Patrick Mineault

A few years ago, with @tyrellturing.bsky.social @patrickmineault.bsky.social and others, we showed one single objective (prediction) was sufficient for explaining specialized pathways of mouse visual cortex: tinyurl.com/2vsjbve7

Super cool to see the same observation now for human visual cortex.

Brain Surfaces of 70 primate species

1
To predict the behaviour of a primate, would you rather base your guess on a closely related species or one with a similar brain shape? We looked at brains & behaviours of 70 species, you’ll be surprised!

🧵Thread on our new preprint with @r3rt0.bsky.social , doi.org/10.1101/2025...

RFI: Neuroscience & the Path to Safe AGI Safe Artificial General Intelligence through Neuroscience

RFI: How can brain science shape frontier-scale AGI? The Amaranth Foundation wants your 2-page ideas on how brain science can guide & govern frontier-scale AGI—incl. short-timeline (2025-2028) scenarios. Send ≤2-page ideas by 29 Aug. Link: blog.amaranth.foundation/p/rfi-neuros...

How can we best use AI in science?

Myself and 9 other research fellows from @imperial-ix.bsky.social use AI methods in domains from plant biology (🌱) to neuroscience (🧠) and particle physics (🎇).

Together we suggest 10 simple rules @plos.org 🧵

doi.org/10.1371/jour...

LLM Daydreaming Proposal & discussion of how default mode networks for LLMs are an example of missing capabilities for search and novelty in contemporary AI systems.

Curious what cognitive scientists think about this daydreaming proposal from Gwern: gwern.net/ai-daydreaming

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...

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...

Who Does What in Deep Learning? Multidimensional Game-Theoretic Attribution of Function of Neural Units Neural networks now generate text, images, and speech with billions of parameters, producing a need to know how each neural unit contributes to these high-dimensional outputs. Existing explainable-AI ...

🚨Preprint Alert: Who Does What in Deep Learning? Multidimensional Contribution of Neural Units using Game Theory: arxiv.org/abs/2506.19732
The result of my MSc thesis is out with @kayson.bsky.social @fatemehhadaeghi.bsky.social @patrickmineault.bsky.social @kordinglab.bsky.social, Claus C. Hilgetag

Thrilled to announce I'll be starting my own neuro-theory lab, as an Assistant Professor at @yaleneuro.bsky.social @wutsaiyale.bsky.social this Fall!

My group will study offline learning in the sleeping brain: how neural activity self-organizes during sleep and the computations it performs. 🧵

Does anybody have a good review or blog post on advances in neurosymbolic and program induction methods over the last 2 years? I'm looking for something technical that has the depth of explanation I would expect from, say, Lil' Log. My usual tools have failed me.

Compete in the Berggruen Prize Essay Competition | $50,000 Prize Submit your essay for a chance to win $50,000 and be published by Berggruen Press. Explore fundamental questions and bridge cultural perspectives. Make your mark on the future of philosophy and cultur...

A $50k essay competition about consciousness... Any takers? Via @erikhoel.blogsky.venki.dev berggruen.org/essay-compet...

This is beautiful. Awe-inspiring science.

Woo! Thanks for subscribing, all!

Very cool! However, the code link is broken.

🪰 A team of Janelia researchers & collaborators have created a new set of tools for studying motor control of wing movements in fruit flies, providing an important #openscience resource for future studies of fly flight & courtship 🧠
hhmi.news/3H3ohn8

Screenshot of Figure 1, which has two panels labeled (a) and (b). The caption states the following. Figure 1: Overview of our study. (a) Experiment 1: We explore whether forward passes show mechanistic signatures of competitor interference, first preferring a salient competing intuitive answer before preferring the correct answer. (b) Experiment 2: We systematically investigate the ability of dynamic measures derived from forward passes to predict indicators of processing load in humans.

Excited to share a new preprint w/ @michael-lepori.bsky.social & Michael Franke!

A dominant approach in AI/cogsci uses *outputs* from AI models (eg logprobs) to predict human behavior.

But how does model *processing* (across layers in a forward pass) relate to human real-time processing? 👇 (1/12)

The catastrophic risks of AI — and a safer path Yoshua Bengio — the world's most-cited computer scientist and a "godfather" of artificial intelligence — is deadly concerned about the current trajectory of the technology. As AI models race toward fu...

When I realized how dangerous the current agency-driven AI trajectory could be for future generations, I knew I had to do all I could to make AI safer. I recently shared this personal experience, and outlined the scientific solution I envision @TEDTalks⤵️
www.ted.com/talks/yoshua...

Applications are now open for a special cohort of the Brains Accelerator targeting ambitious AI research programs, with a special focus on security and governance capabilities!

spec.tech/ai-brains

1/

Titans: Learning to Memorize at Test Time Over more than a decade there has been an extensive research effort on how to effectively utilize recurrent models and attention. While recurrent models aim to compress the data into a fixed-size memo...

arxiv.org/abs/2501.00663

New Paper: Continuous Thought Machines

pub.sakana.ai/ctm/

Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence.

Thread ↓

I'm very pleased to share our latest study:
‘Emergence of Language in the Developing Brain’,
by L Evanson, P Bourdillon et al:
- Paper: ai.meta.com/research/pub...
- Blog: ai.meta.com/blog/meta-fa...
- Thread below 👇

I just tried the Gemini features in Google Sheets. It's pretty good! You can ask it to perform a simple analysis, and it will do it. I asked it to do a logistic regression, and it imported statsmodels, ran it, printed the results and interpreted them.

Excited to dig into this, from @jeanremiking.bsky.social 's group. Compares neural representations of language across age. The stimulus? Le Petit Prince. ai.meta.com/research/pub...

I think we're overdue for a rollup of the best papers of the past year in NeuroAI. Candidates?

Check out our new paper!
Vision models often struggle with learning both transformation-invariant and -equivariant representations at the same time.

@hafezghm.bsky.social shows that self-supervised prediction with proper inductive biases achieves both simultaneously. (1/4)
#MLSky #NeuroAI

Have you tried Il Laboratorio del Gelato in the Lower East Side? Some of the best ice cream I’ve had

Why I'm more worried about AI safety now than 6 months ago Exponentials are all you need

New post: things have been moving very fast in AI. But has safety caught up to capabilities? open.substack.com/pub/naix/p/w...

*Peak Twitter circa 2022*, I agree with you it's been trash for a while

In the peak days of Twitter I could get some NeuroAI news with occasional politics. Now the mix is reversed, it's all politics and occasionally NeuroAI

My issue is that the Following Feed is already so noisy I don't want to make the problem worse by increasing my follows

Any good neuroscience/NeuroAI discord servers I should know about? I feel like the neuroscience town square on bsky is a bit of a ghost town