flatironinstitute.org

16 21

Eghbal Hosseini @eghbal-hosseini.bsky.social · Mar 26

I couldn’t recommend a better place to learn and work if you’re considering neuroscience grad school and want to interact with phenomenal scientists!

Nancy Kanwisher @nancykanwisher.bsky.social · Mar 26

I’m hiring a full-time lab tech for two years starting May/June. Strong coding skills required, ML a plus. Our research on the human brain uses fMRI, ANNs, intracranial recording, and behavior. A great stepping stone to grad school. Apply here:
careers.peopleclick.com/careerscp/cl...
......

Technical Associate I, Kanwisher Lab

MIT - Technical Associate I, Kanwisher Lab - Cambridge MA 02139

careers.peopleclick.com

Reposted by Eghbal Hosseini

Ev Fedorenko @evfedorenko.bsky.social · Mar 18

So excited to have our work on conlangs out in PNAS: www.pnas.org/doi/10.1073/... Congrats, Saima, Maya, and the rest of the crew -- well done!
Here is the MIT news story:
news.mit.edu/2025/esperan...

To the brain, Esperanto and Klingon appear the same as English or Mandarin

MIT research finds the brain’s language-processing network also responds to artificial languages such as Esperanto and languages made for TV, such as Klingon on “Star Trek” and High Valyrian and Dothr...

news.mit.edu

18 56

Eghbal Hosseini @eghbal-hosseini.bsky.social · Mar 15

Very important work by @rarefin.bsky.social, Oscar Skean, @yann-lecun.bsky.social and @shwartzzzivravid.bsky.social on richness of information content in intermediate layers in LLMs. highly recommended

NYU Center for Data Science @nyudatascience.bsky.social · Mar 13

Recent work using info theory & geometric analysis by @rarefin.bsky.social, Oscar Skean, & CDS' @shwartzzzivravid.bsky.social and @yann-lecun.bsky.social shows intermediate layers in LLMs often outperform the final layer for downstream tasks.

nyudatascience.medium.com/middle-layer...

Middle Layers Excel: New Research Challenges Final-Layer Focus in Language Models

New research shows that intermediate layers in LLMs often capture better representations than the final layer.

nyudatascience.medium.com

Reposted by Eghbal Hosseini

Cory Shain @coryshain.bsky.social · Mar 11

🚨 First preprint from the lab! 🚨 Josh Rozner (w/@weissweiler.bsky.social and @kmahowald.bsky.social) uses counterfactual experiments on LMs to show that word distributions can provide a learning signal for diverse syntactic constructions, including some hard cases.

Constructions are Revealed in Word Distributions

Construction grammar posits that constructions (form-meaning pairings) are acquired through experience with language (the distributional learning hypothesis). But how much information about…

arxiv.org

2 10 36

Reposted by Eghbal Hosseini

Aran Nayebi @anayebi.bsky.social · Feb 25

What should count as a good model of intelligence?

AI is advancing rapidly, but how do we know if it captures intelligence in a scientifically meaningful way?

We propose the *NeuroAI Turing Test*—a benchmark that evaluates models based on both behavior and internal representations.

3 13 39

Reposted by Eghbal Hosseini

The Transmitter @thetransmitter.bsky.social · Feb 24

Many neuroscience students are steeped in an experiment-first style of thinking that leads to “random walk science.” Let’s not forget how theory can guide experiments towards deeper insights, writes @gershbrain.bsky.social.

#neuroskyence

www.thetransmitter.org/theoretical-...

Breaking the barrier between theorists and experimentalists

Many neuroscience students are steeped in an experiment-first style of thinking. Let’s not forget how theory can guide experiments.

www.thetransmitter.org

4 55 140

Reposted by Eghbal Hosseini

Nature @nature.com · Feb 6

Nature research paper: Children’s arithmetic skills do not transfer between applied and academic mathematics

https://go.nature.com/40PT3pT

Children’s arithmetic skills do not transfer between applied and academic mathematics - Nature

Children who learn maths working in markets and children who learn maths only from school were both unable to transfer their skills to new contexts, highlighting a need to reconsider how maths is taught in school.

go.nature.com

3 26 81

Reposted by Eghbal Hosseini

Gatsby Computational Neuroscience Unit @gatsbyucl.bsky.social · Feb 5

📢 Applications are open for 2025 Analytical Connectionism!

Topical focus: bias in learning
Learn key connectionist models & analytical tools for neural-network analysis, and work on novel projects with faculty & postdoc mentors

⏰ Apply by 18 April

ℹ️ www.analytical-connectionism.net/school/2025

2025 School on Analytical Connectionism (25 Aug - 5 Sep, London).
Apply by April 18, 2025.

Learn about techniques to analyse neural networks and to develop connectionist theories of higher-level cognition; interact with leading researchers in neuroscience, psychology and machine learning.

Supported by the Gatsby Unit and SWC.

5 13

Reposted by Eghbal Hosseini

Erin Grant @eringrant.me · Jan 16

Our representational alignment workshop returns to #ICLR2025! Submit your work on how ML/cogsci/neuro systems represent the world & what shapes these representations 💭🧠🤖

w/ @thisismyhat.bsky.social @dotadotadota.bsky.social, @sucholutsky.bsky.social @lukasmut.bsky.social @siddsuresh97.bsky.social

Dota Tianai Dong @dotadotadota.bsky.social · Jan 16

🚨Call for Papers🚨
The Re-Align Workshop is coming back to #ICLR2025

Our CfP is up! Come share your representational alignment work at our interdisciplinary workshop at
@iclr-conf.bsky.social

Deadline is 11:59 pm AOE on Feb 3rd

representational-alignment.github.io

9 20

Reposted by Eghbal Hosseini

Joao Barbosa @jbarbosa.org · Jan 12

Check our latest in which we leverage shape metrics to compare neural geometry across regions, sessions or subjects and how their differences predict behavior.

w/ Nejatbakhsh, Duong, @sarah-harvey.bsky.social, Brincat, @siegellab.bsky.social, @earlkmiller.bsky.social & @itsneuronal.bsky.social

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Jan 11

Quantifying Differences in Neural Population Activity With Shape Metrics https://www.biorxiv.org/content/10.1101/2025.01.10.632411v1

3 37 100

Eghbal Hosseini @eghbal-hosseini.bsky.social · Jan 6

Thanks for the note, reported!

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

Thank you, @lampinen.bsky.social! I completely agree—understanding the divergences can guide us toward better alignment. Great point about the context effect; additional context could serve as a scaffold for aligning representations across models.

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

Universal representations in biological and artificial networks hold great potential for computational neuroscience and AI interpretability.

We’d love to hear your thoughts! 💬
(13/13)

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

Bottom line: high-performing ANNs converge on similar, brain-like representations, likely due to shared input statistics with humans.

Our findings highlight a promising path for NeuroAI: studying convergences across ANNs to uncover more brain-like representations. (12/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

4️⃣
𝗔. Language: perceived frequency and meaning generality decrease from high-agreement to low-agreement stimuli.

𝗕. Vision: valence and clutter increase, while aesthetics decreases. (11/n)

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

What features distinguish universal from individualized representations?

We asked human participants to rate stimuli on various features in both language and vision. (10/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

3️⃣
We also tested universality from brains to ANNs in the visual domain:
𝗔. High-agreement stimuli across brains also showed higher ANN agreement and alignment.

𝗕. Low-agreement stimuli led to substantially worse ANN-to-brain alignment. (9/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

2️⃣
We tested this in language (𝗔) and vision (𝗕) with 3 stimulus sets: high-agreement, random, and low-agreement.

Higher agreement led to stronger ANN-to-brain alignment, thus modulating agreement directly impacts brain alignment, supporting representation universality. (8/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

Next, we asked: does the degree of ANN agreement predict brain alignment?

If representation agreement across ANN models captures the universal part of representations (shared between 🤖 and the 🧠), modulating representation agreement should also modulate brain alignment. (7/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

1️⃣
𝗔. Low-agreement stimuli robustly reduced ANN agreement compared to previous stimuli.

𝗕. When ANNs disagree, their representations are less predictive of fMRI responses, supporting representation universality and the link between model agreement and brain alignment. (6/n)

1 1

Eghbal Hosseini @eghbal-hosseini.bsky.social · Dec 27

We developed a method to test ANNs as competing hypotheses by finding a set of natural stimuli that reduces agreement across ANN representations.

Our question: how does reducing agreement across ANNs affect their alignment with the brain? We tested this in language. (5/n)

1 1