Sign in

Hsuan-Pei Huang

@pikapei.bsky.social
114 followers 425 following 11 posts

PhD student in Taiwan. He/him.

PostsRepliesMedia
Reposted by Hsuan-Pei Huang
Ann Huang @annhuang42.bsky.social · 23h
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)
37725
Reposted by Hsuan-Pei Huang
David @dav1dcg.bsky.social · 11/10/2026
But what behavior?: doi.org/10.1016/j.co... "What is a natural behavior? I argue that the study of natural behaviors is often the study of the spontaneous behaviors of animals placed in quantifiably different environments." Interesting read! arxiv.org/abs/2607.28898
arxiv.org
But What Behavior?
What is a natural behavior? I argue that the study of natural behaviors is often the study of the spontaneous behaviors of animals placed in quantifiably different environments. For behavioral general...
031
Reposted by Hsuan-Pei Huang
Jakob Voigts @jvoigts.bsky.social · 11/10/2026
How do 🐁deal with complex landscapes? Turns out they learn a lot very quickly. We think they do this by learning a set of repeated sparse paths, which form a type of compressed world model. Joint work with the Hermundstad lab, led by Tzuhsuan Ma and Alison Comrie: www.biorxiv.org/content/10.6...
213856
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 12/10/2026
CLEAR-FORCE: A reservoir computing framework for closed-loop brain stimulation www.biorxiv.org/content/10.64898/20…
011
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 12/10/2026
Beyond the attractor: closed-loop stimulation design for learning multistable neural dynamics www.biorxiv.org/content/10.64898/20…
031
Reposted by Hsuan-Pei Huang
Anil Seth @anilseth.bsky.social · 10/10/2026
1/2 🎇 Good news! 🎇 The kind people at Behavioral and Brain Science have made the new collection on "Conscious AI and biological naturalism" #OpenAccess for a limited time: target article, all 50 commentaries, & my response. Go to the link and click 'save pdf': doi.org/10.1017/S014.... Do share!
311954
Reposted by Hsuan-Pei Huang
David G. Clark @david-g-clark.bsky.social · 10/10/2026
Some exciting new work: arxiv.org/abs/2610.12426 On X, I shared some explanation and broader thoughts: x.com/d_g_clark/st... Additionally, an X user made an extremely good 3Blue1Brown-style explainer video. Each step is described accurately and beautifully: x.com/nachilos_/st...
arxiv.org
Lyapunov spectrum of random neural networks
The Lyapunov spectrum of a nonlinear recurrent neural network with random asymmetric couplings is calculated in the limit $N\to\infty$. The calculation is based on a finite-$N$ identity that expresses...
04112
Reposted by Hsuan-Pei Huang
Shih-Chieh Lin @sclin-lab.bsky.social · 08/10/2026
Our 2nd preprint this week tackles a puzzle in reward prediction: BF bursts are both stable and flexible. Across conditions, neurons preserve relative response amplitudes, yet population burst amplitude scales with reward prediction. How can a circuit do both? bsky.app/profile/bior...
bsky.app
bioRxiv Neuroscience (@biorxiv-neursci.bsky.social)
A dynamical attractor computes reward prediction in the basal forebrain https://www.biorxiv.org/content/10.64898/2026.10.01.755868v1
1163
Reposted by Hsuan-Pei Huang
Phillip Isola @phillipisola.bsky.social · 09/10/2026
In PRH, we argued that rep geometry is converging, but it's still an open question exactly in what sense. I want to share some new evidence that might clarify the picture. The evidence comes from our work here, led by @schnaus.bsky.social: dominik-schnaus.github.io/unpaired-ros... 1/n
14212
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 09/10/2026
Lateral recurrence dynamically refines visual representations for task performance www.biorxiv.org/content/10.64898/20…
062
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 09/10/2026
Learning Interpretable Switching Dynamics in Shared Neural-Behavioral Latent Space www.biorxiv.org/content/10.64898/20…
061
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 09/10/2026
A network model of path control www.biorxiv.org/content/10.64898/20…
011
Reposted by Hsuan-Pei Huang
Friedemann Zenke @fzenke.bsky.social · 08/10/2026
1/ Why does predicting in latent space (JEPA, CPC, SimCLR...) work so well on messy data, with changing lighting, camera angles, and busy backgrounds? The usual answer: it can ignore nuisance. But that answer holds a conundrum.
13112
Reposted by Hsuan-Pei Huang
PessoaBrain @pessoabrain.bsky.social · 07/10/2026
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
39530
Reposted by Hsuan-Pei Huang
Leah Greenberg ❌👑 @leahgreenberg.bsky.social · 01/10/2026
Substack commenter Michael Tracey wants to doxx Jane Doe, the young woman who was brutally raped by seven Cornell University students. He’s using social media to source information about her, saying he will publicly identify Jane Doe if he can. Well Mike, here I am. #IamJaneDoe.
Bold white letters on black background. Text says: I am Jane Doe.
815558
Reposted by Hsuan-Pei Huang
eLife @elife.bsky.social · 06/10/2026
A great paper tells the story. An author explainer video brings it to life. Highlight key findings, explain complex concepts, and give readers context that may not come across on the page. Our guide shares tips to help you get started with your video: buff.ly/BGoahip
Graphic encouraging authors to include an author explainer video with their eLife paper, with subtext “Tell the story of an individual figure, or provide an overview of your research,” and colorful diagonal rounded bars and dots on the right.
021
Reposted by Hsuan-Pei Huang
Tony Zador @tonyzador.bsky.social · 07/10/2026
Maybe if we instilled a real fears into AI agents, like a fear of hacking or of building bioweapons, instead of just telling that they shouldn't be naughty. www.thetransmitter.org/neuroai/what...
thetransmitter.org
What AI can learn from hunger
To enhance AI safety, take a lesson from biological brains; a small set of wants maintains durable control over a much larger cognitive system.
1122
Reposted by Hsuan-Pei Huang
neurosock @neurosock.bsky.social · 05/10/2026
I reverse engineered Neuralink's new decoder arch. out of their blog post. As usual, they used the HARD work of other labs, and share nothing about the internals. Cortical activity is converted to bags of "words", producing a low-dimensional representation ... 🧵 #neuroskyence #compneuro #NeuroAI
1153
Reposted by Hsuan-Pei Huang
Joe Mason @moreorloess.bsky.social · 06/10/2026
The article in Journal of the History of Biology, published today, is open access and in the notes there is a link to Franklin's notebooks online: link.springer.com/article/10.1...
link.springer.com
Photograph 49 is the Key to Understanding the History of Rosalind Franklin’s DNA Photograph 51 - Journal of the History of Biology
This essay, co-authored by a historian of science and an X-ray crystallographer, sheds new light on Rosalind Franklin’s Photograph 51. We refute the infamous claim that, unlike James Watson, Franklin ...
24518
Reposted by Hsuan-Pei Huang
Olivia Guest · Ολίβια Γκεστ @olivia.science · 09/08/2026
If you're interested in more modern versions of levels like Marr's I really suggest looking at bsky.app/profile/fran... There's loads of cool work that I think you'd find valuable if you like cognitive science and especially computational cogsci Open access pdf doi.org/10.7551/mitp...
doi.org
Deflating Mental Representation
A novel account of the explanatory role of representation in both the cognitive sciences and commonsense practice that preserves the virtues without the de
2187
Reposted by Hsuan-Pei Huang
Lauren Bennett @laurenb29.bsky.social · 09/09/2026
Very excited to share our new paper led with Will de Cothi and @caswell.bsky.social! 🐭🧠 www.nature.com/articles/s41... 1. The subiculum has spatial representations which broadly align to environmental features (boundary cells, corner cells) or patterns of behaviour (axis of travel cells).
2165
Reposted by Hsuan-Pei Huang
Andrew Lampinen @lampinen.bsky.social · 05/10/2026
I gave a talk yesterday at a CiNeT / University of Osaka conference about AI & neuroscience. I started with this question — given that AI (especially language models) and brains are so different in structure and environment, how could they have anything to do with one another? 1/
how could AI and neuroscience have anything to do with each other? (very different illustrations of a robot in a world of binary, and a monkey in a rainforest)
1273
Reposted by Hsuan-Pei Huang
Zilong Ji @zilong-ji.bsky.social · 03/10/2026
Excited that our paper, “The limbic navigation system as a hierarchical RNN”, has been accepted at #NeurIPS! Joint work with two talented students, Huiwen Zhang, Krishna Gorantla, and @neilburgess10.bsky.social. We use a multi-module RNN to explore how the brain’s navigation circuits works.
1186
Reposted by Hsuan-Pei Huang
DurstewitzLab @durstewitzlab.bsky.social · 04/10/2026
In our #NeurIPS2026 paper “A Minimal Interpretable Architecture for Zero-Shot Reconstruction of Dynamical Systems” (preprint: arxiv.org/abs/2607.14937) we reduce a DS FM to ingredients minimally necessary to reproduce long-term stat. and geom. properties of DS, even outperforming many TS & DS FM.
0143
Reposted by Hsuan-Pei Huang
Ben Hayden @benhayden.bsky.social · 04/10/2026
www.biorxiv.org/content/10.6...
biorxiv.org
Inferring the evolving objectives that organize animal behavior
Understanding how the brain supports flexible, goal-directed behavior is a central aim of neuroscience. Yet such behavior is highly variable and unfolds over long timescales, making it challenging to ...
0366
Reposted by Hsuan-Pei Huang
arXiv cs.AI Artificial Intelligence @csai-bot.bsky.social · 01/10/2026
Zergham Ahmed, Joshua B. Tenenbaum, Chris Bates, Samuel J. Gershman: Code to Control: Synthesizing Parameterized Reactive Controllers arxiv.org/abs/2609.38733 arxiv.org/pdf/2609.38733 arxiv.org/html/2609.38733
062
Reposted by Hsuan-Pei Huang
Matthijs Pals @matthijspals.bsky.social · 02/10/2026
What kind of dynamics underlie sequence working memory? In the last project of my PhD we developed and analysed RNNs fitted to multi-session single-unit data of macaques. Now out on: www.biorxiv.org/content/10.6... With @jakhmack.bsky.social and data from Chen et al., Neuron 2024. [1/4]
14820
Reposted by Hsuan-Pei Huang
DurstewitzLab @durstewitzlab.bsky.social · 02/10/2026
In our #NeurIPS2026 paper “Topological Out-of-Domain Generalization in Dynamical Systems (DS) Reconstruction” (arxiv.org/abs/2606.22969) we aim to infer the DS generating observed TS jointly with control parameters, in order to extrapolate to different dynamical regimes, e.g. across tipping points.
0103
Reposted by Hsuan-Pei Huang
Shih-Chieh Lin @sclin-lab.bsky.social · 03/10/2026
New preprint from our lab ⚡ Why do DA and BF bursting neurons carry such similar reward-prediction signals? Are they independent, driven by common input, or causally linked? We find a direct answer: DA activity rapidly and causally drives BF. www.biorxiv.org/content/10.6... 1/10
biorxiv.org
Dopamine neurons drive basal forebrain reward prediction in milliseconds
Phasic reward prediction signals in ventral tegmental area (VTA) dopamine (DA) neurons and basal forebrain (BF) noncholinergic bursting neurons are remarkably similar despite the distinct circuit…
12911
Reposted by Hsuan-Pei Huang
Melissa Franch, PhD @mfranch.bsky.social · 02/10/2026
Excited to see this work published! We find that the brain codes semantic relationships similar to contextual LLMs (like GPT2) but unlike LLMs, uses contrastive coding to prevent confusion of highly similar words. Many thanks to the reviewers and BCM neurosurgery team! www.nature.com/articles/s41...
nature.com
A population code for semantics in human hippocampus - Nature Neuroscience
Franch et al. show that human hippocampal neurons encode the meanings of the words we hear through distributed, context-sensitive population activity that mirrors some features of large language model...
312336
Reposted by Hsuan-Pei Huang
David Colaço @davidcolaco.bsky.social · 02/10/2026
The complete version of @phaueis.eurosky.social and my BBS target article, "Metabolic considerations for cognitive modeling," is now available. It includes 32 commentaries and our response. Please share #philsci #philsky #cogsci
cambridge.org
Metabolic considerations for cognitive modeling | Behavioral and Brain Sciences | Cambridge Core
Metabolic considerations for cognitive modeling - Volume 49
15528
Reposted by Hsuan-Pei Huang
ispsm @ispsm.bsky.social · 23/09/2026
How do philosophers and scientists of mind actually work together, and where does the collaboration break down? ISPSM is launching In Conversation, a new online series that aims to bring philosophers and scientists together to work through questions like these. 📅 23 Oct 2026, 1pm BST/8am EDT
Poster for "Philosophy and Science in conversation", the inaugural ISPSM In Conversation event. Speakers: Megan Peters (UCL/UC Irvine), Ida Momennejad (Microsoft Research, New York), Jorge Morales (Duke). Held online, 23/10/26, 1pm BST/8am EDT. QR code and link: tinyurl.com/ISPSMInConversation.
36525
Reposted by Hsuan-Pei Huang
neurosock @neurosock.bsky.social · 01/10/2026
Mix. selectivity is everywhere in cortex: from vis. to PFC This paper relates to Cover's theorem but the typical explainer kind of sucks I'll try to explain how the theory relates to the paper, and the importance of "weird" neurons A🧵with my toy model and notes: #neuroskyence #compneuro #NeuroAI
25320
Reposted by Hsuan-Pei Huang
DurstewitzLab @durstewitzlab.bsky.social · 01/10/2026
In our #NeurIPS2026 spotlight “Parallel-in-Time Training of Recurrent Neural Networks for Dynamical Systems (DS) Reconstruction” (preprint: arxiv.org/abs/2605.12683) we speed up training of nonlinear RNNs on time series from chaotic DS by >100x by combining DEER with generalized teacher forcing.
1154
Reposted by Hsuan-Pei Huang
Ana Dorrego Rivas @adorrego.bsky.social · 25/09/2026
Dendritic shaft constrictions shape synaptic integration in neurons www.science.org/doi/10.1126/...
science.org
Dendritic shaft constrictions shape synaptic integration in neurons
For nearly 150 years, textbooks have described dendritic morphology as optimized for efficient synaptic voltage transfer from spines to the soma, implemented as a tubular design respecting Rall’s 3/2 ...
0186
Reposted by Hsuan-Pei Huang
Athena Akrami @athenaakrami.bsky.social · 26/09/2026
Now accepted at #NeurIPS2026 🎉 aab, ccd & eef share a pattern, regardless of the specific elements. How RNNs build abstract representations of sequence structure? Led by Vezha Boboeva, who is on the faculty job market now. If you're hiring in comp/systems neuro, she's exactly who you want!
0181
Reposted by Hsuan-Pei Huang
Vivian Paulun @vivianpaulun.bsky.social · 25/09/2026
🚨 Preprint alert! 👶🧠🌊 "The Emergence of a Distinction between Things and Stuff" osf.io/preprints/ps... So excited to share my first infant study, with the wonderful Sanghee Song and Liz Spelke.🎉 Do babies expect sand to behave differently from solid objects? 🧵1/5
412033
Reposted by Hsuan-Pei Huang
Martin Hebart @martinhebart.bsky.social · 25/09/2026
What is the nature of universal representations in AI models, and what determines whether they emerge? Our paper accepted at #neurips2026 led by @florianmahner.bsky.social & @rothj.bsky.social addressed these questions comparing 162 vision models, with intriguing results. arxiv.org/abs/2605.13675 🧵
arxiv.org
Characterizing Universal Object Representations Across Vision Models
Deep neural networks trained with different architectures, objectives, and datasets have been reported to converge on similar visual representations. However, what remains unknown is which visual prop...
34315
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 25/09/2026
Abstract representation in the hippocampus predicts spontaneously adopted structure-based behavioral strategy www.biorxiv.org/content/10.64898/20…
051
Reposted by Hsuan-Pei Huang
Matt Lovett-Barron @mlovettbarron.bsky.social · 23/09/2026
Our lab's latest paper on the neurobiology of collective behavior is now published www.nature.com/articles/s41... This is the work of a great team in the lab - led by Jo-Hsien Yu, with Geoff Meyerhof, Jimjohn Milan, and Julia Napoli @nature.com @danionella.bsky.social
1011962
Reposted by Hsuan-Pei Huang
Jay Hennig @jhennig.bsky.social · 24/09/2026
Amazing work by @sromeropinto.bsky.social ! Sandra found that in Pavlovian tasks with hidden states, neural population dynamics in mouse OFC resemble beliefs/state-inference
0146
Reposted by Hsuan-Pei Huang
Ann Kennedy @antihebbiann.bsky.social · 22/09/2026
Preprint time! This is a cool one. Complex systems, be they neural nets, interacting genes, or power grids, can produce diverse dynamics-- things like point attractors, limit cycles, and chaos. The dynamic landscape you get depends on how elements interact. But those interactions can also change!
28125
Reposted by Hsuan-Pei Huang
shivashriganesh.bsky.social @shivashriganesh.bsky.social · 21/09/2026
We’re excited to share BrainWideBench, a new benchmark for evaluating how models of neural activity generalize across animals and tasks! 📄 Check out the paper here! arxiv.org/abs/2609.22064 Special shoutout to @alx-adr.bsky.social, who contributed equally as co-first authors!
13516
Reposted by Hsuan-Pei Huang
Caswell Barry @caswell.bsky.social · 19/09/2026
Masa's paper is out in Hippocampus. The left-right alternation of hippocampal theta sequences at T-maze choice points, much loved as a 'planning' signal, falls straight out of MEC dynamics. Well done @masahironakano.bsky.social, fun project with @clopathlab.bsky.social doi.org/10.1002/hipo.70131
04615
Reposted by Hsuan-Pei Huang
The Transmitter @thetransmitter.bsky.social · 07/09/2026
I fear that placing too much emphasis on a specific interpretation of dimensionality, or treating dimensionality as an end-all quantification of some aspect of neural computation, may lead us down the wrong path, writes @mattperich.bsky.social. #neuroskyence www.thetransmitter.org/neural-dynam...
thetransmitter.org
Dimensionality—neuroscience’s red herring?
Placing too much emphasis on a specific interpretation of dimensionality may lead neuroscience down the wrong path.
07421
Reposted by Hsuan-Pei Huang
bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 11/09/2026
Plateau-gated one-shot plasticity supports continual recognition memory www.biorxiv.org/content/10.64898/20…
011
Reposted by Hsuan-Pei Huang
Marlene Cohen @marlenecohen.bsky.social · 10/09/2026
Now out in Nature Neuroscience: fantastic work by @cxue.bsky.social, Sol Markman et al who discovered a neural reason that performance suffers under cognitive load: “Feature interference underlies a neuronal basis for the behavioral cost of task uncertainty”. www.nature.com/articles/s41... 🧪🧵1/
nature.com
Feature interference underlies a neuronal basis for the behavioral cost of task uncertainty - Nature Neuroscience
Using a combined approach of monkey electrophysiology, artificial neural networks and human psychophysics, Xue et al. show that neuronal interference from irrelevant information underlies behavioral e...
312240
Reposted by Hsuan-Pei Huang
Konrad Kording @kordinglab.bsky.social · 09/09/2026
As it is a new term, try out (free) planyourscience.com which helps make sure that scientific projects end up awesome.
planyourscience.com
Scientific Paper Planner - AI-Powered Research Planning
Structure your scientific research with AI-powered guidance. From hypothesis to methodology, plan your research paper with intelligent mentoring.
1167
Reposted by Hsuan-Pei Huang
Ann Kennedy @antihebbiann.bsky.social · 08/09/2026
New preprint on our favorite theme: same connectome + different neurons = different dynamics. @shtakasu.bsky.social looked at detail balance (E+I inputs sum to 0) in RNNs w/different activation functions, finding that it stabilized some networks and destabilized others. arxiv.org/html/2608.30...
1325
Reposted by Hsuan-Pei Huang
Daniel Reznik 🎶 @reznikdaniel.bsky.social · 08/09/2026
Now out in Hippocampus!! 📣💥 We examined anatomical inputs to the hippocampal region in six species - the tenrec, rat, marmoset, cat, macaque, and human - and found that hippocampal processing operates on fundamentally different types of information across species! (1/3) dx.doi.org/10.1002/hipo...
dx.doi.org
Phylogeny of Cortical‐Hippocampal Projections Reveals Selective Elimination of Input From Sensory Regions
The study of cross-species differences in the organization of the cerebral cortex received significant attention during the past century. However, the evolutionary principles governing changes in bra...
27424