Study preregistration: an early example and analysis pubmed.ncbi.nlm.nih.gov/42222493/

"Academics Should Use AI for Writing", says E. J Wagenmakers www.bayesianspectacles.org/why-academic... But he's not saying you should get AI to write *for* you. He's saying you should use AI as a proofreader. "I am meticulous with my writing but ChatGPT had found numerous mistakes."

"Neuroscience in Quantum Marketing" gopalkatoch.com/what-is-quan... What in the AI generated slop is this?!

Monosynaptic connections link functionally similar regions in human cortex www.biorxiv.org/content/10.6... This seems very important! Full of surprises, but here's one: " resting-state fMRI connectivity of a site was only weakly correlated with the stimulation-evoked fMRI connectivity" #fmri

How bumblebees cope without a queen "Queen-less colonies... contained a subset of highly influential workers with elevated movement, spatial centrality, and reproductive activity" www.biorxiv.org/content/10.1... Bumblebees have unlocked government: "Oligarchic Republic"

"Considering children - a population known to be more cognitively flexible than adults - as a proxy psychedelic group..." pubmed.ncbi.nlm.nih.gov/42190161/ Hmm

The psychology of 'simps': "The fear of being single contributes significantly to the motivation behind male simping behaviors." pubmed.ncbi.nlm.nih.gov/42178873/ 🫠

"Forget About Looksmaxxing. Brainmaxxing Is the Real Rage." Can we please just stop maxxing

I think they call this Phineas Gage-core

"While machine learning can classify sex from brain activation, volume, or behavior, these data types provide orthogonal information" www.nature.com/articles/s41... Hmm, interesting. There are sex differences in brain and in behaviour but those differences are not correlated.

What would happen if AI actually became significantly more intelligent than us? How would we know? Let's say an AI generates "a proof" of one of the big mathematics problems - the Riemann hypothesis or P=NP. The AI says it's a proof, but human mathematicians can't understand it. What do we do?

It's been a big week for AI-in-science. Nature published a number of major papers on agentic AI scientists (www.nature.com/articles/d41...), and OpenAI announced that an AI had solved a significant mathematical problem www.scientificamerican.com/article/ai-j...

"The brains are already almost devoid of the coordinated neural firing necessary even for minimal consciousness": disembodied human brains used for drug testing www.science.org/content/arti... Would you volunteer for this? 🤔

Does anyone else find this stock image deeply unsettling? stock.adobe.com/uk/search?k=... It's certainly creative though!

Children acquire object category representations from their everyday experiences in the first few years of life. What do the inputs to this learning process actually look like? New preprint! arxiv.org/abs/2605.14990

More thoughts on DINO, a self-supervised vision model which I believe should be of high interest to cognitive scientists and neuroscientists. DINO isn't new - it was introduced in 2021 which makes it ancient as far as deep learning goes - but I think it's underappreciated. bsky.app/profile/neur...

A new view of depression as decisional reference point pathology: "Elevation of the reference point would lead a person to experience once pleasurable activities as negative" www.pnas.org/doi/10.1073/... When 🙂becomes 🙁 because your reference point is stuck at 😁

‘Comically bad’ datasets used to train clinical models for stroke and diabetes retractionwatch.com/2026/05/18/k... Some datasets should be seen by a statistician, others should be seen by a garbageman

Enhanced processing of cartoons in infant visual cortex www.biorxiv.org/content/10.6... This fMRI study compared the original Lion King to the "realistic" remake!

"Zed has experienced lifelong, dynamic facial distortions in which features move, droop, disappear, expand, shrink, and rotate." www.biorxiv.org/content/10.6...

What a stimulus predicts, not what it depicts, determines striatal reward signals www.biorxiv.org/content/10.6... This is interesting. nAcc responds most strongly to stimuli that predict reward, but vmPFC responds more to inherently pleasant (erotic 😳) stimuli.

We are implementing a similar policy at @arxiv.bsky.social. If there is incontrovertible evidence of LLM slop in a paper, this means the authors did not take the time to read the LLM output and we can't trust anything else in the paper. Penalty is 1 year ban from arXiv followed by...

"Health companies co-opt identity politics and social movements in their marketing, transforming marginalized communities into profitable niche markets." pubmed.ncbi.nlm.nih.gov/42139867/ Neurodiversity will get commercialized and exploited soon. It's already started, but it will get much worse

Likely. All plausible hypotheses. We’re now officially discussing within the Lakatosian program of neuroconnectionism as nicely coined by @adriendoerig.bsky.social et al. www.nature.com/articles/s41...

so yes, there's A LOT that could be learned in a self-supervised manner. In fact there's overviews claiming this is what happens in the brain; see www.nature.com/articles/s41... / www.biorxiv.org/content/10.6... Computational plausibility? Yes. Algorithmic/implementational plausibility? Idk.

Yesterday I argued that cognitive "modules" could be seen as deep nonlinear networks, which could interact via linear (or linear-ish) mappings between their outputs, because deep networks can output rich representations suitable for linear decoding. bsky.app/profile/neur... But you might ask...

Just in case you, or those reading this thread, are not aware of it: i think these ideas map very well to the concept of communication subspaces pubmed.ncbi.nlm.nih.gov/30770252/

Here's some quick thoughts on something I've been thinking about for a while: One of the books that most influenced me about cognitive science was The Modularity of Mind by Jerry Fodor (1983). (1/N)

Procrastination, productivity and inspiration: how research is like designing video games www.nature.com/articles/d41...

Amplifying this thread esp the convo b/n @micahgallen.com and @jtheriault.bsky.social for those who say science engagement on Bluesky is dead

Neural network models can perform, or at least simulate, complex cognitive functions despite having no relation to allostasis. You could however put a neural network in the service of allostasis. They are a tool. To me that's what the brain is: a set of neural networks, serving allostasis.

Some further thoughts on "It's not the thought that counts": It's important to distinguish "thinking" and "cognition". Brains do cognition - this is true of animals as well as humans, as all animals with brains can learn and make decisions. Thinking, on the other hand, is probably uniquely human.

"The Brain is a Prediction Machine: When Leaders Leave Gaps, the Brain Fills Them In" www.blanchard.com/blog/the-bra... I never thought we'd get to the point where the Bayesian Brain was turned into management-speak but here we are!

A psychoanalytic interpretation of neurobiological reductionism: the idea that mental life is reducible to physical brain states as "a defense against anxieties associated with relational ambivalence and the unknown" www.tandfonline.com/doi/full/10.... That's a big hmmm from me.

Thanks to the internet, anyone can now review anything, and it turns out that we think famous historical wonders are pretty good, but far from perfect. Angkor Wat? Sistine Chapel? 4.7⭐/5. (I know that the review partly reflects the tourist experience rather than the 'wonder itself')

"The author argues that psychotherapy's moral panic about AI reflects a splitting of the human from the technological that mirrors colonialist Subject/Other dualisms." pubmed.ncbi.nlm.nih.gov/42105337/ Ah, so I think the author is doing a little light trolling.

this is a fabulous example of the "picture of a description" issue with generated imagery. You can pretty easily back out the prompt that was used to generate this and the actual image provides you absolutely zero information over and above what just reading the prompt would have.

This is an AI generated image to illustrate the idea that "Scientists Discover The Brain Starts Life 'Too Full' And Then Trims Itself Down" www.ndtv.com/science/scie... The paper in question was in mice! But even if it weren't, this offers zero insight.

"Up to 46% of neurons in prenatal cortex showed aberrant genomes"! These mutant neurons tend to die out during development, so they end up much less common, but some do remain www.biorxiv.org/content/10.1... This is really fascinating - individual neurons can be genetically very different!

"Negative experiences with supervisors were among the most frequently reported reasons for leaving among former researchers (48%)" www.biorxiv.org/content/10.6... Well, the problem is that people don't become scientific supervisors because they'd be good at being scientific supervisors...

"The common conception that the brain is primarily for thinking, or other cognitive processes, is potentially misleading... [the brain's actual function is] allostasis: the predictive regulation of competing demands from internal bodily systems" www.cell.com/neuron/fullt... hmm.

So I came across a nootropic supplement called MindLabPro, which is billed as "The world's first Universal Nootropic™" uk.mindlabpro.com Given that Universal Nootropic™ is a trademark™, it's not surprising that it's the first Universal Nootropic™, since no one else can legally use that term...

There was a recent paper arguing a similar point regarding representational structures and Unwelten: akoepke.github.io/cave_umwelten/

"Major AI models converge as they model reality better... because there's only one reality to model!" towardsdatascience.com/how-major-re... Well, we are only training our AIs on one "reality", which is the human reality (more specifically the 21st century internet user reality 99% of the time...)

How much for a fake authorship? Ad database reveals secrets of scientific fraud www.nature.com/articles/d41... I'm selling co-authorship on my Bluesky posts. Bluesky only allows single authors at present but if you pay me $1000 I will privately honor you as a co-contributor in spirit.

What Does 'Human-Centred AI' Mean?, asks @olivia.science pubmed.ncbi.nlm.nih.gov/42073945/ Good question. I think it's meaningless, but maybe reassuring. Some therapists like to say they are "client-centered" or "patient-centered"... it has no real meaning but it's a signal about their values.

If you got to rename a brain area, what would you rename? I'm not a fan of the word "cerebellum". The little brain? Not very creative, plus it has more neurons than the rest put together! I'm not sure what I'd call it, though...

Scientific Objectivity and the Pursuit of the Public Good pubmed.ncbi.nlm.nih.gov/42068410/ To me scientists, like all citizens, have a duty to advocate for what they think is right. This is not a scientific duty, it's a more general one from which scientists are not exempt.

Made-up non-words produce semantic brain activity "comparable to that of existing words, provided that novel words contain a concatenation of identifiable morphemes." www.biorxiv.org/content/10.6... This is a bluty result, but not too jarfish if you doot ahugeapun, no?

Technological folie à deux: feedback loops between AI chatbots and mental health www.nature.com/articles/s44... This is a real concern. I wonder if it's truly a new concern, though...