They are using people pressing the slop button to train their llm to recognise llm output that humans clock as llm output.

🔔🔔 Russian polls have shown ~75% support for the war in Ukraine since February 2022, essentially flat. New working paper with Kirill Chmel and @nikitasavin.bsky.social shows that number is stable partly because it stopped measuring what people actually think. 🧵 ssrn.com/abstract=711...

Post coding agents, it's amazing to realize how "I need to do this myself because I know what I want" meant exactly the opposite. You did not know what you wanted, that's why you can't verbalize it. You only know that if you type and mechanically follow your intuition, you will get where you want.

In other words, I lack the discernment to evaluate the output of the chatbot that Tao exchanged theories with. Show me an equally opaque transcript of a "conversation" between a chatbot and a crank with AI psychosis whose math made no sense whatsoever. 5/

We change constantly — new roles, new beliefs, new ways of seeing our past. Yet most of us feel like the same person over time. How? Brent Hughes and I explore this in our new paper in Trends in Cognitive Sciences. 🧵 www.cell.com/trends/cogni...

Wired for coherence: a network theory of self

People change across daily situations and life transformations, yet maintain a stable sense of self. Explaining this stability requires understanding how identities, traits, values, and memories inter...

cell.com

The spaces between meetings are not long enough to get meaningful work done. I know this. I know. And yet

a new job market paper! using surveys from ~9.5 million college students over 35 years, I study the role of ideology in career aspirations. I show that ideology structures what careers people want, a pattern much stronger among high-income students. read here: tkeskinturk.github.io/aspirations.pdf

out today! www.pnas.org/doi/10.1073/...

Chaz Firestone@chazfirestone.bsky.social · last mo.

You press ‘send’ on a high-stakes email but hesitate when lifting your finger. You make a chess move but linger on the piece before letting go. Familiar? In @pnas.org, Hanbei Zhou @ruizhegoh.bsky.social @ianbphillips.bsky.social & I show that response *duration* tracks confidence! shorturl.at/A9NVe

My thanks to the New Atlantis, letting me do this series on "How the System Works," and for removing the paywall today. It's maddening how little attention our society--and our political leaders, who take cues from us--pay to the systems that have made things better for so many billions.

How the System Works

A series on the hidden mechanisms that support modern life — and what happens if we don’t maintain them

thenewatlantis.com

It feels a certain way knowing that @mjcrockett.bsky.social and I published the rebuttal to this article… six months ago.

AI Surrogates and illusions of generalizability in cognitive science

Recent advances in artificial intelligence (AI) have generated enthusiasm for using AI simulations of human research participants to generate new know…

sciencedirect.com

This skeet will self destruct@pardoguerra.bsky.social · 4w ago

This is quite the finding for a device thats just a plagiarism machine, or whatever we are calling it this week. www.nature.com/articles/s41...

How do people build new ideas from existing knowledge? By tracking Wikipedia searches, we found that novel ideas emerged when people balanced exploration with exploitation—alternating between distant topics and deep dives into related information.

Frontier lab people love to drop ”based on secret models I can’t show you, I probably won’t have a job in six months” as a funny little joke. Ha ha so funny what a nice little joke

Attaching third-party fact-checks to false news reduces both belief and sharing. But do people want to see this context? In a new preprint, @jaewkwon.bsky.social and I find that people choose to read fact-checks nearly 80% of the time. 1/9 osf.io/preprints/psyarxiv/84d7h

4/ Here's the puzzle. Give young children two tests with the exact same logical structure — one outcome is guaranteed, the others are merely possible — and the same child can breeze through one and flounder on the other. Same logic, wildly different performance. Why?

"I used an LLM to help me write, but it was just to polish the style and grammar" is something you've probably come across, or done yourself. Many conferences and journals and teachers are ok with this. But does it...? Just focus on the style and grammar...?

Visiting my university’s new, expensively-refurbished space for startup incubators and senior management and cannot imagine a better illustration of the American university at this moment in time than a breathlessly press-released “collaboration room” featuring a single desk and chair.

What's more nonsensical: smashing a pumpkin using a number, or growing flowers inside a sneeze? Our paper on graded inconceivability is out now in Cognition! Come for the cognitive science 🧠🔍, stay for the whimsy 🌼🧚! 🔗Journal link: bit.ly/gradedInconCog

Tomer Ullman@tomerullman.bsky.social · last mo.

Now out (for realz) in Cognition: "People Make Graded Judgments About The Inconceivable" (by Hu, Sosa, & me) Free preprint: www.tomerullman.org/papers/grade... Journal link: bit.ly/gradedInconCog @jennhu.bsky.social @cognitionjournal.bsky.social

I was today years old when I learned base #Rstats has a helper function `example()`. Better late than never, and it's only been half a three decades give or take ...

Dirk Eddelbuettel@eddelbuettel.com · last mo.

I was today years old when I learned base #Rstats has a helper function `data.matrix()`. Better late than never, and it's only been three decades give or take ...

Results show a clean crossover: 🤖LLMs were more accurate on genuine riddles than riddle riddles (84.9% vs. 50.7%). 🧑Humans showed the opposite pattern (50.5% vs. 80.5%). ♟️We also found the same crossover in whether they used the correct reasoning strategy.

Climate.us is officially live! After NOAA ended Climate.gov’s day-to-day operations in 2025, former team members built Climate.us to carry that work forward. Their explanations and graphics have always stood out to me as clear, accessible, and transparent. I’m really looking forward to this reboot.

Climate.us Home

independent, nonprofit, and immune to politics

climate.us

Do we perceive animacy itself, beyond its lower-level visual correlates? In @elife.bsky.social, @chazfirestone.bsky.social and I leverage “visual anagrams” — images whose interpretations change with orientation — to suggest the answer is: yes! elifesciences.org/reviewed-pre...

In this new JEP:General paper, we show that geometric shapes are organized as tree structures in a language of thought. We apply several linguistic tests for nested constituents, including structural ambiguity, constituent subparts, and syntactic movement, and find analogues in the visual domain.

Barbu Revencu@barbu-revencu.bsky.social · 2mo ago

Representations of geometric shapes have syntactic structure w/ @maxencepajot.bsky.social and @standehaene.bsky.social is out & open-access in JEP:General doi.org/10.1037/xge0.... For an overview, see thread below!

Can infants or other animals represent "mutually exclusive possibilities"? In a new paper in JEP:G, we argue for specifying: in thinking or seeing? We show that in object perception (shared with infants and many animals), the answer is yes. (w Peter Mazalik & Roman Feiman) osf.io/preprints/ps... /1

One of the first studies from my PhD is out now in JEP:G 🥳We tested whether people can infer the truth from teachers who were either helpful, misleading, or randomly sampling. With Keith Ransom and @perfors.net psycnet.apa.org/fulltext/202...

I'm with Aaron on this: virality is just a terrible epistemic criterion, and the way to fix this is for us collectively to learn that the socials are for fun & networking but mostly not for epistemics

direct.mit.edu/opmi/article... maybe there's a way to study metacognitive monitoring w/o collecting confidence? could be handy for studying animals and infants?

Response Time as a Proxy for Decision Confidence: Insights From Type-2 ROC Analysis

Abstract. This study explores the use of response time (RT) data in type-2 receiver operating characteristic (ROC) analysis, a method traditionally used to examine the relationship between confidence ...

direct.mit.edu

⭐️From Emily Liquin: More Time and Effort, Same Curiosity: Expected Effort Does Not Impact Curiosity

More Time and Effort, Same Curiosity: Expected Effort Does Not Impact Curiosity

Abstract. Why do people feel curious about some questions but not others? Recent accounts of curiosity argue that curiosity should be highest when learning is likely to occur and likely to be rapid. H...

doi.org

New paper accepted as a proceedings of the Cognitive Science Society: "Inferring arithmetic skills from speed and accuracy”! We tested whether people were optimal in their inference of others' math skills. doi.org/10.31234/osf...

When you look at a complex scene, you perceive various ensembles — sets of items of various sizes, numbers, etc. Are some more salient than others? In a new paper (w/ samiyousif.bsky.social), we provide an answer to this question! osf.io/preprints/ps...

Pointing out that “everyone else is already doing it!” is a classic way to motivate prosocial behavior. But how can we encourage actions which aren't popular yet? In a new paper in @jexpsocpsych.bsky.social, Mike Norton and I had a lot of fun testing a method using "collective streaks." (1/5)