Aaron Milstein

It's awkward that every awesome advance will help enable the robot apocalypse though

The U.S. Is Funding Fewer Grants in Every Area of Science and Medicine A quiet policy change means the government is making fewer bets on long-term science.

Under Trump, the NIH is giving fewer grants to fewer scientists. The grants awarded are smaller and scientists have less time to spend them.

Projects in cancer, diabetes, aging, neurological disorder, and more are going unfunded.

“Make America Healthy Again.”

Everyone looks thrilled to be there

Raises hand

Salesforce, the Aristocrats!

Stop discounting the essential role of frymbrial in autism

Stranger Things and Wicked too

There's new guidance with respect to #foreignawards at #NIH. Please share with your networks. #foreignsubs #funding

This field is on fire. So many cool methods these days. Happy Thanksgiving friends!

It's the lmgtfy of our era

Then you can't say "highly sparse." You'll have to call it "lowly dense."

Even with Brunel, sparsity is 1-f

How prior experience biases new learning for generalization? new preprint is out:
1) behavioral and hippocampal generalization co-emerged in a novel context.
2) BTSP was involved in forming cross-context stable representations, as well as aligning novel neural dynamics to familiar subspace.

a woman is surrounded by police officers and says i volunteer . ALT: a woman is surrounded by police officers and says i volunteer .

I volunteer as an editor!

Hmmmm. I have only ever seen it and used it in neuroscience as 1 - density which is indeed fraction of zeros. "Highly sparse" is literally a high value of the sparsity metric.

Yes but it would be the difference between Nature Communications and Nature Neuroscience, with a special emphasis on systems, circuits, and computation, which no one has done yet but obviously deserves its own journal!

Oh man. Science Neural Circuits would be my new favorite journal.

Infuriating are sites that won't allow .PDF and then have PDF generators that fuck up the document formatting.

Oh, the answer to RFK is to bring on more scientific illiterates?

I mean... I've been trying for years to regain login access so I can edit my tree. And the link to neurotree from the article didn't work. How did you build a time machine to let this guy write this article from 15 years ago?

Andrea is one of over 170 U.S. citizens who have detained by ICE. It should never have happened, and we will hold Kristi Noem and the Trump Administration accountable.

So the finding is a corroboration of prior work that deep networks without biological architectures, sparsity constraints, or learning rules build distributed representations robust to ablation of the most selective cells, and moderate selectivity can be found at initialization?

I'm definitely on board with looking at RNN dynamics! Even better with Dale's law constraints...

Ok, I can see how temporal jitter would throw off sliceTCA, but I think first finding cell x time latents that reconstruct trials, and then asking which latents align with decision variables could also highlight "non-task" dimensions

What are your thoughts on sliceTCA

🚨new paper: the head-direction circuit as a model of primary thalamocortical system. Check this out 👇

kudos to @adrian-du.bsky.social for the huge amount of work put into this opinion piece, starting with the beautiful figures comparing the different thalamocortical systems 🤩

Neural dynamics outside task-coding dimensions drive decision trajectories through transient amplification https://www.biorxiv.org/content/10.1101/2025.11.20.689599v1

It's about the IC Directors being political appointments, "FFS"

Sorry, WTAF?

Hi Jakob, have you considered implementing a lightweight wrapper around sbi as an optuna sampler? Was just thinking this would be a zero energy barrier for users to replace their current favorite sequential/population-based hyperparameter optimization algorithm