Konrad Kording
@kordinglab.bsky.social
@Penn Prof, deep learning, brains, #causality, rigor, http://neuromatch.io, Transdisciplinary optimist, Dad, Loves outdoors, 🦖 , c4r.io
Planyourscience.com now applauds the good changes you make - and pushes back on the less good one.
February 9, 2026 at 8:43 PM
Planyourscience.com now applauds the good changes you make - and pushes back on the less good one.
neuroAI comparisons of ANNs to brains do have a range of problems. Even more than I had realized. And I was worried before: www.biorxiv.org/content/10.1...
www.biorxiv.org
February 9, 2026 at 2:13 PM
neuroAI comparisons of ANNs to brains do have a range of problems. Even more than I had realized. And I was worried before: www.biorxiv.org/content/10.1...
Is Bluesky dying? For me it used to drive about as much engagement for maths science content as twitter. Now twitter is up by a factor of five.
February 9, 2026 at 1:24 PM
Is Bluesky dying? For me it used to drive about as much engagement for maths science content as twitter. Now twitter is up by a factor of five.
Well-predicting machine learning in no way means that you can understand how the world works.
open.substack.com/pub/kording/...
open.substack.com/pub/kording/...
Forward vs Inverse problems: why high performance machine learning usually means little about how the world works
Understanding causality from machine learning is unfortunately usually impossible; life sciences take note
open.substack.com
February 7, 2026 at 6:31 PM
Well-predicting machine learning in no way means that you can understand how the world works.
open.substack.com/pub/kording/...
open.substack.com/pub/kording/...
Reposted by Konrad Kording
Why don’t neural networks learn all at once, but instead progress from simple to complex solutions? And what does “simple” even mean across different neural network architectures?
Sharing our new paper @iclr_conf led by Yedi Zhang with Peter Latham
arxiv.org/abs/2512.20607
Sharing our new paper @iclr_conf led by Yedi Zhang with Peter Latham
arxiv.org/abs/2512.20607
February 3, 2026 at 4:19 PM
Why don’t neural networks learn all at once, but instead progress from simple to complex solutions? And what does “simple” even mean across different neural network architectures?
Sharing our new paper @iclr_conf led by Yedi Zhang with Peter Latham
arxiv.org/abs/2512.20607
Sharing our new paper @iclr_conf led by Yedi Zhang with Peter Latham
arxiv.org/abs/2512.20607
Reposted by Konrad Kording
Reproducibility and open science are great, but they don't necessarily equate to rigor. You can perfectly share a study and still draw weak conclusions. True rigor lives in the questions we ask, the designs we choose, and the inferences we make.
February 2, 2026 at 5:15 PM
Reproducibility and open science are great, but they don't necessarily equate to rigor. You can perfectly share a study and still draw weak conclusions. True rigor lives in the questions we ask, the designs we choose, and the inferences we make.
Without looking it up, what kind of data could realistically support a statement like "This reveals that the human brain may be simpler than previously thought, with potential implications for human cognition and disease."
February 2, 2026 at 1:37 AM
Without looking it up, what kind of data could realistically support a statement like "This reveals that the human brain may be simpler than previously thought, with potential implications for human cognition and disease."
We should use AI not to remove friction but we should use AI to *add* friction. The problem is not a lack of good ideas. The problem is that bad ideas are everywhere. Adding AI carelessly will just add more bad ideas that sound like good ideas. As if we didn't already have enough of that.
“The idea is to put ChatGPT front and center inside software that scientists use to write up their work in much the same way that chatbots are now embedded into popular programming editors.
It’s vibe coding, but for science.”
It’s vibe coding, but for science.”
OpenAI’s latest product lets you vibe code science
Prism is a ChatGPT-powered text editor that automates much of the work involved in writing scientific papers.
www.technologyreview.com
January 29, 2026 at 1:15 AM
We should use AI not to remove friction but we should use AI to *add* friction. The problem is not a lack of good ideas. The problem is that bad ideas are everywhere. Adding AI carelessly will just add more bad ideas that sound like good ideas. As if we didn't already have enough of that.
Reposted by Konrad Kording
I serve on a review committee for an accelerator that has this problem too. Applicants are submitting ~a dozen pages of AI content they've often not fully read.
The solution in my opinion, is to SHORTEN submission requirements, and I think the solution for scientific publishing is similar.
1/
The solution in my opinion, is to SHORTEN submission requirements, and I think the solution for scientific publishing is similar.
1/
What a fucking disaster.
Not only are OpenAI and the likes unleashing this shit on the world, they are deliberately leaning into the harms that their business model causes with products like Prism.
Not only are OpenAI and the likes unleashing this shit on the world, they are deliberately leaning into the harms that their business model causes with products like Prism.
Research Notes of the AAS in particular, which was set up to handle short, moderated contributions especially from students, is getting swamped. Often the authors clearly haven’t read what they’ve submitting, (Descriptions of figures that don’t exist or don’t show what they purport to)
January 28, 2026 at 11:43 PM
I serve on a review committee for an accelerator that has this problem too. Applicants are submitting ~a dozen pages of AI content they've often not fully read.
The solution in my opinion, is to SHORTEN submission requirements, and I think the solution for scientific publishing is similar.
1/
The solution in my opinion, is to SHORTEN submission requirements, and I think the solution for scientific publishing is similar.
1/
Reposted by Konrad Kording
This tool is extremely helpful and I think all scientists could benefit from it. Good at all stages of the research process, especially the very beginning.
It's also a great of example of how we *should* be building AI tools for science. Not to replace scientific thinking, but to hone it.
It's also a great of example of how we *should* be building AI tools for science. Not to replace scientific thinking, but to hone it.
First week where planyourscience.com had more than 1000 organic users. I know this is nothing by software standards but it is quite something by academic standards. And to everyone using it - tell me what could/should be better. And use it to export a first draft ;)
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.
planyourscience.com
January 28, 2026 at 6:16 PM
This tool is extremely helpful and I think all scientists could benefit from it. Good at all stages of the research process, especially the very beginning.
It's also a great of example of how we *should* be building AI tools for science. Not to replace scientific thinking, but to hone it.
It's also a great of example of how we *should* be building AI tools for science. Not to replace scientific thinking, but to hone it.
First week where planyourscience.com had more than 1000 organic users. I know this is nothing by software standards but it is quite something by academic standards. And to everyone using it - tell me what could/should be better. And use it to export a first draft ;)
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.
planyourscience.com
January 28, 2026 at 3:36 PM
First week where planyourscience.com had more than 1000 organic users. I know this is nothing by software standards but it is quite something by academic standards. And to everyone using it - tell me what could/should be better. And use it to export a first draft ;)
Isfahan AI event. Sounded great. Diagnosing MS is useful across the world. journals.lww.com/jmss/fulltex...
Isfahan is one of the cities where crackdown is extreme. Wishing the best for the scientists there. www.bbc.com/news/article...
Isfahan is one of the cities where crackdown is extreme. Wishing the best for the scientists there. www.bbc.com/news/article...
journals.lww.com
January 28, 2026 at 2:03 PM
Isfahan AI event. Sounded great. Diagnosing MS is useful across the world. journals.lww.com/jmss/fulltex...
Isfahan is one of the cities where crackdown is extreme. Wishing the best for the scientists there. www.bbc.com/news/article...
Isfahan is one of the cities where crackdown is extreme. Wishing the best for the scientists there. www.bbc.com/news/article...
Why on earth does tDCS do anything given how low the fields are? Here is a nice paper: nature.com/articles/s41... -- also from one of the labs whose webpage is in the dark due to their Tehran location.
Effects of transcranial electrical stimulation of the cerebellum, parietal cortex, anterior cingulate, and motor cortex on postural adaptation - Scientific Reports
Scientific Reports - Effects of transcranial electrical stimulation of the cerebellum, parietal cortex, anterior cingulate, and motor cortex on postural adaptation
nature.com
January 27, 2026 at 9:42 PM
Why on earth does tDCS do anything given how low the fields are? Here is a nice paper: nature.com/articles/s41... -- also from one of the labs whose webpage is in the dark due to their Tehran location.
Psychophysics to understand the visual cortex: frontiersin.org/journals/sys... - another paper by a great group from Tehran whose webpage is gone now.
Frontiers | Early Visual Processing of Feature Saliency Tasks: A Review of Psychophysical Experiments
Visual attention has an undeniable impact on our perception. Human brain is unable to process all of the information it faces with in a scene so, the most sa...
frontiersin.org
January 26, 2026 at 5:53 PM
Psychophysics to understand the visual cortex: frontiersin.org/journals/sys... - another paper by a great group from Tehran whose webpage is gone now.
This Gershman and Ullman paper is just so cool in showing that humans really can't not view correlation without assuming causality. Summary of my quixotic life in neuroscience: gershmanlab.com/pubs/Gershma...
gershmanlab.com
January 22, 2026 at 2:12 AM
This Gershman and Ullman paper is just so cool in showing that humans really can't not view correlation without assuming causality. Summary of my quixotic life in neuroscience: gershmanlab.com/pubs/Gershma...
As we were celebrating psychophysics the other day, here is a fun example in the multi-source auditory domain. So beautiful. So clean. Bayes all the way. journals.plos.org/ploscompbiol...
Perceptual clustering in auditory streaming
Author summary Perceiving the world requires humans to organize perceptual stimuli according to the likely sources that generated them, requiring inference about these sources and their relationship w...
journals.plos.org
January 22, 2026 at 1:12 AM
As we were celebrating psychophysics the other day, here is a fun example in the multi-source auditory domain. So beautiful. So clean. Bayes all the way. journals.plos.org/ploscompbiol...
I love this so much. After pushback on his recent "Medicine is the only field that reaches 6 sigma" with "my field, psychophysics is so awesome" he posted this. Hurray all Psychophysicists. LETS CELEBRATE PSYCHOPHYSICS. An island of large effects is us!
Psychophysics is a human-facing science with interventions arguably more robust than medicine.
1000 Hurts
Psychophysics is a human-facing science with interventions arguably more robust than medicine.
www.argmin.net
January 15, 2026 at 4:14 PM
I love this so much. After pushback on his recent "Medicine is the only field that reaches 6 sigma" with "my field, psychophysics is so awesome" he posted this. Hurray all Psychophysicists. LETS CELEBRATE PSYCHOPHYSICS. An island of large effects is us!
Reposted by Konrad Kording
@kordinglab.bsky.social and I ran a summer school last year to help young profs (<5 yrs) in systems/comp neuro thrive.
compneurosci.com/Neuro4Pros/i...
It was great! Now we want to know if you'd be interested in participating if we did this again this year?
Let us know!
compneurosci.com/Neuro4Pros/i...
It was great! Now we want to know if you'd be interested in participating if we did this again this year?
Let us know!
Neuro4Pros summer school
Neuroscience Leadership Training
compneurosci.com
January 15, 2026 at 3:51 PM
@kordinglab.bsky.social and I ran a summer school last year to help young profs (<5 yrs) in systems/comp neuro thrive.
compneurosci.com/Neuro4Pros/i...
It was great! Now we want to know if you'd be interested in participating if we did this again this year?
Let us know!
compneurosci.com/Neuro4Pros/i...
It was great! Now we want to know if you'd be interested in participating if we did this again this year?
Let us know!
Another cool paper from Iran: nature.com/articles/s41... Prenatal stress makes mouse offspring more anxious, hyperactive, forgetful, and more drawn to morphine while laterodorsal tegmentum cholinergic neurons become hyperexcitable. From Kerman, a city with reported violent crackdowns on protests.
The hyperexcitability of laterodorsal tegmentum cholinergic neurons accompanies adverse behavioral and cognitive outcomes of prenatal stress - Scientific Reports
Scientific Reports - The hyperexcitability of laterodorsal tegmentum cholinergic neurons accompanies adverse behavioral and cognitive outcomes of prenatal stress
nature.com
January 15, 2026 at 2:18 PM
Another cool paper from Iran: nature.com/articles/s41... Prenatal stress makes mouse offspring more anxious, hyperactive, forgetful, and more drawn to morphine while laterodorsal tegmentum cholinergic neurons become hyperexcitable. From Kerman, a city with reported violent crackdowns on protests.
An open source rodent forelimb paradigm cell.com/heliyon/full... Cool work from Tehran!
An open-source automated rodent paradigm for forelimb movement in motor control, executive functions, and learning studies
Reaching tasks are frequently employed in studies investigating cognitive functions,
motor control, and brain-computer interfaces. The center-out task, a subtype of reaching
movement, is widely utiliz...
cell.com
January 14, 2026 at 2:30 PM
An open source rodent forelimb paradigm cell.com/heliyon/full... Cool work from Tehran!
Nice writeup. I mostly agree. However in psychophysics effect sizes (at least main effects) tend to be crazy large. Like too big to even put it into numbers large. Eg look at this nature.com/articles/nat... the main effects is slope greater zero.
Client Challenge
nature.com
January 12, 2026 at 5:05 PM
Nice writeup. I mostly agree. However in psychophysics effect sizes (at least main effects) tend to be crazy large. Like too big to even put it into numbers large. Eg look at this nature.com/articles/nat... the main effects is slope greater zero.
Into neuromorphic? Consider the excellent nengo summer school www.nengo.ai/summer-school/
Nengo Summer School
Nengo is a graphical and scripting based Python package for simulating large-scale neural networks.
www.nengo.ai
January 12, 2026 at 4:06 PM
Into neuromorphic? Consider the excellent nengo summer school www.nengo.ai/summer-school/
Our paper that argues that the physical domain with its slow scaling should eventually dominate the fast scaling intelligence domain is now live on ssrn papers.ssrn.com/sol3/papers....
papers.ssrn.com
January 12, 2026 at 2:07 PM
Our paper that argues that the physical domain with its slow scaling should eventually dominate the fast scaling intelligence domain is now live on ssrn papers.ssrn.com/sol3/papers....
Why is it real fun in a mosh pit? It is hard to see normative nor mechanistic reasons.
January 11, 2026 at 3:41 AM
Why is it real fun in a mosh pit? It is hard to see normative nor mechanistic reasons.
This is very cool. Much faster DSA. Although we still don't know what DSA findings actually mean about brains.
Wanna compare dynamics across neural data, RNNs, or dynamical systems? We got a fast and furious method🏎️
The 1st preprint of my PhD 🥳 fast dynamical similarity analysis (fastDSA):
📜: arxiv.org/abs/2511.22828
💻: github.com/CMC-lab/fast...
I’ll be @cosynemeeting.bsky.social - happy to chat 😉
The 1st preprint of my PhD 🥳 fast dynamical similarity analysis (fastDSA):
📜: arxiv.org/abs/2511.22828
💻: github.com/CMC-lab/fast...
I’ll be @cosynemeeting.bsky.social - happy to chat 😉
January 8, 2026 at 5:17 PM
This is very cool. Much faster DSA. Although we still don't know what DSA findings actually mean about brains.