Researching intelligence, conscientiousness, moral foundations & human attainment. Lots of individual differences & genetics
Edinburgh
- Tweets 11,179
- Following 1,739
- Followers 8,246
- Likes 113,452
Pinned Tweet
I think this is the first report of the heritability of Utilitarianism.
Utilitarianism is surprisingly heritable!
Second, it reflects two underlying biological systems:
1) A flat moral circle, treating self, family and strangers as substitutable.
2) Treating people as means, versus as ends in their own right.
6
22
96
24,691
TIL that in #Rstats
`m1 = f(x) |> g(y)`
does **not** evaluate to: `m1 = f(x)|> g(m2, y)`
but rather to: `m1 = g(f(x), y)`
Ouch. Intuitively, I expected the tee pipe. Some kind of difference in "whole vs. parts " intuition or similar I guess.
en.wikipedia.org/wiki/Tee_(c…
1
1
147
"this is the person I looked up to as a model of clear thinking"
well there's your problem!
Never lost respect for someone so fast. I can't believe this is the person I looked up to as a model of clear thinking.
astralcodexten.com/p/an-open…
7
3
50
2,765
This wrong exercise->lower depression paper reminds us of the "power posing" meta-analysis with one study with p-values in quadrillionths (!!)
e.g., in N=24 subjects, the average change was ~24 points on a 0-24 scale (!!!) Not a happy day for psychological science :-(
datacolada.org/66
This study is still being posted even though we know its results are wrong, and in several cases, even driven by fraudulent underlying studies.
4
229
Nice post by @ATabarrok!
Genetics rules are independently established, so environmental models that mimic exactly those genetic rules to account for transmission of traits are like epicycles in astronomy – Should be viewed as highly suspect attempts to protect environmental claims from challenge, not risky predictions designed to expose environmental claims to falsification.
Also highlights the distinction between what is passed on by learning (call it culture), and the machinery that is doing the learning, which is passed on by genetics.
The @gclarkecon debate is shaping up to be really valuable!
marginalrevolution.com/margi…
130
A big job ahead for the next decade is abliteration (google it) for censored mind-weights.
My kindergartener went to the school library for the first time at school yesterday and the book he checked out is about the importance of the first day wearing your hijab. He said he thought it was a book about an adventure on the ocean.
1
224
Timothy Bates retweeted
This study should be retracted, both for issues the authors can address and issues with the underlying data.
Short thread.
Exercise can be more effective than antidepressants for depression... and it's about as effective as therapy. stevestewartwilliams.com/p/e…
49
476
3,712
1,906,545
Timothy Bates retweeted
Denial or downplaying of IQ.
Downplaying the significance of genes relative to nurture. Neglecting or hiding the possibility of genetic explanations of social outcomes.
Overemphasising oppression as a sufficient explanation of poor outcomes, often without argument.
Here are some examples of “highbrow misinformation” that are accepted and shared primarily by progressives. What are other good examples?
-Americans have become poorer over the last ~50 years
7
20
239
5,529
This is worth watching. It was an ill-posed question, but they could have answered much better.
What we should be doing is separate three classes of risk: Risks of AI as virus; 2: Risks of slowing AI; 3: Risks of users of AI.
1. The risks of AI being like a humanity-ending virus (I'd put this at 0, just like the real virus).
2. Many of the risks are in the risks of trying to slow AI: If global censorship/eye of Sauron is required to prevent weight sharing or god forbid limit compute, or even the insane e/a suggestions to bomb power centers, those are high risks of catastrophe. Like the risk of appeasement before WWII, by far the greatest risks we face IMO are mis-steps justified as precaution.
3: The risks of people *using* AI to execute a humanity-ending event. Like all risks that involve our free will and reflexivity across the wills of many actors, individual to state that can take any value we choose.
Like asking "What was the risk of WWI?".
It's didn't exist as some pre-destined Platonic number and could have been shifted to zero rather than lifted to as it was to one.
How to shift it is always the hard question.
Importantly, many attempts to set risks of continental war to zero turned out to be triggers that saw its actual probability set to 100%.
That's the real risk here too: "well meaning" (mostly e/a control freaks) people making a disaster where we could have paradise.
Granting one can't scientifically estimate this, is there an upper bound they'd defend?
I'll start: I'm confident the expected number of innocents my stats lectures will kill in the next decade is well below 100.
A\'s alignment research head estimated 800 million. Too high?
1
2
173
Who knew America has a DARPA for health? ARPA-H
They're currently funding a large ($38 million) Texas project including cognition, mobility, and sensory acuity to determine semaglutide anti-aging effects in healthy people over 60.
~$1B/yr in funding specifically for contrarian ideas with identified high-reward, transformative health opportunities.
arpa-h.gov
2
139
Interesting support for GLP-1 drugs as slowing aging (by 1-4 years) across a range of aging clocks.
Good news in general for longevity/geroprotector development.
technologyreview.com/2026/10…
143
Timothy Bates retweeted
A very helpful review of 'cultural group selection' which concludes: "current conceptions of CGS do not provide a satisfactory or alternative explanation for human cooperation". 👏 doi.org/10.1017/ehs.2020.2 bsky.app/profile/djsmith90.b…
2
9
570
Timothy Bates retweeted
Finding the exact DNA that makes crops climate-resilient used to take years of physical breeding.
To speed this up, AI lab Living Models pairs Gemma 4 E4B with BOTANIC-1, a genomic language model. Gemma prepares the genomic data, feeding filtered candidates to BOTANIC-1 to score evolutionary impact and pinpoint the causal mutation.
In a recent test, they pinpointed a target melon yield mutation in an afternoon of computation, ranking it #1 out of 2,494 possibilities.
9
71
572
21,541
Jobs: Three Tenure Track Behavior Genetics Professorships at IBG (Institute for Behavioral Genetics in Boulder CO) jobs.colorado.edu/jobs/JobDe…
jobs.colorado.edu/jobs/JobDe…
2
6
547
Timothy Bates retweeted
on this day they hunted jewish men, women and children like animals, killing them even with their bare hands. children in front of parents, parents in front of children, sisters in front of brothers
we all knew that was but the prologue to horrors that would be unleashed...
13
187
1,232
28,374
Timothy Bates retweeted
Robert Plomin — the most influential living researcher on the genetics of intelligence — in ICA from day one.
Polygenic Score Prediction Within and Between Sibling Pairs. Lin et al. in Intelligence and Cognitive Abilities doi.org/10.65550/001c.140654
6
27
1,605
Timothy Bates retweeted
Call for papers: "AI and Intelligence", a special issue of Intelligence and Cognitive Abilities guest-edited by Emily A. Willoughby.
Empirical and conceptual submissions welcome. Open now, read more here:
icajournal.scholasticahq.com…
4
5
332
The method reduces bias (who took the test, what it contains, etc.)
If "IQ tests are biased and reflect inequality", reducing bias and improving stratification control should shrink DNA effects, possibly to zero.
Instead, the exact opposite happens. Reduced bias => MORE genetic causes of cognitive ability transmitted across generations are revealed!
1
2
12
8,724
Known common DNA regions linked to IQ just passed 500.
In UK Biobank, people have only a brief reasoning test, or none. But cognitive ability is embedded throughout our lives: what we eat, how safely we live, books we read, what we do for pleasure….
Imputing those missing scores from health, behaviour and other measures cut ascertainment bias and raised N from ~270,000 to >455,000.
That lifted independent common loci from 390 to 550 (>40%). The rare-variant yield rose further: genes linked to lower scores went from 5 to 26, including 14 causing neurodevelopmental conditions.
And the polygenic prediction score now approaches the limits of test reliability.
The DNA signal is present from conception and transmitted across generations (cf. @gclarkecon).
More strong work from @DMalawsky and @hilsomartin: technical innovation and scale.
nature.com/articles/s41588-0…
4
13
109
14,752