Incredibly interesting paper by @tknapen.bsky.social and colleagues. Retinotopic reference frame for the win!

Children acquire object category representations from their everyday experiences in the first few years of life. What do the inputs to this learning process actually look like? New preprint! arxiv.org/abs/2605.14990

🚨 The actual paper is now out on @cp-trendscognsci.bsky.social: www.sciencedirect.com/science/arti... In this opinion piece, @duncanastle.bsky.social and I explore the human brain through the lens of optimality and Pareto-optimality theory.

New preprint with @lingqiz.bsky.social: Neurodata Without Boredom: Benchmarking Agentic AI for Data Reuse arxiv.org/abs/2605.12808 1/10

Yesterday I argued that cognitive "modules" could be seen as deep nonlinear networks, which could interact via linear (or linear-ish) mappings between their outputs, because deep networks can output rich representations suitable for linear decoding. bsky.app/profile/neur... But you might ask...

1/ Check out our new modeling paper on predictive learning in the hippocampus! 🧠 www.biorxiv.org/content/10.6... A central idea in neuroscience is that the brain makes sense of the world by learning to predict its environment 🧠➡️🌍.

A huge effort and a new take on what constitutes the "feedforward pass" across the visual system. Make sure to check out our new preprint:

Excited about our new preprint: “The illusory simplicity of the feedforward pass: evidence for the dynamical nature of stimulus encoding along the primate ventral stream” arxiv.org/abs/2604.12825 Work with Sushrut Thorat, Anna Mitola, Paolo Papale, Peter König & Tim Kietzmann 🧵 thread below

I’ve officially aged out of 40 Under 40 eligibility. Officially signing up for 50 Under 50.

1/7 New paper accepted as ICML spotlight arxiv.org/abs/2605.03517! We unify self-supervised learning (SSL) algorithms (e.g., contrastive, VICReg, stopgrad) via latent distribution matching (LDM), which matches an induced latent distribution to an explicit latent model.

The "Easy" and the "Hard" problem of understanding. Solutions to the easy problem of understanding (EPU) are inherently prediction-friendly, measurable, and reportable. Solutions to the hard problem of understanding (HPU) are phenomenological, subjective, and beyond prediction. 1/2

Autonomous error detection is enabled by conflict-dependent forward models in human medial frontal cortex www.biorxiv.org/content/10.6... whoa…

Fascinating new research from UWaterloo Psychology’s @ccolomb.bsky.social Clara Colombatto and colleague (@smfleming.bsky.social Steve Fleming) examines attributions of metacognitive states in human-AI interactions. More here: www.nature.com/articles/s44...

Very excited to share our latest paper led by @nbinish.bsky.social in @natneuro.nature.com We demonstrate how a communication subspace channels higher-dimensional PFC dynamics into lower-D motor activity to enable efficient behavior using human iEEG. www.nature.com/articles/s41... #neuroskyence

the ppl we thank in acknowledgement sections are "a more relevant predictor of publication success than formal collaborations (i.e., coauthorship), even after matching for gender, seniority, methodological orientation, geographical location, and institutional prestige." www.pnas.org/doi/10.1073/...

Our new study (Debray*, Karami* et al.) explores how the brain encodes basic math concepts (integers, fractions, shapes). Samuel Debray led the 7T fMRI & behavioral work; I led the MEG. @danielavalerio.bsky.social @chrplr.bsky.social @standehaene.bsky.social www.biorxiv.org/content/10.6...

For many Iranian students, Neuromatch was one of the few bridges to the international neuroscience community. Iranian students and TAs have contributed so much to it. Many of us, including me, found neuroscience through Neuromatch. This decision by Trump’s gov is cruel, unjust and a loss for science

The Taliban banned girls from attending school. The Trump administration banned Iranians from attending an online summer school. Different reasons, same shit!

Neuromatch has been denied our Office of Foreign Assets Control "Iran license" renewal. We had renewed successfully before, but now our request is “inconsistent with current U.S. foreign policy.” I am absolutely gutted by this denial.

🧠Introducing OmniMouse: One of the largest datasets in neuroscience along with a systematic study of scaling properties of brain models Co-led by 🤩 K. Willeke, @pollytur.bsky.social and @alexrgil14.bsky.social Models trained on up to 3M neurons, >150B tokens

The uninterrupted flow of the interviewees' stories and thoughts is surprisingly pleasant in this podcast; even knowing that the person on the other side of the interviews has so much he could add.

I’ve been really enjoying Tom Griffith’s "The Cognition Project" lately. A couple of observations: - Hearing the history of the cognitive revolution directly from the giants who lived it is priceless. - Recent AI advances have sort of robbed us of the joy of being amazed by human cognition. 🧠🤖

#NeurIPS developing sub-community-specific reviewing standards is a huge plus this year. Also ... looking at the suggestions in my sub-communities, it's super interesting to see just how much expectations and maturity vary across different sub-fields.

"The tragedy of the em dash" The title of a book I’ll be writing soon.

www.sciencedirect.com/science/arti... very nice paper on a mice model of blindsight i'm not sure if i'm fully convinced by their saliency control, but maybe i'm just picking hair let me explain: (a thread to follow)

New preprint! 🧠 How do RNNs learn abstract rules from sequences, independent of specific stimuli? By Vezha Boboeva, with Alberto Pezzotta & George Dimitriadis "From sequences to schemas: low-rank recurrent dynamics underlie abstract relational representations" www.biorxiv.org/content/10.6...

New preprint from my lab! We study how reinforcement learning & selective attention interact. To do so, we built a set of models describing different ways that value & reward prediction error can modulate top-down attention. We compare model outcomes to monkey data from a color value learning task

Check the lecture notes I contributed to: a lecture note from Analytical Connectionism 2024 “A Computational Basis of Natural Intelligence” (PMLR, Analytical Connectionism 2024), based on lectures by Jonathan D. Cohen.

A neurodevelopment-inspired warm-up strategy to address uncertainty calibration: networks are briefly trained on random noise and labels before exposure to real data, leading to well-calibrated confidence and strong detection of unknown inputs. Cool results! #NeuroAI www.nature.com/articles/s42...

My first foray into explicitly trying to bridge Marr’s levels, with @bealebrains.bsky.social. Inspired by Hahn and Wei’s models (pubmed.ncbi.nlm.nih.gov/38360947/), we wondered how the brain could instantiate sensory inference with efficient /decoding/ properties. 1/

Bitter reviewer rant alert! 🚨 Nothing kills the motivation to do review like being "uninvited" when you’re only two days late; especially after keeping the editor updated. I’m nearly finished, I take the work seriously, and I’m doing it on top of a million other things 🙄

Wonderful work by Amin!

Want a dataset to test ideas on neural basis of decision making or how areas interact as we make choices? Check out our data published today @rudebecklab.bsky.social. >16,000 single neurons from 22 anatomically confirmed areas in macaques performing a decision task. www.nature.com/articles/s41...

What are the systems in neuroscience that we really have something that we can call “explanation” at all relevant levels, other than reflexive feed-forward like circuits. Here are a few that I would argue are getting there. Obviously not complete explanations but genuinely satisfying.

A perfect antidote to the Eon story. The RL component is exactly what Eon’s press release tried to sweep under the rug.

🧵 New preprint led by @bingbrunton.bsky.social, @elliottabe.bsky.social, @lawrencehu.bsky.social We gave a worm brain control of a fly body and it walked What did we learn? Nothing, other than deep reinforcement learning is effective We call it the digital sphinx www.biorxiv.org/content/10.6...

How do neural circuits in the brain implement normalization? 🧠 In our new paper, we show that just normalizing sensory input isn't enough. Crucially, we must also normalize the error signals! 🧵👇 Paper: arxiv.org/abs/2603.17676

Here's a lovely #blueprint on a new study from our lab led by @royeyono.bsky.social. tl;dr: it implies that there may be interneurons whose role is to normalize credit assignment signals during learning. #neuroscience 🧪

Excellent summary of the issues surrounding the 'uploaded' fruit fly. As I noted in my response to @theroberthart.bsky.social, given the flashy claims, it’s our responsibility to set a high bar for evidence. It definitely needs to be more than just video demos.

New paper! We introduce JEDI, Jointly Embedded Dynamics Inference for neural dynamics. arxiv.org/abs/2603.10489. JEDI flexibly infers dynamical principles (across behaviors/contexts) from neural population data through RNNs constrained at single-neuron resolution to reproduce that data.

Pleased to share that our paper "Representation Biases: Variance is Not Always a Good Proxy for Importance" is now out as Theory/New Concepts paper in eNeuro! www.eneuro.org/content/13/3... 1/

Viral claim on X: This is a "real uploaded animal" with “91% behavior accuracy” @theroberthart.bsky.social with @shahabbakht.bsky.social @asbates.bsky.social @anayebi.bsky.social @birchlse.bsky.social @tom-mcclelland.bsky.social: Is it though?

To my colleagues: as you review the next round of applications, please take into account the unique challenges these candidates have faced. I’m personally witnessing how hard they strive to stay connected despite the brutality of the regime and a devastating war.

Today I received a note from a grad student who lives in Tehran. Her note gives you firsthand experience of what it’s like to live in a city that is being bombed, and what it’s like to be young and feel despair about your future. rezashadmehr.blogspot.com/2026/03/hope...

This is super cool! The neuroscience implications are also interesting. Putting developmental stage aside, does the adult brain even need complicated and expensive credit assignment to learn? Perhaps in situations where the animal is pushed outside its ecological niche (eg experiment tasks)?

Sharing “Neural Thickets”. We find: In large models, the neighborhood around pretrained weights can become dense with task-improving solutions. In this regime, post-training can be easy; even random guessing works Paper: arxiv.org/abs/2603.12228 Web: thickets.mit.edu 1/

Heading to #COSYNE2026 Sadly missing most of the main conference, but will be there for the workshops. Let’s catch up if you’re around! 🧠🤖

(1/n) Presenting our NeurIPS’25 work at #cosyne2026 poster 3-109 We introduce a flexible framework to infer how animals learn new decision-making tasks from scratch without assuming REINFORCE/Q-learning. Paper: arxiv.org/abs/2509.04661 Joint w/ @vgeadah.bsky.social and @jpillowtime.bsky.social

New paper hot off the (pre-)press! We dig into the evolutionary origins of neural computations for behavioral control across mice, monkeys, and humans: www.biorxiv.org/content/10.6.... As our lab's first foray into comparative analysis of neural dynamics, I’m super excited about this work! 1/18

The ‘cyber fly’ is all over the internet and seems to have deeply captured the public imagination. Did I miss the conversation about it here? 🧠🤖