Dan Goodman

New preprint (well, very updated). 🤖🧠🧪 We find that an abstract model of neuromodulation lets spiking neural networks perform much better, particularly in challenging noisy environments, using less energy. Relevant to #neuroscience and #neuromorphic computing. 🧵👇 www.biorxiv.org/content/10.1...

80 replies when I saw this, of which precisely one is not negative. That one is not positive btw.

We are hiring at @neuromatch.bsky.social ! Come join us to help build the Computational Behaviour course!!

I'm happy to share my first contribution to The Transmitter. From its inception, AI has borrowed ideas and concepts from the brain's operations and mechanisms. The two field have a lot to gain from each other and we should work to make the bridge wider and stronger.

I'm suddenly seeing a lot more genuine email going into my spam folder. I assume this is because the rise of LLM-written spam means that email providers have had to dial up the sensitivity of their filters. Another great contribution from AI.

I've seen this rush to use oral assessment either as a replacement or addition to coursework (because of LLMs) and I've been resisting it for the reasons below, but also because I think it inevitably favours self confident people, which isn't what we should be assessing.

"Take Ofwat, the water regulator... the words river, pollution and sewage don’t appear in its general duties." Another brilliant post by @laurenleek.eu combining data analysis and incisive systems level thinking about society.

New blog post: "we should resist ... using journal publication record as a key indicator of applicant suitability, because it’s a system that shifts costs from the rich to the poor, slows down scientific progress as a whole, and bakes in discrimination" thesamovar.github.io/zavarka/hiri...

There's a paper going around saying that AI is now better at peer review than humans. I won't link to it because I assume that like all these AI papers it's just some sort of marketing. My only comment is that we don't have to critique this paper because we already know human PR isn't very good.

🚨 New preprint! 🧵 Brain regions 🧠 integrate information over different timescales ⏱️: short in sensorimotor cortex, long in association cortex. But do regions change their timescales when we engage in a task? And how does that flexibility develop? 👇 www.biorxiv.org/content/10.6...

Excited and proud to share this work, now published in PNAS. www.pnas.org/doi/10.1073/... In this work, we identify neuromodulation as a high-dimensional reconfiguration signal that not only changes individual functional properties but also the statistical architecture linking them.

Very happy to see that neural heterogeneity is getting more attention. It's a great time to be studying this.

Lots of interesting articles suddenly appearing on the topic of how we can understand systems as complex as a brain, whose very nature is to be flexible. All converging on the idea that the type of answers we've been looking for are wrong. I guess the race is on to find a new type of answer.

Brilliant news that this book is free for anyone to download. Haven't read it yet, but scanned the conclusion. Many things I don't agree with as a strong advocate of dramatic changes to the way we do peer review, but super clear and reasoned argument. 👏

Autumn view from my office just before heading home. #photography

How much representational diversity exists around a trained network? A lot, it turns out! Moving a ViT’s weights by 1.3% while preserving its predictions, its representations end up less similar to the original than a random-init ViT’s ‼️ New paper: Hessian Null Space Continuation (HNC) 💫🧵 (1/12)

I am really glad this work from the Statistical Physics of Cognition project is finally out as a preprint. It settles a lot of confusion in the work on the Brain Criticality Hypothesis by separating out the impact of stimulus and arousal on driving brain activity towards or away from criticality.

This is a great read. I will say that I think big-O notation and complexity analysis is useful and not just a mathematical game. But it's not a substitute for actually thinking. It's not the full story. Most of the textbooks you read about it do make this clear, even if people forget it in practice.

Hahahaha they generated their fake data using seed(42) which let someone exactly reproduce their pipeline. 😂

I love the return of autumn. Colder weather means it's time for orange coloured leaves and hearty orange coloured soup.

"This technology that is ruining education has incredible potential to save education, if only I could figure out how to use it better to improve my students’ education." buff.ly/cEmLBex

New post on Transformation: Cracks appear in OpenAI’s wall of proofs https://www.quantamagazine.org/updates/transformation/#167798

Thinking about and studying the way science functions and how it could do better is important work. It's amazing that the UK has a specific metascience unit to do this. But what a shame that instead of doing so they're investing in projects to study how AI can automate research evaluation.

In a world of ongoing post publication review you don't need retraction because when you open the paper you see all the comments showing the problems. The idea of retraction inherently builds in the idea of a reliable authority that is precisely the mechanism that makes fraudulent papers a problem.

Math isn't being devastated. A bunch of medium hard to quite hard problems that humans have made up and worked on extensively are getting finished off by AI. AI is also a great way for mathematicians to learn more math. The future is unpredictable, but I wouldn't declare devastation just yet.

This is going straight to the top of the to-read pile! Very excited to see this.

𝗛𝗲𝘁𝗲𝗿𝗮𝗿𝗰𝗵𝘆 𝗶𝗻 𝘁𝗵𝗲 𝗯𝗿𝗮𝗶𝗻 New preprint with Andrea Gambarotto. Neuroscience often treats control as a hierarchy: the structures behind a behavior hold a stable rank. We argue this mistakes a configuration for an architecture. 🧵 arxiv.org/abs/2610.04643 #neuroskyence

Fascinating insights into the early history of Zotero, including the fact that the name is the Albanian word for "to learn well or master". Hats off to everyone involved in this project. Also if you're not using Zotero, why not? Dump that closed rubbish you're using now and switch.

I first wrote about the need to recognize peer reviews as acts of scholarship back in 2006. It's a dream deferred, but not a dream denied. We're going to make this happen.

Applications are now open for the SWC Systems Neuroscience PhD Programme. Join us in London! 🎓 World-class neuroscience training 💰 Fully funded 4-year programme 🖥️Close links to @gatsbyucl.bsky.social Apply by 9 Nov: www.sainsburywellcome.org/web/content/...

Join us for 𝗖𝗼𝗻𝘃𝗲𝗿𝗴𝗲𝗻𝗰𝗲, a multidisciplinary gathering designed for researchers, funders, technologists, and policymakers to meet, exchange ideas & make new connections. 📅 June 16–18, 2027 | Montreal 👉 𝗦𝗮𝘃𝗲 𝘆𝗼𝘂𝗿 𝘀𝗲𝗮𝘁 | 𝗣𝗮𝗿𝘁𝗻𝗲𝗿 𝘄𝗶𝘁𝗵 𝘂𝘀 www.convergescience.org #𝗖𝗼𝗻𝘃𝗲𝗿𝗴𝗲𝗻𝗰𝗲𝗖𝗼𝗻𝗳 #OpenScience #CallFor…

🚨 Calling all brains — biological, artificial, and everything in between! Abstract submissions are now open for the UCL NeuroAI Annual Conference 2026 🧠🤖 ⏰ Abstract deadline: 16 Oct, 12pm Come share your work with the UCL NeuroAI community! forms.cloud.microsoft/Pages/Respon...

I am able to take on new doctoral students through this amazing initiative! If you are interested please reach out to me by email to discuss.

Registration and abstract submission are now open for UCL's NeuroAI conference @ucl-neuroai.bsky.social. Talks from: 🟠 Wolfgang Maass 🟣 @apalmigiano.bsky.social 🔵 Ellie Pavlik 🟠 Bernardo Sabatini 🟣 @nskat.bsky.social 🔵 Eleni Vasilaki And more! gordonstreetticketing.com/event/2026/1...

📣 New preprint from the lab! 🎉 www.biorxiv.org/content/10.6...

It's that time of year where the view from my window gets quite distracting. #photography

Just uploaded my #BernsteinConference talk on abstract models of neuromodulation in #SpikingNeuralNetworks. No video was taken, so it's just the slides. neural-reckoning.org/talk_2026_be...

Today, I'm launching my research group 🟣Neural forms🟠 @imperialcollegeldn.bsky.social We study how structure shapes function in artificial and biological neural networks. www.neural-forms.org 🤖🧠🧪 🧵 1/3

#SilentSunday #photography

Excited to first share @alishdipani.bsky.social 's study. Alish asked a (deceptively) simple Q: Are category-selective units in ANNs actually like category-selective regions in the brain? (why is this imp? read until the end). #neuroskyence #neuroscience #brain #neuroai

This was my first time going to #BernsteinConference and I really enjoyed it. Would definitely recommend going next year. Particularly enjoyed the workshops and posters. Overall vibe was 👇