For the past few years, I've been trying to create a healthier ecosystem for online dating. The idea is that instead of browsing people's pictures, you are first shown their Tile -- a rectangle where you write 7 things about yourself. Join us at NotAZombie.net! And help spread the word!

Brains (and RNNs) are noisy and it turns out this has mega implications for working memory🧠 Helping this project (my first in NeuroAI!) was a learning curve so I’m super grateful to @fatihdin4en and team for including me. Preprint: biorxiv.org/content/10.64898…

A thalamus–brainstem attractor network drives history-biased decisions nature.com/articles/s41586-0…

What actually is the sensible recommendation now for tech/adjacent people who can't find a job in the field? Keep grinding the application process? Wait until the market improves? Retrain in something totally different? If so, what?

Using the MICrONS mouse visual cortex connectome from @AllenInstitute, we explored how local dendritic spine organization relates to large-scale cortical connectivity. One of our findings is that longer pyramidal axons form a higher fraction of synapses onto dendritic spines. Moreover, along individual axons, spine targeting increases with distance from the soma. Great collaboration with @deangeckt, Michael Reimann, @yusterafa, @Segev_Lab. biorxiv.org/content/10.64898…

New work by @deangeckt from our lab in collaboration with the Yuste lab analyzing a piece of cortical connectivity

So proud of this paper led by Conor Dorian, in collaboration with Jiannis Taxidis and Ahmet Arac where we used 2P opto and calcium imaging to show BTSP creating non-spatial representations in hippocampus during working memory task performance. nature.com/articles/s41467-0…

Pick a character from history or fiction and come up with 7 short clever phrases for each that they'd use as a dating profile. If it's good, I may put it on the landing page at NotAZombie.net.

Well that can't be good.

The modern world is predominantly a result of both lack of control and an abdication of responsibility, rather than particularly malevolent control.

In a new Stanford study, law professors by far preferred Gemini 2.5 Pro's responses over those written by their peers when they were unaware of who wrote the answers.

Clearer picture

Can someone start a journal called “Cell Atlases” so that the rest of the journals can go back to publishing interesting things?

Its really embarrassing living in a time where "I'm [arbitrary age] and worried I haven't achieved anything" is so deeply relatable when most people who ever lived basically just grew crops or roamed the plains and then died in ultimate obscurity. It feels dumb thinking like that

Why does deep learning generalize? What does weight decay really do? Can algorithmic information theory address these questions? In my latest preprint, I give a proof that the minimum neural weight norm matches the minimum program length (aka Kolmogorov Complexity), up to a logarithmic factor. In other words, the neural network with the smallest possible weight norm (that fits the data) must encode the shortest program (that fits the data). The result only holds for fixed-precision neural nets: infinite precision nets can store infinite information with finite (small) weights. arxiv.org/abs/2605.10878

LLMs represent concepts as vectors. Strikingly, taxonomies (organism → animal → bird) appear as hierarchies in embedding space. Led by my student @AndresNava, we show this comes from co-occurrence statistics alone.📄 arxiv.org/abs/2605.23821

Neural representation of action symbols (=recombinable representations of discrete units of motor behaviour) in primate frontal cortex nature.com/articles/s41586-0… 👉action elements (strokes) exhibit invariance, categorial structure and recombination into novel sequences

(Whether or not this is a good idea is an exercise left for the reader.)