Saurabh

GitHub - esennesh/numpyro_template: Numpyro probabilistic programming projects made easy Numpyro probabilistic programming projects made easy - esennesh/numpyro_template

Apropos of absolutely nothing, I've been slowly but surely recreating half of PyTorch Lightning in Jax with Numpyro to use as a way of organizing codebases. github.com/esennesh/num...

🧪🧬🌏🦠🔭

‘A bombshell’: doubt cast on discovery of microplastics throughout human body Exclusive: Some scientists say many detections are most likely error, with one high-profile study called a ‘joke’

🎓⚕️🧪🔬📰
The way we do science today is very vulnerable to clickbait and sensationalist headlines

Remember the microplastics in your brain, in male testes and in your bloodstream?

It didn't make sense with what we knew, but it made great entertainment. #BadScience
www.theguardian.com/environment/...

"Robustness" Is Not Replication We must be clear about what we are doing

"Our job is to find the correct choice of specifications, not to see how many changes a result is robust to. A result can be completely correct, and yet not be robust to even small changes; conversely, a result can be robust to many different changes, and yet be wrong."

By @captgouda24.bsky.social

A Population Vector Model of Visual Working Memory for Real-World Scenes https://www.biorxiv.org/content/10.64898/2026.01.23.701256v1

Modeling non-dual awareness via constraint closure: a reinterpretation of groundlessness Abstract. Non-dual awareness (NDA) refers to a shift in consciousness in which the usual distinction between subject and object dissolves, and experience i

Modeling non-dual awareness via constraint closure: a reinterpretation of groundlessness url: academic.oup.com/nc/article/2...

Whether you love causal graphs, hate causal graphs, or just want to find out what the heck a causal graph is, you’ll want to watch my talk on how to *actually* make causal graphs.

Sign up at the link below👇🏼

How can teachers learn to teach effectively and equitably? - Journal of Mathematics Teacher Education Journal of Mathematics Teacher Education -

Grateful for this beautiful review of our book, "Teacher Learning of Ambitious and Equitable Mathematics Teaching: A Sociocultural Approach"

link.springer.com/article/10.1...

On Misunderstanding the Poverty of the Stimulus Argument The Poverty of the Stimulus Argument is not (really) an argument, and usage-based approaches reliant on LLMs to refute it are weakened.

I wrote up some thoughts on why the Poverty of the Stimulus Argument - a mainstay in linguistics and other areas of CogSci - is not (really) an argument.

I also argue, controversially, that usage-based reliance on LLMs to refute the POSA are weakened. 🧵

vincentcarchidi.substack.com/p/on-misunde...

This is so, so good

Industry Influence in High-Profile Social Media Research To what extent is social media research independent from industry influence? Leveraging openly available data, we show that half of the research published in top journals has disclosable ties to indus...

Ever wonder what proportion of high profile social media research is tied to the tech industry?

New from me, @cailinmeister.bsky.social, @jevinwest.bsky.social and @carlbergstrom.com.

Thread tomorrow.

arxiv.org/abs/2601.11507

As an editor of a scientific journal, this essay feels very familiar: it’s about how AI coding tools are making it harder to maintain quality in open source software. The solution cannot be to just use AI agents to evaluate too, right? What historical parallels can we learn from?

Science journals retract 500 papers a month. This is why it matters A small team of volunteers is tracking thousands of falsified studies, including cases of bribery, fraud and plagiarism

Science journals retract 500 papers a month. This is why it matters

Things, to be and the computer metaphor From birth, we learn to communicate through language—but language misleads us. It convinces us that we truly know what things are, that we comprehend the abstract essence of a thing. This illusion sha...

The Thing Trap: How Language Locked Us Into the Wrong Model of Mind

ocrampal.com/things-to-be...

#philosophy #science #psychology #AI #intelligence #physics #biology #philmind #philsci #philsky #philpsy #neurosky #neuroskyence

Reliability and validity of multi-band multi-echo fMRI https://www.biorxiv.org/content/10.64898/2026.01.16.699838v1

Investigating the methodological foundation of lesion network mapping - Nature Neuroscience The lesion network mapping method links diverse brain lesions to similar functional brain networks, reflecting general brain organization rather than disorder-specific circuits.

Wake up, babe, heir to the dead fish fMRI study just dropped. www.nature.com/articles/s41... thx @sifu.tweety.fish @jimthommo.bsky.social for the tip. Whooooof.

Methodological misalignment in cognitive measurement

Context in decision making

Context in brain

On Jan 15, 2026 I attended the webinar “Methodological Misalignment in Cognitive Measurement”, by Olha Sobetska.

It offered a sharp reminder that many so-called cognitive errors are really measurement problems.

Thank you for organizing @johanneskleiner.bsky.social

#conscio #cognition

ECogS 2026 What does embodied cognitive science have to say about recent developments in AI, and what might these developments reveal about the nature and limits of embodied cognition? Under the theme “From Embo...

ECogS 2026: International Conference on Embodied Cognitive Science
Nov 9–13, 2026 | OIST, Japan

Theme: “From Embodied AI to LLMs and Back”

Details forthcoming: www.oist.jp/conference/e...

#embodiedcognition #AI #conference

𝗪𝗵𝗮𝘁 𝗮𝗿𝗲 𝘄𝗲 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴?
That I don't know but D Brooks discusses some ideas of my book, The Entangled Brain: '"neuronal ensembles distributed across multiple brain regions,” which, like a murmuration of starlings, “forms a single pattern from the collective behavior."'
www.nytimes.com/2026/01/16/o...

It won't actually exist for another month or so, but because it now 'exists' on amazon, I'll humbly observe that, after working through this book, your student/trainee would be able to read and understand all but two or three papers in this week's J. Neurosci. Check it out:

Congrats to @markbrandonlab.bsky.social and his lab for this exciting discovery about the hippocampus and reward:

Those really big effect sizes are one target for establishing what’s-what in particle physics and should be for us too. And then, of course, there’s the importance of understanding how human behavior & brains work /1
arxiv.org/html/2411.00...

A decade of power posing: where do we stand? | BPS Tom Loncar examines power posing’s 10-year journey to date, and the challenges that still remain.

(Cheeky ref: www.bps.org.uk/psychologist...)

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!

Does morphology support novel word learning through statistical learning? Most novel words that speakers learn are morphologically complex (e.g. columnist, whistleblower). Rather than hindering learning, such complexity might provide facilitation. Affixes (e.g. pre-, -ne...

Does morphology support novel word learning through statistical learning?

www.tandfonline.com/doi/full/10....

Celebrating the Second Year of Linux Man-Pages Maintenance Sponsorship - Linux.com Sustaining a Core Part of the Linux Ecosystem The Linux Foundation has announced a second year of sponsorship for the ongoing maintenance of the Linux manual pages (man-pages) project, led by Alejandr...

At Google, we know we are built on an open ecosystem. Supporting maintainers as critical infrastructure is vital, so we're celebrating a second year of sponsoring Alejandro Colomar’s work on the Linux man-pages. 🛠️

www.linux.com/news/celebra...

Have the Linux man-pages helped your workflow?

On the referential capacity of language models: An internalist rejoinder to Mandelkern & Linzen Abstract. Mandelkern and Linzen (2024) argue that words generated by language models (LMs) are linked to causal histories of use within human linguistic communities, and ultimately to their referents....

New paper in Computational Linguistics with @giosuebaggio.bsky.social.

Responding to Matt Mandelkern & @tallinzen.bsky.social's arguments about the "referential" capacities of LLMs by leveraging some internalist critiques.

direct.mit.edu/coli/article...

On Smallest Effect Size of Interest: Its Development and a Simplified Procedure: https://osf.io/g2yv5

An Audit of Social Science Survey Experiments Abstract. Survey experiments have become a popular methodology for causal inference across the social sciences. We study the efficacy of survey experiment

This is a belated post about our paper in @poqjournal.bsky.social.

We analyzed 100 survey experiments fielded by TESS (tessexperiments.org), using only information from the proposals to identify intended hypotheses.

Here are some of the things we learned:

If you're interested in a short behind-the-scenes of the retraction of a mega-metascience paper—the first of its kind (the retraction, not the paper)—this might be for you! Had to gloss over so many details to observe allotted time but happy to share an extended version of my slide deck with links.