I definitely have pretty marked color deficit (protanomaly), very noticeable when I compare my color judgments with my partner, yet I still got 0.006 ?

Whoops forgot the relevant feeds #PsychSciSky #neuroskyence

A comment on Piray’s recent paper on low statistical power in computational modelling studies:

Are computational modelling studies severely underpowered?

I don’t think the current analyses justify that conclusion

Thread + commentary 👇

Yes, that analysis is discussed in the commentary. My concern is about the claim that it corresponds to a meaningful “medium” effect size in the context of model comparison. I don’t think the analogy to Cohen’s benchmark is well justified here.

Without explicit effect-size assumptions, power simulations are hard to interpret. And if those assumptions are poorly chosen, the resulting power estimates can be highly misleading.

Commentary here: osf.io/preprints/ps...

My point is not that the studies reviewed by Piray are well powered — they may well be underpowered.
The point is that power analysis always depends on assumptions about how strongly one model is favoured in the population, and those assumptions need to be explicit.

Chart showing that power is low (around 0.5) under Piray’s method, but remains above 0.9 when the empirically estimated population-level preference is attenuated by up to 50%.

I applied this to 3 studies from Piray’s review (selected only due to availability of empirical Dirichlet params in public materials). Power remained high under substantial attenuation, only dropped to the low levels estimated by Piray under 90% attenuation of the empirical population preferences

To illustrate this in this short commentary I propose an approach for sensitivity analysis: start from the empirical population model frequencies, then progressively attenuate them toward a uniform Dirichlet.

This is an interesting paper on power in computational modelling, with several useful points.

But I think the headline conclusion — that 60–80% of studies are underpowered — rests on a very specific effect-size assumption that is hard to justify and unlikely to be realistic in most cases.

1/5

human-AI misalignment in the wild

common knowledge among italians :) and in most cases, it’s better to drain the pasta even earlier than that and finish it in the pan with the sauce and some pasta water

yes but the correlation really needs to be -1 for all lines to cross in precisely the same spot, which would indicate some issues with the model

it can happen if don't have by-participant random slopes for the effect of `variable` and the fixed-effect coefficient is ≈ -1

WARN-D machine learning competition is live » Eiko Fried If you share one single thing of our team in 2026—on social media or per email with your colleagues—please let it be this machine learning competition. It was half a decade of work to get here, especi...

After 5 years of data collection, our WARN-D machine learning competition to forecast depression onset is now LIVE! We hope many of you will participate—we have incredibly rich data.

If you share a single thing of my lab this year, please make it this competition.

eiko-fried.com/warn-d-machi...

One day something will happen — a wedding, a graduation, a job promotion, Jake Paul getting knocked out, Andrew Tate getting knocked out the next day — and you’ll want the right outfit to celebrate. Buy that special outfit now, so you’re ready when that next unexpected moment arrives.

Ongoing decimation of British universities part 252:

Apx. 1000 academic staff at University of Essex just received formal ‘risk of redundancy’ letters via email.

Please share @ucuessex.bsky.social @ucu.org.uk

A diagram showing that easily encoded context cues can be stored in memory and modulate the encoding of hard-to-recognise images.

@orisshenyan.bsky.social and me are happy to share our most recent preprint - we find that prior knowledge protects high-level perception from low-level visual deficits common in healthy ageing.

@tessamdekker.bsky.social @mlisi.bsky.social

www.biorxiv.org/content/10.6...

can't stop watching this clip of a tesla Optimus teleoperator taking his headset off before properly logging out the robot

It's official: 400 redundancies and a campus closure. At-risk staff will get a letter next week. Absolutely no serious consideration of any of the union's counterproposals.

Please share widely.

❗️Our next workshop will be on Nov 13 6 pm CET titled Building & Customising Stat Models with Stan and R: Introduction to Bayesian Inference by @mlisi.bsky.social

Register or sponsor a student by donating to support Ukraine!
Details: bit.ly/3wBeY4S
Please share!
#AcademicSky #EconSky #RStats

We are hiring, come work with us!

Please share widely 🙏 and do get in touch if you have questions about the Department

Deadline: December 1st

#neurojob #psychscisky #neuroskyence #VisionScience #academicsky

Europa Ineffable on X: "In a recent panel at what looks like Labour conference, Sunder Katwala, wearing an England shirt, appeared to compare the R🦍 GANGS to a football match: "it's a game of two halves" 🥴 https://t.co/b5YNKyaQtu" / X In a recent panel at what looks like Labour conference, Sunder Katwala, wearing an England shirt, appeared to compare the R🦍 GANGS to a football match

A very worrying escalation of online harassment.

Somebody has used a video of me saying "racism in football is a game of two halves" + put a fake voiceover on it saying "what I would say to the girls of Rotherham is it is a game of two halves"

x.com/EuropaIneffa...

Dariia Mykhailyshyna - Workshops for Ukraine Feedback on the past workshops (if you want to learn how to make wordclouds, check out Text Data Analysis workshop below)

I’ll be giving a 'Workshops for Ukraine' session on Building and Customising Statistical Models with Stan and R: An Introduction to Bayesian Inference — online on Nov 13.

Open to all, with donations supporting Ukrainian organisations.

👉 sites.google.com/view/dariia-...

#stats #rstats #statssky

Plot showing unemployment rates in college grads and people without a degree, showing that when you compare groups based on when they entered the job market, people with a degree have substantially less unemployment rates which display slower increase in 2025.

Turns out a university degree does still make a substantial difference to job prospects, once you account for the conditions when people enter the job market. Nice analysis by the FT.

on.ft.com/4q6Y8pm

A statement from the President of the Royal Society | Royal Society A statement from Adrian Smith, President of the Royal Society, regarding threats to the values that allow science to flourish.

Deeply disappointing. Once again the @royalsociety.org has avoided explaining how *actions* by Elon Musk that clearly repudiate the Society's values are consistent with the code of conduct that Fellows must adhere to. royalsociety.org/news/2025/10...

Amateur view. AI bubble shows that there is too much money chasing too few opportunities for genuine innovation. The solution is to take a medium/long view and put much more funding into basic science in universities rather than cutting as is happening now. (But I would say that wouldn’t I?)

❗️Our next workshop will be on October 2nd, 6 pm CEST, on Effective and Useful Feature engineering by @emilhvitfeldt.bsky.social

Register or sponsor a student by donating to support Ukraine!
Details: bit.ly/3wBeY4S
Please share!
#AcademicSky #EconSky #RStats

Research Assistant at UCL Recruiting now: Research Assistant on jobs.ac.uk. Click for details and explore more academic job opportunities on the top job board

We’re hiring! We’re looking for two RAs to study neuroplasticity in sight loss, sight rescue and development in children and adults @ucl.ac.uk using a wide range of neuroimaging and behavioral methods. Please help spread the word! Apply by 16 Oct! t.ly/q3aYe #neurojobs #NeuroSkyence

Biggest record of hope against hate Add your name to show you’re with us >>

All sign this please @hopenothate.org.uk secure.hopenothate.org.uk/page/178037/...

Research Assistant at UCL An academic position as a Research Assistant is being advertised on jobs.ac.uk. Click now to find more details and explore additional academic job opportunities.

We’re hiring! We’re looking for two RAs to work with @tessamdekker.bsky.social and @ronimaimon.bsky.social on visual neuroscience experiments with children and adults, both with and without eye disease using cutting-edge neuroimaging and behavioral methods. Apply by 16 Oct! t.ly/q3aYe #neurojobs

New paper out! 📣

Syrian refugee children (4-9 yrs) show age-related shifts in judging trustworthiness from faces — older kids tend to see faces as less trustworthy; yet judgments are not tied to child or maternal mental health.

doi.org/10.1017/S095...

#PsychSciSky
@rhulpsychology.bsky.social