Marius Mercier

PhD Candidate in Cognitive Sciences | @ENS_Ulm. Currently working on the dynamics of impression formation (e.g competence judgment).

[Marius Mercier @MariusMercier]

"What goes on four feet in the morning, two feet at midday, three feet in the evening?"

This riddle is thousands of years old, but why does it stick with us?

In our new paper, we argue that the feeling of insight might explain the lasting success of some cultural products.

[Koenfucius 🔍 @koenfucius]

Research by @mariusmercier et al suggests people can predict, with broad accuracy, from the (rarity of) conspiracy theories someone endorses/rejects what other conspiracy theories they would endorse/reject, and infer related personality traits:

https://t.co/5WfcURvrlA

[Marius Mercier @MariusMercier]

@kepano @obsdmd Native pdf highlights within obsidian :)

[Marius Mercier @MariusMercier]

📄 Preprint: https://t.co/BRxQr37Odl

💾 Data, code, materials, and preregistrations: https://t.co/fJQVfnTcqK

Huge thanks to my co-authors, all senior authors with equal contribution: @OliverWithAnI, @hugoreasoning, @TadegQuillien

[Marius Mercier @MariusMercier]

🎉New preprint: Bayesian Competence Inference guides Knowledge Attribution and Information search

In our new preprint, we examine how people infer others’ competence from sparse evidence, and the computational principles underlying these judgments.

[Marius Mercier @MariusMercier]

⚙️ We analyze behavior both at the aggregated level and at the trial level, showing that participants’ accuracy is not merely a wisdom-of-crowds artifact.

🤖 Our Bayesian model outperforms several plausible heuristic and explains ~90% of the variance at the aggregated level.

[Marius Mercier @MariusMercier]

More in the paper:
📊 We combine a rich behavioral dataset (N=1,916; 113,400 knowledge-attribution judgments; 2,960 search choices) with a large and diverse stimulus set (N >1200).

[Tanay Katiyar @adigitaltanay]

My first paper, out in PsychReview!! Along with Amy, Nik and Adrian, @RealAdamHunt & I revisit an old theoretical question using concepts from evo psychiatry and anthro:

Why is evidence linking social media to mental health so mixed?

A 🧵

[Edgar Dubourg @EdgarDubourg]

Mon livre est en librairie depuis hier !

[Marius Mercier @MariusMercier]

Report from google:

https://t.co/emaFBcDttC

[Ethan Mollick @emollick]

We have data on the environmental impact per AI prompt:
Gemini: 0.00024 kWh & 0.26 mL water
ChatGPT: 0.0003 kWh & 0.38 mL
...the same energy as one Google search in 2008 & 6 drops of water.

Seems to be improving, too: Google reports a 33x drop in energy use per prompt in a year.

[Marius Mercier @MariusMercier]

Quite interesting: LLM’s inference is becoming more and more efficient. Current estimates say that the median text query is equivalent to less than 1 minutes of streaming (~0.35 Wh). But this doesn’t include the costs of training the models…

https://t.co/xKLN4H6JSR

[Marius Mercier @MariusMercier]

@jeremie_beucler Congrats !! 🎉

[Marius Mercier @MariusMercier]

@jvaldpmqa Oui, c'est vrai que l'élégance d'une preuve a une place particulière en mathématique. Merci pour le partage de l'article, c'est assez surprenant

[Marius Mercier @MariusMercier]

@jvaldpmqa (Pour être clair, je ne soutiens pas l'argumentaire du type à qui tu répondais)

[Marius Mercier @MariusMercier]

@jvaldpmqa Je ne connais pas du tout la littérature sur le sujet, mais les attendus d’une démonstration mathématique semblent plus explicites que ceux concernant les bonnes tournures de phrase à employer dans une dissertation, non ?