Romain Brette
LightNews LightNews
romainbrette.bsky.social
Romain Brette
@romainbrette.bsky.social
610 followers 110 following 180 posts
Looking at protists with the eyes of a theoretical neuroscientist.
Looking at brains with the eyes of a protistologist.
(I also like axon initial segments)

Forthcoming book: The Brain, in Theory.

http://romainbrette.fr/
Posts Replies Media Videos
Pinned
romainbrette.bsky.social
Romain Brette @romainbrette.bsky.social · Nov 18
On the publisher's website:
press.princeton.edu/books/paperb...
The Brain, In Theory
Why engineering and computational analogies are poorly suited to the study of biological cognition
press.princeton.edu
romainbrette.bsky.social
Not everything useful is computation. Example: the clock.

press.princeton.edu/books/paperb...
An elementary example is a clock. We can certainly say that a clock implements a computation, namely counting. But this is not what makes it a clock. What makes it a clock is that each tick takes exactly one second, and this is not obtained by computation but from the physical properties of the device. For example, in a traditional clock (figure 4.10), it is the dynamics of the pendulum that sets the tick frequency.
December 31, 2025 at 7:32 AM
Reposted by Romain Brette
wsj.com
AI ran our office vending machine for several weeks. It lost hundreds of dollars, gave away a PlayStation, bought a live fish—and taught us a lot about AI agents, writes Joanna Stern.
We Let AI Run Our Office Vending Machine. It Lost Hundreds of Dollars.
An AI agent ran a snack operation in the WSJ newsroom. It gave away a free PlayStation, ordered a live fish—and taught us lessons about the future of AI.
on.wsj.com
December 18, 2025 at 11:06 AM
Reposted by Romain Brette
rensec.bsky.social
You mean I can't just make up people instead of studying people? www.science.org/content/arti...
AI-generated ‘participants’ can lead social science experiments astray, study finds
Data produced by “silicon samples” depends on researchers’ exact choice of models, prompts, and settings
www.science.org
December 18, 2025 at 8:14 AM
Reposted by Romain Brette
jedbrown.org
I think it's also important to recognize that LLM "summaries" are not epistemically grounded in the ideas of the source material. When they're correct, it's more because *other* writing (in the training corpus) contains similar language. That makes it especially bad when applied to novel results.
December 17, 2025 at 6:18 AM
romainbrette.bsky.social
The belief that an LLM can help you write science reveals such magical thinking about language (and LLMs). If your writing is unclear, an LLM won't find its meaning hidden in the words. Only you can explain what you mean; the LLM does not read your mind. It completes with what *others* have written.
December 17, 2025 at 8:27 AM
romainbrette.bsky.social
Wow. Majority uses AI for peer reviewing, and authoring. But not everyone, so the conclusion of the publisher (Frontiers, same group as Nature): this leaves "major potential untapped". Same poll: 90% have concerns AI tools are misused by researchers, and by publishers. So: there's a "trust gap".
December 17, 2025 at 8:10 AM
Reposted by Romain Brette
babrahamse.bsky.social
My favorite euphemism is "responsible use of AI" . if it were as good a thing as everyone says it it's, you wouldn't need to be super responsible to use it
December 16, 2025 at 11:04 PM
romainbrette.bsky.social
GenAI is such a harmful technology.
andrew.heiss.phd Andrew Heiss @andrew.heiss.phd · 22d
Grading and googling hallucinated citations, as one does nowadays, and now that LLMs have been around for a while, I've discovered new horrors: hallucinated journals are now appearing in Google Scholar with dozens of citations bc so many people are citing these fake things
December 16, 2025 at 11:26 AM
romainbrette.bsky.social
The "what else" argument is really something quite bizarre. The brain is a computer, because what else, a magic box?
Like a cat must be a horse, because what else, an elephant?
December 11, 2025 at 12:22 PM
romainbrette.bsky.social
The philosophical misconception behind the LLM cult (or why LLMs will always bullshit)

romainbrette.fr/the-philosop...
The philosophical misconception behind the LLM cult (or why LLMs will always bullshit)
There is this idea that if a large language model (LLM) is trained on a large corpus of text, then it knows whatever knowledge is in that corpus. Improving the performance is then essentially a mat…
romainbrette.fr
December 9, 2025 at 8:31 AM
romainbrette.bsky.social
Look, the universal praise of ERC among scientists is based on a fallacy: it's great to have an ERC grant, so we should have more. No, we shouldn't. Money doesn't just pop up. Big individual grants means concentrating money on a few, and so reducing the perimeter of science.
tank-silvia.bsky.social Silvia Manrique @tank-silvia.bsky.social · Dec 3
“A program with a 0.5% success rate is not elite. It’s wasteful. Thousands of highly trained scientists will sink dozens of hours into writing proposals with vanishing odds of success. This is not just inefficient — it’s disrespectful of scientific labour.” Very well said @kamounlab.bsky.social!👏🏼👏🏼👏🏼
kamounlab.bsky.social Sophien Kamoun @kamounlab.bsky.social · Dec 3
I just published: ERC-Plus: jackpot science or missed chance to fix academia?

My reflections on ERC-Plus, Europe’s newest ultra-competitive research grant and what it tells us about the academic culture we’re building.

medium.com/p/erc-plus-j...
December 4, 2025 at 8:46 AM
romainbrette.bsky.social
Just wondering: does anyone find the "Significance" paragraph remotely useful? (e.g. in PNAS papers)
December 4, 2025 at 7:40 AM
romainbrette.bsky.social
Exactly. The relation between theory and experiment is an interaction, not theory->experiment (preregistration) or experiment->theory (exploration).
devezer.bsky.social Berna Devezer @devezer.bsky.social · Dec 3
Penicillin was discovered bec Fleming followed up on the initial observation by designing studies to understand the mechanism producing it. The confirmatory (preregistered) tests vs. exploratory (non-preregistered) results distinction that is being normalized is exacerbating this misconception.
December 3, 2025 at 7:34 PM
romainbrette.bsky.social
No, chatbots are not scientists
romainbrette.fr/no-chatbots-...
No, chatbots are not scientists
I can’t believe I need to write this, but apparently, I do. Many journals and preprint sites are now overwhelmed with chatbot-generated submissions. This is bad, but at least it gets characterized …
romainbrette.fr
December 2, 2025 at 10:33 AM
romainbrette.bsky.social
I have a PhD project open for application about "Habituation of Paramecium, the “swimming neuron”" on the DIM C-Brains programme, for a PhD in Paris. Open for students with a Master outside France. Get in touch if interested!
dim-cbrains.fr/en/phd-progr...
DIM C-BRAINS
Cognition and Brain Revolutions: Artificial Intelligence, Neurogenomics, Society
dim-cbrains.fr
November 28, 2025 at 11:38 AM
romainbrette.bsky.social
En toute décontraction, on s'apprête en France à recenser les opinions politiques des profs de fac.
November 25, 2025 at 10:37 AM
romainbrette.bsky.social
About "PhD-level AI". This is such a misunderstanding of scientific work. What a PhD is about is not technical skills or encyclopedic knowledge. It's about learning to prove (or disprove) claims. At core, it's the development of an ethical attitude to knowledge. Exactly what an LLM cannot provide.
November 24, 2025 at 3:38 PM
romainbrette.bsky.social
Remember obscurantism is a political agenda, not a failure of scientists to do public outreach.
luckytran.com Dr. Lucky Tran @luckytran.com · Nov 19
I really hate it when scientists keep saying that “we need to rebuild trust in science,” because it implies that scientists are to blame for the mistrust rather than the millions of dollars of dark money that have funded political attacks on science in order to advance a far right agenda.
November 20, 2025 at 8:53 AM
romainbrette.bsky.social
Commercial publishers poison academia. They must be banned.
hansonmark.bsky.social Mark A. Hanson @hansonmark.bsky.social · Nov 11
We wrote the Strain on scientific publishing to highlight the problems of time & trust. With a fantastic group of co-authors, we present The Drain of Scientific Publishing:

a 🧵 1/n

Drain: arxiv.org/abs/2511.04820
Strain: direct.mit.edu/qss/article/...
Oligopoly: direct.mit.edu/qss/article/...
A table showing profit margins of major publishers. A snippet of text related to this table is below. 1. The four-fold drain 1.1 Money Currently, academic publishing is dominated by profit-oriented, multinational companies for whom scientific knowledge is a commodity to be sold back to the academic community who created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis, which collectively generated over US$7.1 billion in revenue from journal publishing in 2024 alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit margins have always been over 30% in the last five years, and for the largest publisher (Elsevier) always over 37%. Against many comparators, across many sectors, scientific publishing is one of the most consistently profitable industries (Table S1). These financial arrangements make a substantial difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor & Francis revenues were generated in North America, meaning that North American researchers were charged over US$2.27 billion by just two for-profit publishers. The Canadian research councils and the US National Science Foundation were allocated US$9.3 billion in that year. A figure detailing the drain on researcher time. 1. The four-fold drain 1.2 Time The number of papers published each year is growing faster than the scientific workforce, with the number of papers per researcher almost doubling between 1996 and 2022 (Figure 1A). This reflects the fact that publishers’ commercial desire to publish (sell) more material has aligned well with the competitive prestige culture in which publications help secure jobs, grants, promotions, and awards. To the extent that this growth is driven by a pressure for profit, rather than scholarly imperatives, it distorts the way researchers spend their time. The publishing system depends on unpaid reviewer labour, estimated to be over 130 million unpaid hours annually in 2020 alone (9). Researchers have complained about the demands of peer-review for decades, but the scale of the problem is now worse, with editors reporting widespread difficulties recruiting reviewers. The growth in publications involves not only the authors’ time, but that of academic editors and reviewers who are dealing with so many review demands. Even more seriously, the imperative to produce ever more articles reshapes the nature of scientific inquiry. Evidence across multiple fields shows that more papers result in ‘ossification’, not new ideas (10). It may seem paradoxical that more papers can slow progress until one considers how it affects researchers’ time. While rewards remain tied to volume, prestige, and impact of publications, researchers will be nudged away from riskier, local, interdisciplinary, and long-term work. The result is a treadmill of constant activity with limited progress whereas core scholarly practices – such as reading, reflecting and engaging with others’ contributions – is de-prioritized. What looks like productivity often masks intellectual exhaustion built on a demoralizing, narrowing scientific vision. A table of profit margins across industries. The section of text related to this table is below: 1. The four-fold drain 1.1 Money Currently, academic publishing is dominated by profit-oriented, multinational companies for whom scientific knowledge is a commodity to be sold back to the academic community who created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis, which collectively generated over US$7.1 billion in revenue from journal publishing in 2024 alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit margins have always been over 30% in the last five years, and for the largest publisher (Elsevier) always over 37%. Against many comparators, across many sectors, scientific publishing is one of the most consistently profitable industries (Table S1). These financial arrangements make a substantial difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor & Francis revenues were generated in North America, meaning that North American researchers were charged over US$2.27 billion by just two for-profit publishers. The Canadian research councils and the US National Science Foundation were allocated US$9.3 billion in that year. The costs of inaction are plain: wasted public funds, lost researcher time, compromised scientific integrity and eroded public trust. Today, the system rewards commercial publishers first, and science second. Without bold action from the funders we risk continuing to pour resources into a system that prioritizes profit over the advancement of scientific knowledge.
November 18, 2025 at 7:14 PM
romainbrette.bsky.social
My book "The Brain, in Theory" on the publisher's website (out in April 2026):
press.princeton.edu/books/paperb...
The Brain, In Theory
Why engineering and computational analogies are poorly suited to the study of biological cognition
press.princeton.edu
November 18, 2025 at 8:25 AM
romainbrette.bsky.social
Wow I had no idea this existed. Bacteria that build centimeter-scale electrical wires to breathe.
November 18, 2025 at 8:23 AM
romainbrette.bsky.social
The Nobel prize in economics. The fact that it's not a "true" Nobel is anecdotical. What matters is that a prize (Nobel or other) is given by people, and not by an omniscient God of Science. So it's a social recognition within an academic field, not a badge of Truth.
1/3
November 14, 2025 at 8:09 AM
Reposted by Romain Brette
cianodonnell.bsky.social
gave a short lecture this morning on principles of computational modelling, always try to stress the point made by @romainbrette.bsky.social that adding details to a model does not automatically make it more realistic.

The wooden airplane model has more 'details' but only the paper model can fly
Slide with title 'Details vs realism' showing a picture of a wooden model airplane on left and a paper airplane on right.
August 25, 2025 at 10:20 AM
Reposted by Romain Brette
irisvanrooij.bsky.social
AI slop and the destruction of knowledge irisvanrooijcogsci.com/2025/08/12/a...
AI slop and the destruction of knowledge
This week I was looking for info on what cognitive scientists mean when they speak of ‘domain-general’ cognition. I was curious, because the nuances are relevant for something I am researching at t…
irisvanrooijcogsci.com
August 12, 2025 at 10:12 PM
romainbrette.bsky.social
Why large language models are not conscious.
romainbrette.fr/notes-on-con...
Notes on consciousness. (XI) Why large language models are not conscious
The current discussions about AI are plagued with anthropomorphism. Of course, the name “artificial intelligence” has probably something to do with the matter. A conscious artificial intelligence m…
romainbrette.fr
August 20, 2025 at 10:18 AM