Alex Williams
@itsneuronal.bsky.social
Asst Prof at NYU + Flatiron Institute
Computational + Statistical Neuroscience
https://neurostatslab.org/
Computational + Statistical Neuroscience
https://neurostatslab.org/
Reposted by Alex Williams
Our next paper on comparing dynamical systems (with special interest to artificial and biological neural networks) is out!! Joint work with @annhuang42.bsky.social , as well as @satpreetsingh.bsky.social , @leokoz8.bsky.social , Ila Fiete, and @kanakarajanphd.bsky.social : arxiv.org/pdf/2510.25943
November 10, 2025 at 4:16 PM
Our next paper on comparing dynamical systems (with special interest to artificial and biological neural networks) is out!! Joint work with @annhuang42.bsky.social , as well as @satpreetsingh.bsky.social , @leokoz8.bsky.social , Ila Fiete, and @kanakarajanphd.bsky.social : arxiv.org/pdf/2510.25943
Reposted by Alex Williams
ATTN: Two faculty positions are available in my department (neuroscience) at the University of Minnesota. This is a general search with no topic focus. November 20th application deadline.
Apply here: hr.umn.edu/jobs/Find-Job
Assistant Prof job code: 364920
Associate/Full Prof job code: 364921
Apply here: hr.umn.edu/jobs/Find-Job
Assistant Prof job code: 364920
Associate/Full Prof job code: 364921
October 30, 2025 at 3:59 PM
ATTN: Two faculty positions are available in my department (neuroscience) at the University of Minnesota. This is a general search with no topic focus. November 20th application deadline.
Apply here: hr.umn.edu/jobs/Find-Job
Assistant Prof job code: 364920
Associate/Full Prof job code: 364921
Apply here: hr.umn.edu/jobs/Find-Job
Assistant Prof job code: 364920
Associate/Full Prof job code: 364921
Reposted by Alex Williams
Thrilled that our paper is out today in Nature!
www.nature.com/articles/s4...
www.nature.com/articles/s4...
Sensory expectations shape neural population dynamics in motor circuits
Nature - Experiments with human volunteers and macaques show that expectations produced by probabilistic cueing of future sensory inputs shape motor circuit dynamics in order to increase the...
www.nature.com
October 29, 2025 at 4:11 PM
Thrilled that our paper is out today in Nature!
www.nature.com/articles/s4...
www.nature.com/articles/s4...
Reposted by Alex Williams
I wrote an explainer about the mass firings—the RIFs.
The RIFs all year have been clearly illegal. Russell Vought is using them to lawlessly clean out federal agencies. Lower court judges have said so. But the Calvinball Supreme Court has allowed them. 1/
The RIFs all year have been clearly illegal. Russell Vought is using them to lawlessly clean out federal agencies. Lower court judges have said so. But the Calvinball Supreme Court has allowed them. 1/
The mass firings have been lawless: now and all year
The first step in stopping a lawless Supreme Court is being clear about what it’s doing
scienceandfreedomalliance.substack.com
October 29, 2025 at 12:02 PM
I wrote an explainer about the mass firings—the RIFs.
The RIFs all year have been clearly illegal. Russell Vought is using them to lawlessly clean out federal agencies. Lower court judges have said so. But the Calvinball Supreme Court has allowed them. 1/
The RIFs all year have been clearly illegal. Russell Vought is using them to lawlessly clean out federal agencies. Lower court judges have said so. But the Calvinball Supreme Court has allowed them. 1/
Reposted by Alex Williams
Pleased to share new work with @sflippl.bsky.social @eberleoliver.bsky.social @thomasmcgee.bsky.social & undergrad interns at Institute for Pure and Applied Mathematics, UCLA.
Algorithmic Primitives and Compositional Geometry of Reasoning in Language Models
www.arxiv.org/pdf/2510.15987
🧵1/n
Algorithmic Primitives and Compositional Geometry of Reasoning in Language Models
www.arxiv.org/pdf/2510.15987
🧵1/n
October 27, 2025 at 6:13 PM
Pleased to share new work with @sflippl.bsky.social @eberleoliver.bsky.social @thomasmcgee.bsky.social & undergrad interns at Institute for Pure and Applied Mathematics, UCLA.
Algorithmic Primitives and Compositional Geometry of Reasoning in Language Models
www.arxiv.org/pdf/2510.15987
🧵1/n
Algorithmic Primitives and Compositional Geometry of Reasoning in Language Models
www.arxiv.org/pdf/2510.15987
🧵1/n
Reposted by Alex Williams
Fisher meets Feynman! 🤝
We use score matching and a trick from quantum field theory to make a product-of-experts family both expressive and efficient for variational inference.
To appear as a spotlight @ NeurIPS 2025.
#NeurIPS2025 (link below)
We use score matching and a trick from quantum field theory to make a product-of-experts family both expressive and efficient for variational inference.
To appear as a spotlight @ NeurIPS 2025.
#NeurIPS2025 (link below)
October 27, 2025 at 12:51 PM
Fisher meets Feynman! 🤝
We use score matching and a trick from quantum field theory to make a product-of-experts family both expressive and efficient for variational inference.
To appear as a spotlight @ NeurIPS 2025.
#NeurIPS2025 (link below)
We use score matching and a trick from quantum field theory to make a product-of-experts family both expressive and efficient for variational inference.
To appear as a spotlight @ NeurIPS 2025.
#NeurIPS2025 (link below)
Reposted by Alex Williams
Want the freedom of a fancy fellowship, but not the year-long wait or arduous application?
Come join my lab! Work on neuroscience and AI, explore your creativity, be independent or work closely with me, collaborate widely, and have a lot of fun!
my.corehr.com/pls/uoxrecru...
Come join my lab! Work on neuroscience and AI, explore your creativity, be independent or work closely with me, collaborate widely, and have a lot of fun!
my.corehr.com/pls/uoxrecru...
October 23, 2025 at 10:46 AM
Want the freedom of a fancy fellowship, but not the year-long wait or arduous application?
Come join my lab! Work on neuroscience and AI, explore your creativity, be independent or work closely with me, collaborate widely, and have a lot of fun!
my.corehr.com/pls/uoxrecru...
Come join my lab! Work on neuroscience and AI, explore your creativity, be independent or work closely with me, collaborate widely, and have a lot of fun!
my.corehr.com/pls/uoxrecru...
Hotter take: We should only be reporting nonasymptotic confidence intervals using empirical Bernstein inequality or something similar.
Dividing by sqrt(n) is incorrect unless n = infinity.
(To be clear, I am 100% trolling)
Dividing by sqrt(n) is incorrect unless n = infinity.
(To be clear, I am 100% trolling)
What's with machine learning researchers always reporting standard deviation instead of standard error? My understanding is that the error bars are typically used to back up inferential claims about significant differences between sample means (although statistical tests are rare, another problem).
October 27, 2025 at 1:51 AM
Hotter take: We should only be reporting nonasymptotic confidence intervals using empirical Bernstein inequality or something similar.
Dividing by sqrt(n) is incorrect unless n = infinity.
(To be clear, I am 100% trolling)
Dividing by sqrt(n) is incorrect unless n = infinity.
(To be clear, I am 100% trolling)
Reposted by Alex Williams
Looking for a PhD program where you can study computational neuroscience? NYU has a fantastic array of researchers covering the field:
groups.google.com/g/systems-ne...
groups.google.com/g/systems-ne...
Doctoral studies in Computational/Theoretical Neuroscience at NYU
groups.google.com
October 25, 2025 at 2:19 PM
Looking for a PhD program where you can study computational neuroscience? NYU has a fantastic array of researchers covering the field:
groups.google.com/g/systems-ne...
groups.google.com/g/systems-ne...
Reposted by Alex Williams
A new essay on the crazy, all or nothing approach to work happening in AI today, the looming human costs, and the lack of a finish line.
I wouldn't say it's okay, but I'm not sure how to fix it.
www.interconnects.ai/p/burning-out
I wouldn't say it's okay, but I'm not sure how to fix it.
www.interconnects.ai/p/burning-out
Burning out
The international AI industry's collective risk.
www.interconnects.ai
October 25, 2025 at 2:35 PM
A new essay on the crazy, all or nothing approach to work happening in AI today, the looming human costs, and the lack of a finish line.
I wouldn't say it's okay, but I'm not sure how to fix it.
www.interconnects.ai/p/burning-out
I wouldn't say it's okay, but I'm not sure how to fix it.
www.interconnects.ai/p/burning-out
Reposted by Alex Williams
Excited to share our new work with @engeltatiana.bsky.social!
RNNs are often used to explore how the brain may solve specific tasks. We show that, depending on the architecture, RNNs find distinct circuit solutions, behaving differently when exposed to novel stimuli.
www.nature.com/articles/s42...
RNNs are often used to explore how the brain may solve specific tasks. We show that, depending on the architecture, RNNs find distinct circuit solutions, behaving differently when exposed to novel stimuli.
www.nature.com/articles/s42...
Single-unit activations confer inductive biases for emergent circuit solutions to cognitive tasks - Nature Machine Intelligence
Recurrent neural networks are widely used to model brain dynamics. Tolmachev and Engel show that single-unit activation functions influence task solutions that emerge in trained networks, raising the ...
www.nature.com
October 24, 2025 at 7:14 PM
Excited to share our new work with @engeltatiana.bsky.social!
RNNs are often used to explore how the brain may solve specific tasks. We show that, depending on the architecture, RNNs find distinct circuit solutions, behaving differently when exposed to novel stimuli.
www.nature.com/articles/s42...
RNNs are often used to explore how the brain may solve specific tasks. We show that, depending on the architecture, RNNs find distinct circuit solutions, behaving differently when exposed to novel stimuli.
www.nature.com/articles/s42...
Reposted by Alex Williams
The NIH institute director firing last Friday is very bad.
I made a video explainer about why.
Stay for last post, w link to @science.org story from @jocelynkaiser.bsky.social
1/4
🧪
I made a video explainer about why.
Stay for last post, w link to @science.org story from @jocelynkaiser.bsky.social
1/4
🧪
October 23, 2025 at 4:02 AM
The NIH institute director firing last Friday is very bad.
I made a video explainer about why.
Stay for last post, w link to @science.org story from @jocelynkaiser.bsky.social
1/4
🧪
I made a video explainer about why.
Stay for last post, w link to @science.org story from @jocelynkaiser.bsky.social
1/4
🧪
Reposted by Alex Williams
🧠🌟🐭 Excited to share some of my postdoc work on the evolution of dexterity!
We compared deer mice evolved in forest vs prairie habitats. We found that forest mice have:
(1) more corticospinal neurons (CSNs)
(2) better hand dexterity
(3) more dexterous climbing, which is linked to CSN number🧵
We compared deer mice evolved in forest vs prairie habitats. We found that forest mice have:
(1) more corticospinal neurons (CSNs)
(2) better hand dexterity
(3) more dexterous climbing, which is linked to CSN number🧵
October 22, 2025 at 8:41 PM
🧠🌟🐭 Excited to share some of my postdoc work on the evolution of dexterity!
We compared deer mice evolved in forest vs prairie habitats. We found that forest mice have:
(1) more corticospinal neurons (CSNs)
(2) better hand dexterity
(3) more dexterous climbing, which is linked to CSN number🧵
We compared deer mice evolved in forest vs prairie habitats. We found that forest mice have:
(1) more corticospinal neurons (CSNs)
(2) better hand dexterity
(3) more dexterous climbing, which is linked to CSN number🧵
Reposted by Alex Williams
Delighted to share our new paper, now out in PNAS! www.pnas.org/doi/10.1073/...
"Hierarchical dynamic coding coordinates speech comprehension in the brain"
with dream team @alecmarantz.bsky.social, @davidpoeppel.bsky.social, @jeanremiking.bsky.social
Summary 👇
1/8
"Hierarchical dynamic coding coordinates speech comprehension in the brain"
with dream team @alecmarantz.bsky.social, @davidpoeppel.bsky.social, @jeanremiking.bsky.social
Summary 👇
1/8
PNAS
Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...
www.pnas.org
October 22, 2025 at 5:21 AM
Delighted to share our new paper, now out in PNAS! www.pnas.org/doi/10.1073/...
"Hierarchical dynamic coding coordinates speech comprehension in the brain"
with dream team @alecmarantz.bsky.social, @davidpoeppel.bsky.social, @jeanremiking.bsky.social
Summary 👇
1/8
"Hierarchical dynamic coding coordinates speech comprehension in the brain"
with dream team @alecmarantz.bsky.social, @davidpoeppel.bsky.social, @jeanremiking.bsky.social
Summary 👇
1/8
Reposted by Alex Williams
🚨Pre-print alert! 🚨https://www.biorxiv.org/content/10.1101/2025.10.17.683171v1
Our new study tackles the question: do all neurons in motor cortices (MC) encode movement & coordinate as we move? Answering this question will be key for effectively targeting motor representations in BCIs.
Our new study tackles the question: do all neurons in motor cortices (MC) encode movement & coordinate as we move? Answering this question will be key for effectively targeting motor representations in BCIs.
The spatiotemporal structure of neural activity in motor cortex during reaching
Intracortical brain-computer interfaces (BCI) leverage knowledge about neural representations to translate movement-related neural activity into actions. BCI implants have targeted broad cortical regi...
www.biorxiv.org
October 20, 2025 at 6:54 PM
🚨Pre-print alert! 🚨https://www.biorxiv.org/content/10.1101/2025.10.17.683171v1
Our new study tackles the question: do all neurons in motor cortices (MC) encode movement & coordinate as we move? Answering this question will be key for effectively targeting motor representations in BCIs.
Our new study tackles the question: do all neurons in motor cortices (MC) encode movement & coordinate as we move? Answering this question will be key for effectively targeting motor representations in BCIs.
Reposted by Alex Williams
Today in Nature Machine Intelligence, Kazuki Irie & I discuss 4 classic challenges for neural nets — systematic generalization, catastrophic forgetting, few-shot learning, & reasoning. We argue there is a unifying fix: the right incentives & practice. rdcu.be/eLRmg
October 20, 2025 at 1:18 PM
Today in Nature Machine Intelligence, Kazuki Irie & I discuss 4 classic challenges for neural nets — systematic generalization, catastrophic forgetting, few-shot learning, & reasoning. We argue there is a unifying fix: the right incentives & practice. rdcu.be/eLRmg
Reposted by Alex Williams
This one hits close to home. Jim is my office neighbor. Absolutely devastating.
A Cornell University researcher has been developing an artificial heart for children for more than 20 years. Now, his research is on hold and his lab is shut down.
This 4-year-old's heart is failing. A federal grant that might help him was canceled
A Cornell University researcher has been developing an artificial heart for children for more than 20 years. Now, his research is on hold and his lab is shut down.
n.pr
October 12, 2025 at 10:19 PM
This one hits close to home. Jim is my office neighbor. Absolutely devastating.
Reposted by Alex Williams
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...
www.nature.com/articles/s41...
October 12, 2025 at 12:42 AM
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...
www.nature.com/articles/s41...
Reposted by Alex Williams
Over the past year, my lab has been working on fleshing out theory + applications of the Platonic Representation Hypothesis.
Today I want to share two new works on this topic:
Eliciting higher alignment: arxiv.org/abs/2510.02425
Unpaired learning of unified reps: arxiv.org/abs/2510.08492
1/9
Today I want to share two new works on this topic:
Eliciting higher alignment: arxiv.org/abs/2510.02425
Unpaired learning of unified reps: arxiv.org/abs/2510.08492
1/9
October 10, 2025 at 10:13 PM
Over the past year, my lab has been working on fleshing out theory + applications of the Platonic Representation Hypothesis.
Today I want to share two new works on this topic:
Eliciting higher alignment: arxiv.org/abs/2510.02425
Unpaired learning of unified reps: arxiv.org/abs/2510.08492
1/9
Today I want to share two new works on this topic:
Eliciting higher alignment: arxiv.org/abs/2510.02425
Unpaired learning of unified reps: arxiv.org/abs/2510.08492
1/9
Reposted by Alex Williams
Finally out: our recent work with Nick Betley is a view into how the brain reshapes its behavior in the face of competing survival needs- and also a potential angle on treatment targets for enduring pain.
A brief rundown...
www.nature.com/articles/s41...
A brief rundown...
www.nature.com/articles/s41...
A parabrachial hub for need-state control of enduring pain - Nature
Activity in a set of parabranchial neurons in the mouse brain is increased during chronic pain, predicts coping behaviour, and can be modulated by circuits activated by survival threats.
www.nature.com
October 9, 2025 at 1:15 PM
Finally out: our recent work with Nick Betley is a view into how the brain reshapes its behavior in the face of competing survival needs- and also a potential angle on treatment targets for enduring pain.
A brief rundown...
www.nature.com/articles/s41...
A brief rundown...
www.nature.com/articles/s41...
Reposted by Alex Williams
Six of the nine Nobel Prize winners this year work in the U.S.
Three of the six were born outside the U.S., which is the pattern most years. No country has benefited more from welcoming immigrants from around the world.
www.nobelprize.org
Three of the six were born outside the U.S., which is the pattern most years. No country has benefited more from welcoming immigrants from around the world.
www.nobelprize.org
The official website of the Nobel Prize - NobelPrize.org
The Nobel Prize rewards science, humanism and peace efforts. This is one of the central concepts in the will of Alfred Nobel, and it also permeates the outreach activities that have been developed for...
www.nobelprize.org
October 8, 2025 at 12:37 PM
Six of the nine Nobel Prize winners this year work in the U.S.
Three of the six were born outside the U.S., which is the pattern most years. No country has benefited more from welcoming immigrants from around the world.
www.nobelprize.org
Three of the six were born outside the U.S., which is the pattern most years. No country has benefited more from welcoming immigrants from around the world.
www.nobelprize.org
Reposted by Alex Williams
Do AI reasoning models abstract and reason like humans?
New paper on this from my group:
arxiv.org/abs/2510.02125
🧵 1/10
New paper on this from my group:
arxiv.org/abs/2510.02125
🧵 1/10
Do AI Models Perform Human-like Abstract Reasoning Across Modalities?
OpenAI's o3-preview reasoning model exceeded human accuracy on the ARC-AGI benchmark, but does that mean state-of-the-art models recognize and reason with the abstractions that the task creators inten...
arxiv.org
October 6, 2025 at 9:27 PM
Do AI reasoning models abstract and reason like humans?
New paper on this from my group:
arxiv.org/abs/2510.02125
🧵 1/10
New paper on this from my group:
arxiv.org/abs/2510.02125
🧵 1/10
Reposted by Alex Williams
Federal workers are speaking out on how the Trump regime is harming Americans:
"We stand here today at a moment of the greatest assault of our lifetimes on our freedoms. The US Constitution is being shredded... We are here to call on Congress to stand up and stop it."
youtu.be/uw1KwIBgrCw?...
"We stand here today at a moment of the greatest assault of our lifetimes on our freedoms. The US Constitution is being shredded... We are here to call on Congress to stand up and stop it."
youtu.be/uw1KwIBgrCw?...
Mark Histed, NIH Scientist *speaking in personal capacity
YouTube video by Civil Servants Coalition
youtu.be
October 2, 2025 at 2:46 AM
Federal workers are speaking out on how the Trump regime is harming Americans:
"We stand here today at a moment of the greatest assault of our lifetimes on our freedoms. The US Constitution is being shredded... We are here to call on Congress to stand up and stop it."
youtu.be/uw1KwIBgrCw?...
"We stand here today at a moment of the greatest assault of our lifetimes on our freedoms. The US Constitution is being shredded... We are here to call on Congress to stand up and stop it."
youtu.be/uw1KwIBgrCw?...
Reposted by Alex Williams
The neural control & computation lab is recruiting!
If you're interested in using large-scale neural population recordings to study how the brain learns to produce complex and flexible behaviours, please get in touch.
www.ncclab.ca
If you're interested in using large-scale neural population recordings to study how the brain learns to produce complex and flexible behaviours, please get in touch.
www.ncclab.ca
October 1, 2025 at 11:23 PM
The neural control & computation lab is recruiting!
If you're interested in using large-scale neural population recordings to study how the brain learns to produce complex and flexible behaviours, please get in touch.
www.ncclab.ca
If you're interested in using large-scale neural population recordings to study how the brain learns to produce complex and flexible behaviours, please get in touch.
www.ncclab.ca
Reposted by Alex Williams
Many treat uncertainty = a number. At Apple, we're rethinking this: LLMs should output strings that reveal all information of their internal distributions. We find that Reasoning, SFT, CoT can't do it - yet. To get there, we introduce the SelfReflect benchmark.
arxiv.org/pdf/2505.20295
arxiv.org/pdf/2505.20295
October 1, 2025 at 9:53 AM
Many treat uncertainty = a number. At Apple, we're rethinking this: LLMs should output strings that reveal all information of their internal distributions. We find that Reasoning, SFT, CoT can't do it - yet. To get there, we introduce the SelfReflect benchmark.
arxiv.org/pdf/2505.20295
arxiv.org/pdf/2505.20295