Vase with Irises - 1890 https://botfrens.com/collections/46/contents/15543

LLMs are trained to compress data by mapping sequences to high-dim representations! How does the complexity of this mapping change across LLM training? How does it relate to the model’s capabilities? 🤔 Announcing our #NeurIPS2025 📄 that dives into this. 🧵below #AIResearch #MachineLearning #LLM

Excited to share a new preprint w/ @annaschapiro.bsky.social! Why are there gradients of plasticity and sparsity along the neocortex–hippocampus hierarchy? We show that brain-like organization of these properties emerges in ANNs that meta-learn layer-wise plasticity and sparsity. bit.ly/4kB1yg5

Connectome suggests brain’s synaptic weights follow heavy-tailed distributions, yet most analyses of RNNs assume Gaussian connectivity.  🧵⬇️ Our @alleninstitute.org #NeurIPS2025 paper shows heavy-tailed weights can strongly affect dynamics, trade off robustness + attractor dimension.

**Discovering network dynamics** One more on estimating dynamics of complex systems, this time with symbolic regression doi.org/10.1038/s435...

Total Lunar Eclipse - 26 May 2021 - From Ángel López-Sánchez - https://flic.kr/p/2m1PLyZ

Jorge Luis Borges

Happy birthday Kay Sage. The artist & poet, famous for her surrealist paintings, often referencing architecture & the built environment, was born today in 1898. #surrealism

The Eye of Silence https://www.wikiart.org/en/max-ernst/the-eye-of-silence-1943

What do we talk about when we talk about "readout"? I argued that our overly specialized, modular approach to studying the brain has given us a simplistic view of readout. 🧠📈

Prominence 10-26-10 - From Jason Major (jpmajor.bsky.social) - https://flic.kr/p/8NkYgk

ESA ROSETTA 14 July 2015 - From 2di7 & titanio44 - https://flic.kr/p/vpqn2z

Mimas, Epimetheus and Rings - From Gordan Ugarković (ugordan.bsky.social) - https://flic.kr/p/5oPunj

🧠🚨 How does the hippocampus transform the visual similarity space to resolve memory interference? In this new preprint, we found that the hippocampus sequentially inverts the behaviorally relevant dimensions of similarity 🧵 www.biorxiv.org/content/10.1...

Replace Variational Autoencoder (VAE) with pretrained representation encoders (e.g., DINO, SigLIP, MAE) paired with trained decoders, which they terms as Representation Autoencoders (RAE).

Preprint Alert 🚀 Can we simultaneously learn transformation-invariant and transformation-equivariant representations with self-supervised learning? TL;DR Yes! This is possible via simple predictive learning & architectural inductive biases – without extra loss terms and predictors! 🧵 (1/10)

So excited to see this preprint released from the lab into the wild. Charlotte has developed a theory for how learning curriculum influences learning generalization. Our theory makes straightforward neural predictions that can be tested in future experiments. (1/4) 🧠🤖 🧠📈 #MLSky

Tomorrow the next meeting of MIT #Consciousness Club. This is a Zoom link 🔗⬇️

Max Ernst Showing a Young Girl the Head of his Father https://www.wikiart.org/en/max-ernst/max-ernst-showing-a-young-girl-the-head-of-his-father

Cielo di piombo, ispettore Callaghan...

Jour...sur la grève

The domain of Arnheim (1962) by René Magritte

Kay Sage

Misha Kovalov

Interesting paper! www.nature.com/articles/s41...

Super fun New blog post about @kristorpjensen.bsky.social awesome preprint.

If you know ground truth, you can evaluate LLM annotation by measuring accuracy. But what if the annotation task is subjective and you have many judgments by different observers? This paper offers a method for assessing whether divergence of a single LLM is outside expected human range. +

University of Vienna Ceiling Paintings (Medicine), detail showing Hygieia https://botfrens.com/collections/109/contents/28526

here the trick is to unpack the word *understanding* it’s not about what each neuron does, but the rules shaping the behaviour of biological systems! "rules for development and learning in brains may be far easier to understand than their resulting properties" 👇👌 arxiv.org/abs/1907.06374

Galalite Picture (Gz.III) 1932 by László Moholy-Nagy

Sensory reformatting for a working visual memory www.cell.com/trends/cogni... #neuroscience

Language is a great example - not necessary for intelligence to emerge, but might be enough to bootstrap. Once you’re using language data you’re not building intelligence de novo. I’d bet social interactions are the opposite - you need them for intelligence to emerge but not to bootstrap.

Always find it an interesting question which ingredients are necessary for intelligence to emerge from scratch vs which are necessary if you only want to bootstrap from an existing intelligent system…

Can consciousness arise purely from computation, or does it require the biological mechanisms that allow computing? The tension between computational roles and biological realizers may hold the key. Check this paper by @neddo.bsky.social @mkhochb.bsky.social www.sciencedirect.com/science/arti...

Wow, what a powerhouse of a paper! Physiology from three different sites in auditory cortex too. I'm a little surprised all birds were anesthetized, but still, amazing findings! Fellow ARO members should love this paper.

A study led by Cina Aghamohammadi is now out in ‪@natcomms.nature.com‬! We developed a mathematical framework for partitioning spiking variability, which revealed that spiking irregularity is nearly invariant for each neuron and decreases along the cortical hierarchy. www.nature.com/articles/s41...

Sparks (2001) by Andrew Wyeth

I'll pass them apart https://botfrens.com/collections/69/contents/20034

The world through infant eyes: Evidence for the early emergence of the cardinal orientation bias www.pnas.org/doi/full/10....

New paper dropped! Led by @thosvarley.bsky.social and Pai Vaibhav, we show that flows of information within a #xenobot resemble flows of information observed in adult human brains. Thomas provides caveats and context below. w/ @drmichaellevin.bsky.social, Caitlin Grasso, & Jeantine Lunshof.

AFF Architekten / Lichtenberg District Office of Berlin / Lew-Tolstoi-Schule / Building / 2022

Lunar Occultation of Saturn 9/17/2024 - From Ryan Kinnett (rkinnett.bsky.social) - https://flic.kr/p/2qh5d47

CASSINI (4) - From Jacint Roger (landru79.bsky.social) - https://flic.kr/p/2iKfns6

We’re recruiting a postdoctoral fellow to join our team! 🎉 I’m happy to share that I’ve opened back up the search for this position (it was temporarily closed due to funding uncertainty). See lab page and doc below for details!

Do AI reasoning models abstract and reason like humans? New paper on this from my group: arxiv.org/abs/2510.02125 🧵 1/10

Very excited to share that our work (together with co-first author Shanka Subhra Mondal and @neuroai.bsky.social ) on a brain-inspired architecture for planning with LLMs is now out in Nature Communications! www.nature.com/articles/s41... (thread below)

Composition in Color A https://www.wikiart.org/en/piet-mondrian/composition-in-color-a-1917

Titan on the side - From Val Klavans (valklavans.bsky.social) - https://flic.kr/p/dfiN8k

New preprint! How can you remember an image you saw once, even after seeing thousands of them? We find a role for humble mid-level visual cortex in high-capacity, one-shot learning. doi.org/10.1101/2025.09.22.677855 🧵🧪1/

NGC 1999 and HH1/HH2 in Orion - From Rolf Wahl Olsen - https://flic.kr/p/VmKZn6