Jérémie Beucler
LightNews LightNews
banner
jeremiebeucler.bsky.social
Jérémie Beucler
@jeremiebeucler.bsky.social
150 followers 500 following 25 posts
PhD student with Wim de Neys & Lucie Charles at LaPsyDE; MSc in Cog Sciences at ENS - interested in reasoning & metacognition

https://jeremie-beucler.github.io/
Posts Replies Media Videos
Pinned
jeremiebeucler.bsky.social
Jérémie Beucler @jeremiebeucler.bsky.social · 25d
1/10

🚨 New preprint: Using Large Language Models to Estimate Belief Strength in Reasoning 🚨

When asked: "There are 995 politicians and 5 nurses. Person 'L' is kind. Is Person 'L' more likely to be a politician or a nurse?", most people will answer "nurse", neglecting the base-rate info.

A 🧵👇
Abstract Accurately quantifying belief strength in heuristics-and-biases tasks is crucial yet methodologically challenging. In this paper, we introduce an automated method leveraging large language models (LLMs) to systematically measure and manipulate belief strength. We specifically tested this method in the widely used “lawyer-engineer” base-rate neglect task, in which stereotypical descriptions (e.g., someone enjoying mathematical puzzles) conflict with normative base-rate information (e.g., engineers represent a very small percentage of the sample). Using this approach, we created an open-access database containing over 100,000 unique items systematically varying in stereotype-driven belief strength. Validation studies demonstrate that our LLM-derived belief strength measure correlates strongly with human typicality ratings and robustly predicts human choices in a base-rate neglect task. Additionally, our method revealed substantial and previously unnoticed variability in stereotype-driven belief strength in popular base-rate items from existing research, underlining the need to control for this in future studies. We further highlight methodological improvements achievable by refining the LLM prompt, as well as ways to enhance cross-cultural validity. The database presented here serves as a powerful resource for researchers, facilitating rigorous, replicable, and theoretically precise experimental designs, as well as enabling advancements in cognitive and computational modeling of reasoning. To support its use, we provide the R package baserater, which allows researchers to access the database to apply or adapt the method to their own research.
Reposted by Jérémie Beucler
kobedesender.bsky.social
Brace yourself: neural and computational insights into
the experience of mental effort! Now out in @cerebralcortex.bsky.social Led by Gaia Corlazzoli.

Paper: desenderlab.com/wp-content/u.... Thread ↓↓↓
November 4, 2025 at 12:37 PM
Reposted by Jérémie Beucler
adigitaltanay.bsky.social
My first paper, out in PsychReview!!

Along with @orbenamy.bsky.social, Nik & @jaeggiadrian.bsky.social, @realadamhunt.bsky.social & I revisit an old theoretical question using concepts from evo psychiatry and anthro:

Why do mixed associations exist b/w social media & mental health?

A 🧵
November 4, 2025 at 8:04 AM
Reposted by Jérémie Beucler
lauragwilliams.bsky.social
Delighted to share our new paper, now out in PNAS! www.pnas.org/doi/10.1073/...

"Hierarchical dynamic coding coordinates speech comprehension in the brain"

with dream team @alecmarantz.bsky.social, @davidpoeppel.bsky.social, @jeanremiking.bsky.social

Summary 👇

1/8
PNAS
Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...
www.pnas.org
October 22, 2025 at 5:21 AM
jeremiebeucler.bsky.social
1/10

🚨 New preprint: Using Large Language Models to Estimate Belief Strength in Reasoning 🚨

When asked: "There are 995 politicians and 5 nurses. Person 'L' is kind. Is Person 'L' more likely to be a politician or a nurse?", most people will answer "nurse", neglecting the base-rate info.

A 🧵👇
Abstract Accurately quantifying belief strength in heuristics-and-biases tasks is crucial yet methodologically challenging. In this paper, we introduce an automated method leveraging large language models (LLMs) to systematically measure and manipulate belief strength. We specifically tested this method in the widely used “lawyer-engineer” base-rate neglect task, in which stereotypical descriptions (e.g., someone enjoying mathematical puzzles) conflict with normative base-rate information (e.g., engineers represent a very small percentage of the sample). Using this approach, we created an open-access database containing over 100,000 unique items systematically varying in stereotype-driven belief strength. Validation studies demonstrate that our LLM-derived belief strength measure correlates strongly with human typicality ratings and robustly predicts human choices in a base-rate neglect task. Additionally, our method revealed substantial and previously unnoticed variability in stereotype-driven belief strength in popular base-rate items from existing research, underlining the need to control for this in future studies. We further highlight methodological improvements achievable by refining the LLM prompt, as well as ways to enhance cross-cultural validity. The database presented here serves as a powerful resource for researchers, facilitating rigorous, replicable, and theoretically precise experimental designs, as well as enabling advancements in cognitive and computational modeling of reasoning. To support its use, we provide the R package baserater, which allows researchers to access the database to apply or adapt the method to their own research.
October 16, 2025 at 4:17 PM
Reposted by Jérémie Beucler
sophiebavard.bsky.social
🚨 New preprint! 🚨
Very happy to share our latest work on metacognition with M. Rouault, A. McWilliams, F. Chartier, @kndiaye.bsky.social and @smfleming.bsky.social where we identify contributors to self-performance estimates across memory and perception domains 👇
osf.io/preprints/ps...
October 13, 2025 at 8:08 AM
Reposted by Jérémie Beucler
cognitionjournal.bsky.social
In our study, we investigated how people evaluate everyday socio-political arguments in the context of their prior beliefs about the topics being discussed.
October 1, 2025 at 3:14 PM
Reposted by Jérémie Beucler
kobedesender.bsky.social
Introducing hMFC: A Bayesian hierarchical model of trial-to-trial fluctuations in decision criterion! Now out in @plos.org Comp Bio.
led by Robin Vloeberghs with @anne-urai.bsky.social Scott Linderman

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓

#PsychSciSky #Neuroscience #Neuroskyence
September 25, 2025 at 9:13 AM
Reposted by Jérémie Beucler
kobedesender.bsky.social
Common neural choice signals reflect accumulated evidence, not confidence! Now out in @cerebralcortex.bsky.social w @helenevanmarcke.bsky.social @pierreledenmat.bsky.social @yfvisser.bsky.social @denizerdil.bsky.social a.o.

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓
September 19, 2025 at 10:47 AM
Reposted by Jérémie Beucler
jamiecummins.bsky.social
Can large language models stand in for human participants?
Many social scientists seem to think so, and are already using "silicon samples" in research.

One problem: depending on the analytic decisions made, you can basically get these samples to show any effect you want.

THREAD 🧵
The threat of analytic flexibility in using large language models to simulate human data: A call to attention
Social scientists are now using large language models to create "silicon samples" - synthetic datasets intended to stand in for human respondents, aimed at revolutionising human subjects research. How...
arxiv.org
September 18, 2025 at 7:56 AM
Reposted by Jérémie Beucler
kyleflaw.com
🪦 New in @pnas.org: we analyzed 38 million U.S. obituaries to ask what signals a life well lived:

What values are people most remembered for?

How do legacies shift with cultural events?

How do age and gender shape what it means to have lived well?

www.pnas.org/doi/10.1073/...
An exploration of basic human values in 38 million obituaries over 30 years | PNAS
How societies remember the dead can reveal what people value in life. We analyzed 38 million obituaries from the United States to examine how perso...
www.pnas.org
August 27, 2025 at 2:39 AM
Reposted by Jérémie Beucler
nicolasbeauvais.bsky.social
Happy to share that my first paper is out in Thinking & Reasoning! 📄📢
With Aikaterini Voudouri, @boissinesther.bsky.social & @wimdeneys.bsky.social we show that deliberate reasoning helps not just to correct but also to justify intuitive judgments.

🔗Full paper: shorturl.at/JTeTi
Quick thread below!
August 21, 2025 at 7:46 AM
Reposted by Jérémie Beucler
edgardubourg.bsky.social
Today’s popular fictions can be extremely far from reality: The Lord of the Rings, Avatar, The Legend of Zelda, Avengers: Endgame. But has this always been the case?
August 12, 2025 at 10:26 AM
jeremiebeucler.bsky.social
Curious about the mechanisms behind biased reasoning and metacognition? 🤔

📍 Come see our poster at #CCN2025, Aug 12, 1:30–4:30pm

We show how a biased drift-diffusion model can explain choice, RT and confidence in a base-rate neglect task, revealing why more deliberation doesn’t always fix bias.
poster about why we don't stop and think, showing the main results of the study
August 11, 2025 at 1:06 PM
Reposted by Jérémie Beucler
jfbonnefon.bsky.social
We need data, not guesses, on how future tech may reshape behavior & society. Our new paper with @azimshariff.bsky.social and @iyadrahwan.bsky.social out in @nature.com spells out a framework we call the ❝science fiction science method❞ (sci-fi-sci) +

www.nature.com/articles/s41...
The science fiction science method - Nature
The ‘science fiction science’ method simulates future technologies and collects quantitative data on the attitudes and behaviours of participants in various future scenarios, with the aim of predictin...
www.nature.com
August 6, 2025 at 3:26 PM
jeremiebeucler.bsky.social
1/8

New (and first) paper accepted at JEP:LMC 🎉

Ever fallen for this type of questions: "How many animals of each kind did Moses take on the Ark?" Most say "Two," forgetting it was Noah, and not Moses, who took the animals on the Ark. But what’s really going on here? 🧵
When asked “How many animals of each kind did Moses take on the Ark?”, most people answer “Two”, failing to notice that it was Noah, and not Moses, who took the animals in the Ark. “Fast-and-slow” dual process accounts of such semantic illusions posit that incorrect responders are not sensitive to their error and that overcoming the illusion requires deliberate correction of an intuitive erroneous answer. We present three experiments that force us to revise this dual process view. We used a two-response paradigm in which participants had to give their first, initial answer under cognitive load and time pressure. Next, participants could take all the time they wanted to deliberate and select a final answer. This enabled us to identify the intuitively generated response that preceded the final response given after deliberation. Results show that participants do not necessarily need to deliberate to avoid the illusion and that incorrect respondents consistently display error sensitivity (as reflected in decreased confidence), even when deliberation is minimized. Both reasoning performance and error sensitivity in the initial, intuitive stage tended to be driven by the semantic relatedness between the anomalous word (e.g., “Moses”) and the undistorted word (e.g., “Noah”). We show how this leads to a revised model where the response to semantic illusions depends on the interplay of both incorrect and correct intuitions.
August 6, 2025 at 10:24 AM
Reposted by Jérémie Beucler
ccortial.bsky.social
Excited to share my latest research !

Key findings reveal that a negative emotion triggered reasoning.

#epistemology #Psychology #Cognition #Research #Science #Neuroscience #Reasoning #CognitiveScience #AcademicTwitter

📖 Read the full study: www.tandfonline.com/doi/full/10....
www.tandfonline.com
July 15, 2025 at 7:50 AM
Reposted by Jérémie Beucler
zoepurcell.bsky.social
Do AI builders hold different values from AI users?

We show that AI builders and men are more utilitarian and less supportive of pro-diversity outputs, highlighting ongoing concerns about workforce diversity and whose values are shaping AI.

tinyurl.com/AIcognit
June 11, 2025 at 10:57 AM
Reposted by Jérémie Beucler
lauracharbit.bsky.social
👀 Just out in Thinking & Reasoning with @boissinesther.bsky.social @mts-raoelison.bsky.social @wimdeneys.bsky.social

Curious how intuitive reasoning develops through adolescence?

🔗 www.tandfonline.com/eprint/KFH5K...

Quick summary👇
July 7, 2025 at 5:49 PM
Reposted by Jérémie Beucler
dgrand.bsky.social
🚨In PNAS🚨
The right often accuses fact-checkers of political bias
But we analyzed Community Notes on Musk's X and found posts flagged as "misleading" are 2.3x more likely to be written by Reps than Dems!
The issue is Reps sharing misinformation, not fact-checker bias...
www.pnas.org/doi/10.1073/...
June 17, 2025 at 12:51 PM
Reposted by Jérémie Beucler
pbogdan.bsky.social
🚨New paper!🚨

Meta-analysis on 4M p-values across 240k psych articles: How has psychology changed since the replication crisis began? How is replicability linked to citations, impact factor, and university prestige? 🧵

Paper: journals.sagepub.com/doi/10.1177/...

Interactive: pbogdan.com/meganal
Figure 1 of the paper
April 9, 2025 at 1:15 PM
Reposted by Jérémie Beucler
jofrhwld.bsky.social
this is roughly my gutfeel for interpreting logit coefficients. #stats #rstats
November 18, 2024 at 10:21 PM
Reposted by Jérémie Beucler
lewan.bsky.social
Honest people don’t lie. Or do they? Liars aren’t honest. Or are they?
One puzzling conundrum in contemporary politics is that politicians who seem to be estranged from facts and evidence are nonetheless considered honest by their followers.
1/n
April 10, 2025 at 10:47 AM
Reposted by Jérémie Beucler
stephenjwild.bsky.social
Using this as an excuse to share a great post by @natehaines.bsky.social on modelling and Dunning-Kruger.

haines-lab.com/post/2021-01...
April 5, 2025 at 7:59 PM
Reposted by Jérémie Beucler
jorge-morales.bsky.social
William James's take on psychophysics is *incredible*. Worth reading the whole quote. #psychSciSky #philsky #VisionScience

"But psychology is passing into a less simple phase. Within a few years what one may call a microscopic psychology has arisen in Germany, carried on by experimental methods,
March 17, 2025 at 6:19 PM
Reposted by Jérémie Beucler
wimdeneys.bsky.social
New preprint: “Folk Thinking, Fast and Slow: Intuitive Preference for Deliberation in Humans and Machines”

Pop culture often praises intuition (“Blink”, Steve Jobs). But do we really trust it? Across 13 studies, we find a strong intuitive preference for deliberation.

tinyurl.com/8r54dmyn (1/6)
February 18, 2025 at 4:24 PM