samuel mehr
LightNews LightNews
banner
mehr.nz
samuel mehr
@mehr.nz
8.1K followers 1K following 13K posts
i am a cognitive scientist working on auditory perception at the University of Auckland and the Yale Child Study Center 🇳🇿🇺🇸🇫🇷🇨🇦

lab: themusiclab.org
personal: mehr.nz
intro to my research: youtu.be/-vJ7Jygr1eg
Posts Replies Media Videos
Reposted by samuel mehr
disabilitystor1.bsky.social
its not lost on me that its a lot of academic men defending that very poorly designed Chicago study
January 6, 2026 at 6:32 PM
mehr.nz
Took me a few days to work this out, but the energy I'm going into 2026 with is my 5yo's, when she looked at a terrifying ski chair machine and went "fuck it, let's go"
mehr.nz samuel mehr @mehr.nz · 23h
5yo went skiing for the 3rd time in her life and was very brave and rode the chairlift without them even needing to slow it down so she could get on and i'm not crying you're crying
January 6, 2026 at 6:13 PM
Reposted by samuel mehr
proptermalone.bsky.social
introducing AskBeeves, a Chrome extension for Bluesky that displays how many people you're following are blocked or blocked by a user in their profile. 100% vibecoded with Claude Code, use at own risk and so forth.

github.com/PropterMalon...
Blocked by 34 people you follow and blocking 33 people you follow.
January 6, 2026 at 4:51 PM
Reposted by samuel mehr
edhagen.net
1. My new preprint has its own bluesky account. Why? The problems facing social media & scientific publishing are similar: both are dominated by powerful oligopolies. The @atproto.com tech underlying bluesky that aims to solve the social media prob might also help solve the scientific pub prob 🧪 🧵
menopause-preprint.edhagen.net Menopause evolution preprint @menopause-preprint.edhagen.net · 7h
1. Preprint: Menopause averted a midlife energetic crisis with help from older children and parents: A simulation study. zenodo.org/records/1814...

Menopause is rare, known to occur only in humans and toothed whales: 🧵
Post-reproductive lifespans of humans, African great apes and other primates, and toothed whales. Values represent the life expectancy from the end of the fertile period as a proportion of life expectancy from the beginning of the fertile period. Human and nonhuman primate values are from Wood et al. (2023), except for the largest gorilla value, which is from Smit & Robbins (2025). Toothed whale values are from Ellis, Franks, Nattrass, Currie, et al. (2018). Chart shows that these values are high for humans and toothed whales, and low for primates
January 6, 2026 at 3:12 PM
mehr.nz
from my DMs: "every few weeks a researcher makes me go 'ok so maybe my overzealous IRB has a point'"
January 6, 2026 at 2:33 PM
mehr.nz
Folks getting in the weeds about licensing agreements but this research study engaged me in research, by creating a bespoke dataset about me, before I agreed to participate. They already made the dataset before I was invited to participated. That's not how informed consent works
mehr.nz samuel mehr @mehr.nz · 23h
somebody at UChicago is feeding preprints to LLMs without authors' consent, in a research study

they have the gall to suggest to authors they've opted-in that they volunteer to evaluate the LLMs' suggestions regarding their own work.

lol, lmao even. here is the invite and my reply
Dear Author, I am Honglin Bao (www.hbao.info), a data science PhD student in Knowledge Lab (https://knowledgelab.org/) at the University of Chicago, working on a study about “The Capabilities and Potential of AI for Automating Scientific Idealization: A Large-Scale Human-in-the-Loop Study” (under UChicago IRB25-1372). As AI plays an increasingly significant role in scientific discovery, this project aims to evaluate how far AI helps scientists extend their work in new directions – or if it is not useful in this regard. This is a research project – it has no commercial purpose. I will share the final draft, technical details, and aggregate results from scientists once the study is complete. We came across your preprint "Principles of music perception" and applied custom AI models on it in accordance with the terms of the preprint license, which generated five new extensions (i.e., ideas proposed by AI based on the same context as your paper). Our pilot study of 400 scientists (excluding you) indicates that around 60% found at least one of the proposed extensions thought-provoking and actionable. However, the remaining reported that AI only generates seemingly plausible but vague sentences. We solicit your help in evaluating whether you think AI is useful or not, since we believe you are the best judge of new ideas proposed from your own paper. No matter where you stand regarding AI and science: From principally believing that AI is prone to producing fake and morally problematic science to finding it useful, or something in between, we hope to receive evaluations that reflect a spectrum of views. The AI-proposed extensions are included in the survey link below. If you consent to review them and participate in the survey, we would appreciate it if you could do so within the next 14 days (it will take you < 10mins). If you do not consent, we will not use your research in this study. If you find these ideas valuable, you are welcome to purs… Please remove my data from your study. I want nothing to do with this — and I would also point out that if you had done appropriate background work in the selection of articles you fed to the plagiarism machine, you would have found that the preprint of mine you used is way out of date and has been updated substantially since it was first posted. ——— Samuel Mehr School of Psychology, University of Auckland and Child Study Center, Yale University Be a citizen scientist at themusiclab.org!
January 6, 2026 at 1:40 PM
Reposted by samuel mehr
akkerroos.bsky.social
I love the cartoon by Sketchplanations
sketchplanations.com/sampling-bias
Sampling bias
If you wanted to know who would win an election, then the only reliable (most of the time) way to do that is to ask everyone to vote for real and count them all. Short of doing that, if you want to ge...
sketchplanations.com
January 6, 2026 at 6:00 AM
Reposted by samuel mehr
kenwhite.bsky.social
Has a court ruled on whether, under Section 230, a web site’s AI generating content based on a user prompt is treated as the web site’s speech? Seems to me it should be, just as if the site operator published speech answering someone’s question.

BWEPRA
hairybusiness.bsky.social Hairybusiness @hairybusiness.bsky.social · 1d
I am not a Section 230 expert but being that Twitter is creating and publishing these photos I wonder if they actually have any civil legal protection? @kenwhite.bsky.social ? @akivamcohen.bsky.social ?
eliothiggins.bsky.social Eliot Higgins @eliothiggins.bsky.social · 1d
What X is allowing Grok to do is unlawful or regulatory-illegal in multiple jurisdictions, creates clear DSA violations, and continues despite regulator scrutiny. The images do not need to be pornographic, non-consensual sexualised or intimate depictions are enough.
January 5, 2026 at 4:09 PM
mehr.nz
a person who knows my tastes well gave me a box of excellent marrons glacés for christmas and after a day of running around in the frigid cold, it is involving substantial inhibitory control to refrain from powering thru the whole box
a box of Eataly marrons glacés
January 6, 2026 at 3:17 AM
mehr.nz
somebody at UChicago is feeding preprints to LLMs without authors' consent, in a research study

they have the gall to suggest to authors they've opted-in that they volunteer to evaluate the LLMs' suggestions regarding their own work.

lol, lmao even. here is the invite and my reply
Dear Author, I am Honglin Bao (www.hbao.info), a data science PhD student in Knowledge Lab (https://knowledgelab.org/) at the University of Chicago, working on a study about “The Capabilities and Potential of AI for Automating Scientific Idealization: A Large-Scale Human-in-the-Loop Study” (under UChicago IRB25-1372). As AI plays an increasingly significant role in scientific discovery, this project aims to evaluate how far AI helps scientists extend their work in new directions – or if it is not useful in this regard. This is a research project – it has no commercial purpose. I will share the final draft, technical details, and aggregate results from scientists once the study is complete. We came across your preprint "Principles of music perception" and applied custom AI models on it in accordance with the terms of the preprint license, which generated five new extensions (i.e., ideas proposed by AI based on the same context as your paper). Our pilot study of 400 scientists (excluding you) indicates that around 60% found at least one of the proposed extensions thought-provoking and actionable. However, the remaining reported that AI only generates seemingly plausible but vague sentences. We solicit your help in evaluating whether you think AI is useful or not, since we believe you are the best judge of new ideas proposed from your own paper. No matter where you stand regarding AI and science: From principally believing that AI is prone to producing fake and morally problematic science to finding it useful, or something in between, we hope to receive evaluations that reflect a spectrum of views. The AI-proposed extensions are included in the survey link below. If you consent to review them and participate in the survey, we would appreciate it if you could do so within the next 14 days (it will take you < 10mins). If you do not consent, we will not use your research in this study. If you find these ideas valuable, you are welcome to purs… Please remove my data from your study. I want nothing to do with this — and I would also point out that if you had done appropriate background work in the selection of articles you fed to the plagiarism machine, you would have found that the preprint of mine you used is way out of date and has been updated substantially since it was first posted. ——— Samuel Mehr School of Psychology, University of Auckland and Child Study Center, Yale University Be a citizen scientist at themusiclab.org!
January 5, 2026 at 10:55 PM
mehr.nz
5yo went skiing for the 3rd time in her life and was very brave and rode the chairlift without them even needing to slow it down so she could get on and i'm not crying you're crying
January 5, 2026 at 10:35 PM
mehr.nz
developmental psycholinguistics status check: do your children use 'inside-out' as a verb? as in 'i need to put on my jammies can you inside-out the pants'
January 5, 2026 at 1:26 AM
mehr.nz
starting a band called Guanciale Funk who's in
January 4, 2026 at 3:37 PM
mehr.nz
You never see the faces of the subjects of this excellent short film, the MTurk workers who annotate images for task-based pay on the internet

Many psychologists use 'worker' populations for their research studies, usually as paid participants (and often at very low rates), exactly like this
Opinion | Self-Driving Cars Can’t See Without Their Eyes
www.nytimes.com
January 4, 2026 at 2:13 PM
mehr.nz
when it's very very cold the children can walk on water 🇨🇦
three kids walking on a very frozen lake
January 3, 2026 at 10:09 PM
Reposted by samuel mehr
davidehrlich.bsky.social
1/20/26
December 30, 2025 at 8:22 PM
mehr.nz
when the editor shuts down a crabby reviewer
mad eric adams / chortling bernie sanders meme with 'reviewer 2, who hated the paper' and 'the first author, opening up the decision letter'
January 2, 2026 at 2:18 AM
mehr.nz
a restaurant-kitchen practice I highly recommend to all cooks is recipe logging, wherein, especially for complex dishes, one dates & documents each attempt with alterations and their outcomes

so, my first written date of the year is 01 Jan 2026, for a new years' bejgli (following my xmas batch)
cross section of the two bejgli (walnut and poppyseed) the full bejgli, unsliced my alterations mákos & diós bejgli 01 Jan 2026 Sam 110 g powdered sugar instead in the dough as we ran out of white sugar Increased poppyseed to 200 g bc the package was 200 g Added a half grated apple to each of the two fillings No cinnamon Ground walnuts in cuisinart because I didn't have a nut grinder Another spoon of cocoa in dios filling Dough sat in fridge overnight dried 30 min per coating in warm kitchen (2 yolk) not fridge proofed 45 min after white coating on a warm oven These browned a bit too fast, turned temp down to 275 in last 6 min or so - maybe because bejgli were fully warmed with rising. Perfect texture. Fillings were moister with the apple but I don't like the extra sweetness from it. Maybe reduce sugar in syrup if keeping apple. Definitely return cinnamon to fillings and increase salt in dios. 19 Dec 2025 Sam This makes 4 small bejgli or 2 large ones. It does not require proofing; if you want a breadier texture to the filling, then proof the dough for 2-3 hours in a warm, moist oven while making the fillings rather than just letting the dough rest in the fridge.
January 2, 2026 at 1:36 AM
Reposted by samuel mehr
clapifyoulikeme.favrd.social
this is the most incredible interview i’ve ever read. the aei guy tries so hard to get sliwa to trash mamdani and he simply won’t. he also brings up flaco (rip) unprompted.

www.wsj.com/opinion/free...
Tunku Varadarajan: Will you ask New Yorkers who voted for you to give Zohran Mamdani a chance? Curtis Sliwa: Absolutely. He won a majority. He has a mandate. The Masters of the Universe, the billionaires, did everything they could to make sure he wasn't the mayor. It didn't work. He's a great organizer. I'll give you an example: Melinda Katz, the current District Attorney for Queens County, New York. She's the mother of my two youngest sons. She was running for DA in 2019. She had the institutional support. Moderate Jewish mother. Loyal Democrat. Queens borough president. Perfect résumé. She's running against Tiffany Caban, who wasn't even in elective office at that time. Very radical. You know, "do away with jails, don't arrest anyone." Tiffany Caban won on election night. But she lost the race on the absentee ballots, but only by about 20 votes. You know who the street organizer was for Tiffany Caban? Zohran Mamdani. So, when they talk about him as a trust fund baby, I say no, no, no, no, no. You don't like it when people call him a "nepo baby." He's earned his stripes. He had 100,000 volunteers going door-to-door, and I experienced that in some of the neighborhoods where you would never have seen anybody liberal or progressive go. Howard Beach! In conservative Howard Beach, they're going block to block. They were everywhere. 100,000. I call them Zohranistas. This was the secret to his election, not TikTok. It was good, old-fashioned retail politics. Frankly, Cuomo and not Mamdani was the nepo baby in this race. Born on third base. Felt entitled. Knew the political process, but mailed it in. He didn't have the stomach to actually go out and do retail politics. And shame on him. Do you think Mamdani will make a better mayor than Andrew Cuomo might have done? How would you describe Mamdani's politics? He's not a communist. People call him that. That's ridiculous. He is a democratic socialist. David Dinkins [New York mayor from 1990-93] was a democratic socialist. We've had a long history in New York City of socialism. What's the biggest challenge Mamdani faces as mayor! I don't think the Zohranistas totally understand the fiscal irresponsibility we just went through with the Adams years and how little money there's left. There's not enough money to do all the things that Zohran wants to do. A last question. All through our interview you pronounced Mamdani as "Mandami," transposing the "m" and the "n" in his name. Are you aware of the mispronunciation? Many people, Cuomo included, mangle his name. In the debate, Zohran brought that to my attention, my mispronunciation. I didn't do it on purpose. There's no disrespect. I have a hard time with it. It took people about 10 years to pronounce my last name correctly: "Sleewah.'" They pronounced it Silva, Saliva, Slywah. In the streets, we would have given Zohran a nickname, like I did for all the Guardian Angels. "Z Man" would have been perfect for him. That would have been his handle. Mr. Varadarajan, a Journal contributor, is a fellow at the American Enterprise Institute and at NYU Law School's Classical Liberal Institute.
January 1, 2026 at 11:16 PM
mehr.nz
Top contenders so far

A man who confidently makes an assertion that is plainly false is _________
- brofessing
- floofing
- concockting
mehr.nz samuel mehr @mehr.nz · 5d
the term 'mansplain' is well known but is there also a term for a man confidently making an assertion that, upon reflection, is plainly false?
January 1, 2026 at 7:21 PM
mehr.nz
Canada is ringing in 2026 properly (the low temp today is -30C where we are)
January 1, 2026 at 2:50 PM
mehr.nz
Happy new years to all you people on here. Internet Friends are the best ❤️❤️
January 1, 2026 at 3:42 AM
Reposted by samuel mehr
danielketter.com
GET YOUR KICKS in 2026
December 31, 2025 at 9:22 PM
mehr.nz
the term 'mansplain' is well known but is there also a term for a man confidently making an assertion that, upon reflection, is plainly false?
January 1, 2026 at 12:49 AM