training up the right amt of 'suspiciousness' w/ data/code/analyses has always been hard (it's a key reason my lab uses R vs point & click stats option, but w/AI the lines have blurred). Hard to know when you know enough to know. There's no 'savvy scientist' test, so general purpose advice is hard.
nice post from @mcxfrank.bsky.social; i'm 100% on board w/thinking=writing, keep it human. my key worry is those using AI *without* sufficient knowledge to know when output is wrong/how to tell (esp. trainees). Things that are obviously 6-fingered people to an expert look just fine to a novice.
Figure showing number of competitive grants mentioning women from 2015-2025. The number was rising until recently, with a precipitous drop in the last year.
At the end of 2024, the National Academies put out a report concluding the NIH has woefully underfunded women’s health research, and they suggested $15 B should be invested over the next 5 years.
Here’s what’s happened instead. Hard to study women’s health if you can’t say “women.”
wapo.st/4euUt1c
related hot take: if you're *opting to attend* a talk on zoom and don't have your video on (and have sufficient internet speed to do so), JUST DON'T GO.
Highlighted in red: "Political interference is inappropriately shaping or interfering in the conduct, management, communication, or use of science for political advantage or such that it undermines impartiality, nonpartisanship, or professional judgement"
"HHS works to promote a culture of scientific integrity by creating an empowering environment for innovation and protecting scientists and the process of science from inappropriate interference. Scientific findings and products must not be suppressed, delayed, or altered for political purposes and must not be subjected to political interference or inappropriate influence. The responsible and ethical conduct of research and other scientific activities requires an environment that is safe and free from harassment and discrimination" Highlighted in red are "suppressed, delayed, or altered for political purposes", "subjected to political interference", and "inappropriate influence."
"HHS works to promote a culture of scientific integrity by creating an empowering environment for innovation and protecting scientists and the process of science from inappropriate interference. Scientific findings and products must not be subjected to interference or inappropriate influence and must not be inappropriately suppressed, delayed, or altered. The responsible and ethical conduct of research and other scientific activities requires an environment that is safe and free from harassment and discrimination." Highlighted in green are "subjected to interference or inappropriate influence" and "inappropriately suppressed, delayed, or altered."
Text, with "political" highlighted in red to indicate removal: I. Protecting Scientific Processes Scientific integrity fosters "honest scientific investigation, open discussion, refined understanding, and a firm commitment to evidence" (OSTP 2010). It also enables consideration and documentation of differing scientific opinions. Practices that support scientific integrity may include peer review and open science. Science, and public trust in science, thrives in an environment that prevents political interference and inappropriate influence from impacting scientific data and analyses and their use in decision making. It is the policy of HHS to: 1. Prohibit political interference or other inappropriate influence in the design, proposal, conduct, review, management, evaluation, communication about, and use of scientific activities and scientific information. Prohibit inappropriate restrictions on resources and capacity that limit and reduce the availability of science and scientific products (e.g., manuscripts for scientific journals, presentations for workshops, conferences, and symposia) outside of normal budgetary or priority-setting processes or without scientific, legal, or security justification. 3. Require that leadership and management ensure that covered individuals engaged in scientific activities can conduct their work objectively, free from political interference or other inappropriate influence, and free from retaliation.
HHS just published an update to its Scientific Integrity Policy. Notably, it has removed the concept of political interference, and no longer calls it out as something specifically to be prevented.
www.hhs.gov/sites/defaul... (deletions in red, insertions in green)
construction sign with a tapped up paper reading WARNING: AGGRESSIVE BIRDS AHEAD. Hawk activity reported. Proceed with caution.
caption contest
Aerial view of a beach in Rio de Janeiro with turquoise water meeting a sandy shore, scattered beachgoers, and a large rock formation near the coastline. In the foreground, a beachfront hotel with a pool, lounge chairs, and palm trees is visible. Overlaid text reads: “100 Days to CogSci2026 – Cognitive Inefficiency – July 22–25, Rio de Janeiro,” alongside small gear icons.
100 days to #CogSci2026 in Rio 🇧🇷
As we get closer, it's a great time to start planning your trip and take advantage of the different ways we're working to make the conference more accessible👇
alas you've fallen for the Science Graph Linearity Assumption. 0 graphs: unacceptably unsciency. esp if there's a landscape table that coulda beena graph.
1-7 graphs: excellent, carry on, very science.
>7 graphs: past the information threshold bitrate for most mere mortals. pace George Miller
This is incredible: Google Has a Secret Reference Desk. Here’s How to Use It. I knew some of these but not all, e.g. verbatim mode “returns results for exactly what you typed, stripped of personalization and synonym-swapping.” [cardcatalogforlife.substack.com]
We’re hiring: remote-eligible Provost Postdoctoral Fellow, SEA Lab at Vanderbilt. Great fit for candidates interested in early caregiving, developmental psychopathology, neuroimaging, and ecological methods. Start Aug 1-Dec 1, 2026. Apply: k.humphreys@vanderbilt.edu
ooh i stand corrected, WAY fewer kid vs. llm comparison's at CogDev this past weekend than last time (or some other recent conf's)! It was a blast, with two particularly fabulous keynotes by @kbyers.bsky.social and the (bluesky-less) Lisa Feigenson. Love that post-conference invigoration!
I’m offering a lightly used set of regalia in UC Berkeley colors to a junior faculty member, preference for a mom and 1st gen scholar. DM me your details— I’ll ship for free.
How to join zillions of lexical norms to each word in your language sample the easy way: a quick tutorial and demo reilly-lab.github.io/Jamie_JoinLe...
honest question if you answer fewer emails do you get fewer emails? thanks for your attention to this matter.
Some great developmental science will be happening in Panama this July! Come hear more about what @infantstudies.bsky.social and @fitngin.bsky.social have planned! ☀️🇵🇦
Hey just a year ago this SBE conf got axed & we thought "Doge, yikes!" & now here we are w/a shuttered SBE! How time flies.
p.s. asked NSF SBE POs ab submitting a grant & they said "we've been directed to pass inquiries to NSF's Office of Legislative and Public Affairs".
burning_its_fine.gif
Join me! Here's another way to communicate the impact of NSF's plans to eliminate SBE: email NSF at: olpa-leadership@nsf.gov
I'll be writing to emphasize that their new AI-based mission can't be achieved without SBE (brain/mind) research.
www.nsf.gov/od/olpa
6) totally down for everyone to put a 'who cares' slide in their #CDS2026 talk in montreal and cite this blogpost. but for god's sake ask yourself the same question about the LLMs that are or aren't childlike. (spoiler: they aren't childlike in their wonder)
5) i think i broadly agree w/ @paulbloomatyale.bsky.social. for me: the science is better when it *could have turned out otherwise* in a way that tells you smthng ab the mind, it's order of operations, how learning emerges from the constraints of our cognitive system interacting w/experience, etc.
4) ofc this interfaces with the broader challenge of the 'minimally publishable unit' and what academia rewards and how. but that's a bigger problem than 'showing kids get better at stuff is boring'.
3) for some areas, you need the boring descriptives first because w/o them you can't see where the theory does or doesn't make contact with the 'facts on the ground' in terms of learnability. So yeah, we have some boring papers on our way to the (rarer) good stuff.
2) ymmv, but I still prefer dev talks where i can impute why it might matter (agreeing it's not always possible) to the other new CDS schema on the block: 'kids do this, but whadabout machines? here's what my bespoke and immediately outdated LLM version found! it's the same! or its not! neat!'
1 con't: I think this is at least partly bc it's hard to get funds (& thus data) w/o a solid answer to 'who cares'. As a wee PhD student (~2009) Henry Gleitman (RIP) once accosted me jovially at a party, asking in his thick German accent "I am Obama. Why should I give you money for your research?"
a few thoughts here, premised on agreeing that a chunk of devpsych research is boring.
1) I think niche subfield conf's like CDS (and e.g. for me, BUCLD) exacerbate this, there often *is* a good reason under the hood, just not always one a trainee can quickly frame well in a 15-20' talk or poster.
It’s a metaphor isn‘t it. ISN’T IT.
Apropos of nothing, new theory: in 2parent households the parent that’s better at getting themselves out the door in a timely fashion is also better at getting any children out the door. genetic and environmental factors both likely;). Data to follow.
#Artemis II - We have the first image from yesterdays Lunar flyby captured by the crew on Orion
EARTHSET.
April 6, 2026.
Humanity, from the other side. First photo from the far side of the Moon. Captured from Orion as Earth dips beyond the lunar horizon.
🚀🌕🌍
The Artemis II crew proposed naming a moon crater after Commander Reid Wiseman’s late wife Carroll, who lost her life to cancer 😭
I gave my grad class a sermon week 1 about how they could use AI & get 100% & never come to class, but what’s the point?
Result: by week 13 only 6 of 26 students are still coming to class (but all are submitting perfect homeworks).
Many thoughts, none good
#academicsky #psychscisky #neuroskyence
Students always find this content some of the most actually helpful stuff we cover in my cognition course! and helps them understand why we build in reading checks and in class problems to solve, nice summary!
