Happy and proud to see @rjantonello.bsky.social’s work awarded by SNL!

Our latest paper outlining our ecosystem of tools for mining the neuroimaging literature, is finally officially published in eLife! doi.org/10.7554/eLif...

New Open dataset alert: 🧠 Introducing "Spacetop" – a massive multimodal fMRI dataset that bridges naturalistic and experimental neuroscience! N = 101 x 6 hours each = 606 functional iso-hours combining movies, pain, faces, theory-of-mind and other cognitive tasks! 🧵below

New paper with @rjantonello.bsky.social @csinva.bsky.social, Suna Guo, Gavin Mischler, Jianfeng Gao, & Nima Mesgarani: We use LLMs to generate VERY interpretable embeddings where each dimension corresponds to a scientific theory, & then use these embeddings to predict fMRI and ECoG. It WORKS!

In our new paper, we explore how we can build encoding models that are both powerful and understandable. Our model uses an LLM to answer 35 questions about a sentence's content. The answers linearly contribute to our prediction of how the brain will respond to that sentence. 1/6

The preprint of my 1st project in grad school is up 🙌 We propose a simple, information-theoretic model of how humans remember narratives. We tested it with the help of open-source LLMs. Plz check out this thread for details ➡️ Many thanks to my wonderful advisors! It's been a fun adventure!!

New paper with @mujianing.bsky.social & @prestonlab.bsky.social! We propose a simple model for human memory of narratives: we uniformly sample incoming information at a constant rate. This explains behavioral data much better than variable-rate sampling triggered by event segmentation or surprisal.

Hey, if you need a breather from the news, we have pretty brain pics to distract you! We made an interactive viewer showing images optimized to elicit responses from different places in the brain. Link to viewer in 🧵

Our latest brain-to-image decoding model is now available on HuggingFace: "Dynadiff: Single-stage Decoding of Images from Continuously Evolving fMRI", led by Marlène Careil and Yohann Benchetrit: - Paper: arxiv.org/pdf/2505.14556 - Github: github.com/facebookrese... - Thread: 👇

people need to internalize, very quickly, that federal research grants are a hypercompetitive contracting process not charity, and that what Uncle Sam gets in return for that money is American dominance in the future

Tell Congress to protect crucial medical research funding #saveNIHresearch ucal.us/saveNIHresearch @uofcalifornia.bsky.social @ucsantabarbara.bsky.social

Proud to be a signatory of this statement from 1900 members of NASEM: We call on the administration to cease its wholesale assault on U.S. science, and we urge the public to join this call. docs.google.com/document/d/1...

Excited to share on my first post on Bluesky our new paper in NHB examine the innate and developing aspects of the wiring of the visual system. Congratulations to @emilykubota.bsky.social and the baby MRI team on this important work

This was an interesting commentary to write on work by Binder et al. regarding impaired acoustic phonetic perception after unilateral left hemisphere stroke. academic.oup.com/brain/advanc...

I am incredibly proud to share my first, first-author paper as a postdoc with @benhayden.bsky.social . How does the human hippocampus, known for encoding concepts, represent the meanings of words while listening to narrative speech? www.biorxiv.org/content/10.1...

Trends in Cognitive Sciences Ignoring the cerebellum is hindering progress in neuroscience www.cell.com/trends/cogni...

Our paper on how neural codes track prior events in a narrative and predict subsequent memory for details, led by @collinsilvy.bsky.social, is now out in Communications Psychology! rdcu.be/d93Vc #neuroskyence #psychscisky

Many cognitive neuroscientists ignore the cerebellum but it's long past time to pay more attention to it (and fund more cerebellum grants). Not just for basic science: the cerebellum is commonly affected by pediatric brain tumors, leading to lifelong motor AND cognitive deficits. Must-read paper 👇🏼

Very pleased to share our recent work, in which we use LLMs for automated discovery of interpretable models of animal behavior 🪰🐀🕵️‍♀️ that take the form of Python programs 🐍 See below for a summary of key results by @pcastr.bsky.social!

Two new studies from our team we're particularly happy about: Study 1: ai.meta.com/research/pub... Study 2: ai.meta.com/research/pub... Blog 3: ai.meta.com/blog/brain-a...

I'm excited to share our new paper (with @alexanderhuth.bsky.social) on transferring language decoders across participants and modalities! authors.elsevier.com/a/1kZRD3QW8S... 1/5

looking forward to discussing the relationship between human language processing and LLMs at APS with @alexanderhuth.bsky.social and @neuranna.bsky.social ! 🧠🤖

Just so everyone is clear, the Governor of Texas announces he ordered the protestors at the public university arrested because he is discriminating against their viewpoint.

New paper out in Nature Neuro, wherein we ask: How do perceptual and mnemonic brain networks share information? www.nature.com/articles/s41... Massive work with Ed Silson, Brenda Garcia, and @carolinerobertson.bsky.social

We so do! I wrote that book a few years ago now, and yet that imagined future just keeps getting closer and closer in the real world :(

Love this piece about our brain decoding work (and Rafa Yuste's tireless advocacy for neurorights) that Fletcher Reveley wrote for Undark: undark.org/2024/01/03/b...

Devastatingly sad news. My good friend, colleague, and a treasured scientist at the NIH for over 20 years, Dr. Sean Marrett, has passed away today. I wrote a short obituary for him. www.thebrainblog.org/2023/12/13/a...

At NeurIPS this week, @rjantonello.bsky.social is presenting his work on scaling fMRI language encoding models. We're also sharing code, extracted features, and estimated model weights for some of our best models: github.com/HuthLab/enco... Paper: arxiv.org/abs/2305.11863

I got the exciting chance to talk with @fmri-today.bsky.social on his Neurosalience podcast earlier this year, and the episode is just out today! It was fun, and I hope informative too :) podcasters.spotify.com/pod/show/ohb... www.youtube.com/watch?v=KexB... podcasts.apple.com/us/podcast/s...

Aditya is showing this work at CoNLL today, poster session 1 @ 1:45pm. Check it out if you're there!

TURN!!! IT!!! OFF!!!

Words can't describe the energy in the room at the Ann S. Bowers Women's Brain Health Initiative launch, so I'll leave the heavy lifting to this short film. Humor can subvert, enrage, empower & educate. So take a look and then let's get to work. m.youtube.com/watch?v=tX5u...

Very cool new paper from Kamitani group on decoding visual illusions www.science.org/doi/full/10....

Decoding reach direction in early "visual" cortex of congenitally blind individuals www.jneurosci.org/content/43/4... "reaching in different directions can be reliably decoded from fMRI activity of EVC in congenitally blind"; foveal hi-res & dorsal roles; visual imagery not needed; #neuroscience

At the final SfN poster session Aditya Vaidya will present some new work he’s doing w/ me & @libertysays.bsky.social. It sounds a little nuts, but we’re using in silico experiments on fMRI models to replicate effects that really should only work in ECoG. Poster DD15 Wed PM.

If you're at SfN and you're interested in fMRI and LLMs, go check out @rjantonello.bsky.social's poster "Scaling laws for language encoding models in fMRI" at WW27 Tuesday afternoon! Paper: arxiv.org/abs/2305.11863

Neuroscience and AI interpretability are already diverging in their jargon: "decoder" in neuroscience = "probe" in AII "mixed selectivity" in neuroscience = "superposition" in AII Soon speciation will occur and we will no longer be able to produce productive scientific offspring together. #neuroai

Thrilled to share new work w @taylorwwebb.bsky.social Shanka Mondal: A Prefrontal Cortex-inspired Architecture for Planning in Large Language Models (arxiv.org) LLMs struggle w multi-step planning. We propose a solution inspired by brains: planning via recurrent interactions of PFC subregions. 🧵1/n

For years we have tended to discount activation signals in the white matter as artifact. It's becoming clear now that this was a mistake - white matter shows stimulus-driven activation similar (though much smaller) to gray matter. pubmed.ncbi.nlm.nih.gov/37824529/

Check out this new paper! Led by @mehdiazabou.bsky.social and @evadyer.bsky.social, we show that it is possible to get SOTA brain decoding with transfer across individuals and tasks! The key is a clever way to tokenize spiking data for transformers. #neuroskyence #brain #neurotech

wow

This is wildly impressive! Really awesome that this works with MEG recordings.

can anyone tell me the current state of the "cognitive linguistics" research program? are there big things happening in that world? are there any groups/individuals who are particularly active?

As my first post, delighted to share our #neurips2023 paper: Evaluating Cognitive Maps & Planning in LLMs with CogEval We test cognitive maps & planning in 8 LLMs. Failures like hallucinating invalid paths & falling in loops suggest no emergent zero-shot planning. 1/n 🧵 arxiv.org/abs/2309.15129

What makes an image memorable? Images that are highly arousing and moderately negative are more memorable. But the arousal/valence of an image explains surprisingly little variance in memorability! So proud of the amazing @hartwakeland.bsky.social for leading this work! #PsychSciSky

Open PhD position with Christophe Pallier & myself, to work on the brain and computational bases of language using neuroimaging and deep learning: - application: forms.gle/equJz92rGtiL... - requirements: python, ML, and ideally, neuroimaging experience. Please RT🙏

Our new paper is out now in Nature Communications, highlighting what happens in the brain during a semantic context task when you disconnect the anterior temporal lobe: www.nature.com/articles/s41... 🧠📈

New from Aditya Vaidya arxiv.org/abs/2310.06408 Many have shown that LM and human predictions are very correlated, but we found that's not true when the text repeats! This points to divergence btwn memory mechanisms, which we localize in the model and then modify to get more human-like behavior.

Our MindEye fMRI-to-Image paper got accepted as spotlight for #NeurIPS2023! See you in New Orleans, will be my first time attending NeurIPS :) Updated camera ready paper is also now live on arxiv: arxiv.org/abs/2305.182... Includes new expts, appendix figures, more references to other work 🧠📈