One of the main things neuroscientists do not like to deal with is degeneracy and the irreversibility of the activity -> function relationship, because that would force them to acknowledge that most analysis methods like decoding and functional connectivity are meaningless.

🧠👀 'These findings reveal high-dimensional aspects of cortical representation undetectable with conventional methods, such as RSA, & contradict previous theories suggesting that high-level visual cortex representations are low-dimensional.' #neuroskyence journals.plos.org/ploscompbiol...

Please check out and RT this new position for an exciting collaborative project with @boninlab.bsky.social and @franckp.bsky.social on the links between human-specific mechanisms and neurodevelopmental disorders in cortical circuits.

A structurally profound update to the most comprehensive closed-form model of working memory & RL-like interactions. Really excited to see this evolution - and to learn if this new model might tell us something different about experimental drugs tested with the older task+model. Congrats Anne

Very proud to have played a part in this work. I am confident that measurements of speech contain signal of use for developing new therapeutics, transdiagnostically...

Results so stunningly clear they inspired this classic xkcd (xkcd.com/2400/):

New paper using M/EEG to look at attractive and repulsive serial dependence in working memory, led by the excellent Jiangang Shan, with Jasper Hajonides.

Proud of Amy and Nick, who review technical and theoretical gaps in timescales of dopamine fluctuations. We argue that DA fluctuations are hierarchical control signals, and “tonic” DA relays goal alignment across planning horizons and C-BG hierarchy. www.sciencedirect.com/science/arti...

This is an encouraging result for the use of EEG in a variety of applications going well beyond RSA.

EEG representational analyses are more robust than fMRI. Huh. link.springer.com/article/10.1...

I've updated my literature review of studies of aperiodic neural activity in clinical disorders, adding ~30 papers, taking it to 177 reports across 38 disorders! It's got a review of results so far, discussion of themes & issues, & recommendations for future work! www.medrxiv.org/content/10.1...

1/ New paper by Hame Park, (@AraziAyelet), Bharath Talluri, Marco Celotto, Stefano Panzeri, Alan Stocker & Tobias Donner published in Nature Communications – “Confirmation Bias through Selective Readout of Information Encoded in Human Parietal Cortex”: rdcu.be/etlR7. Here is a summary:

Super talented team working on an incredibly important problem. Congrats!

Denali describing their approach to shuttle large molecules as cargo across the blood brain barrier. We should see new clinical data from approaches like this already in 2025. Exciting time for neurotherapeutics for this and so many other reasons...

Dual targeting of transferrin receptor and CD98hc enhances brain exposure of large molecules https://www.biorxiv.org/content/10.1101/2025.03.24.645085v1

Intellectual ability and cortical homotopy development in children and adolescents https://www.biorxiv.org/content/10.1101/2025.03.24.645014v1

Here is the video I made, inspired by this wonderful tweet. www.youtube.com/watch?v=dSOj...

How did these authors, reviewers and editors all entirely fail to note the fascinating discrepancy beween this finding and the protective effect of lithium for AD? bmcmedicine.biomedcentral.com/articles/10.... There's such an interesting discussion to be had on that! (1/n)

Incredibly talented group and super important work from these teams at Boston Children's - great training opportunity for new PhDs thinking about a postdoc.

As @dereklowe.bsky.social was asking, we need more pharma CEOs to speak up! Merck, Pfizer, Lilly, Amgen - where are your voices?! (BTW- really read to the bottom of this).

The Democrats on the House Science Committee have set up a website to collect stories from fired federal employees, anonymously if desired. Please amplify. (This helps the lawyers establish standing for bringing legal cases against the administration!) democrats-science.house.gov/sciencefirings

Will shadish had a great paper on this where participants were randomized to choose their condition or to be randomly assigned to condition www.statpower.net/Content/MLRM...

New guest post! Excluding "careless" responses is a routine practice in many contexts. But if careless responding itself is an outcome of the variables of interest, that may actually cause trouble.

@briannosek.bsky.social I'm not sure if you're aware but your work on the reproducibility crisis is being cited by the Heritage Foundation to motivate the huge cuts to indirect costs we are seeing at NIH. How do you respond to this? www.heritage.org/sites/defaul...

New preprint: elaborating the domain-general prediction error driven account of paranoia and delusion contents: osf.io/preprints/ps... Awesome to work with @praveensuthaharan.bsky.social Rosa Rossi-Goldthorpe, @juliasheffield.bsky.social, Santiago Castiello, and @celiaheyes.bsky.social on this

More killer stuff from the Roche EEG team. Wow

Super excited to share this paper that just came out today [email protected] led by @sarahbanker.bsky.social, where we show important phenotypic divergence between online subjects who self reported high autism traits and those formally diagnosed by clinicians.

We are very excited to announce the open release of a new 170 participant dataset: Linked MEG, TMS, multimodal 3T MRI, 7T MRI, Connectom MRI, cognitive /questionnaire data. Immense effort from a fabulous CUBRIC team. Paper by Carolyn McNabb et al: doi.org/10.1038/s415...

Null findings are important ‼️ Excited to share my latest work finding model-based planning does NOT improve following ketamine 💉, antidepressant 💊, or iCBT 👩‍💻 treatments 🧵👇🏻

I’m excited to share our new, open-access paper published today in NatMentHealth, “Latent mechanisms of language disorganization relate to specific dimensions of psychopathology”. www.nature.com/articles/s44...

Great new work on the link between thoughts and psychological traits! #neuroskyence #psychscisky

Interesting study on the computational role of prior probability in perceptual choices, finding that priors change the accumulation rate of evidence rather than the baseline. Nicely in line with work by @clarepress.bsky.social & @danieljamesyon.bsky.social: pmc.ncbi.nlm.nih.gov/articles/PMC... 🧠🤖 🧠🟦

Entrainment explains fascinating phenomena like fireflies flashing in sync—but can the brain be 'entrained' by rhythmic stimuli? With @maelanmenetrey.bsky.social, we tested key features of neural entrainment, suggesting specialized tuning mechanisms. #EEG #neuroskyence doi.org/10.1186/s129...

An update on this #scaling matter, for the curious ones. The connectivity from 256k users still confirms the non-power-law shape of the (observed) network (left). Over time, the mean connectivity decreases (right), but more robust analysis is needed to understand biases & mechanisms. 1/

Wish I was attending psychonomics to see this. Intraindividual variation is such an exciting, potentially important, and too often neglected dimension of individual differences.

New findings: We track visual sequences yet the brain's process is unclear. We discovered the sequence-tracking area in monkey's lateral prefrontal cortex is distinct from an adjacent subregion that reacts to other stimulus changes. #neuroscience 🧠 🧪 👩‍🔬

One of the challenges I saw with federated learning was that there’s not much incentive for each individual node to undertake the administrative burden of joining the network. One interesting incentive that I haven’t seen acknowledged is that the individual nodes are in a privileged position to see…

For all my friends who just got here to BlueSky and want some cool people to follow, I've now delved into the wonderful world of Starter Packs (full list at blueskydirectory.com/starter-pack...). Here's a thread of some of the cool ones I found!

NREM sleep improves behavioral performance by desynchronizing cortical circuits. Ed summary: "positive effects of sleep replicated by low-frequency intracortical electrical stim. Computational modeling suggested ... an asymmetric decrease in synaptic conductances" www.science.org/doi/10.1126/...

Strikingly different neurotransmitter release strategies in dopaminergic subclasses https://www.biorxiv.org/content/10.1101/2024.11.20.622499v1

Psychiatry has the most black box warnings of any area. Could the others backfire, too? Deep respect for the regulators responsible for these complex decisions. *Especially* now, better health vigilance systems are needed. Can’t wait 20y to discover the next counterintuitive effect of policy changes

Common genetic variants contribute more to rare diseases than previously thought www.nature.com/articles/d41... (note that the converse is also true!!!)

Scaling Laws for Precision yes, llama models are harder to quantize. They’re “overtrained”, on more data, so quantization removes a lot of critical information arxiv.org/abs/2411.04330

This article gives a phenomenal overview of the history and evolution of best practices in measurement invariance and differential item functioning -- I can't remember the last time I learned so much reading a single paper, highly recommended! www.tandfonline.com/doi/full/10....

Amazing work. A total speculation: superior predictive validity of self-report vs task perf metrics (accuracy, RT, etc) may arise from analogous dissociations btwn the lower-dim probes of '1st-order' task performance metrics, vs higher-dim / '2nd-order' judgements ("How *DO* I feel [wrt X_1...N]?")

AmygdalaGo-BOLT3D: A boundary learning transformer for tracing human amygdala https://www.biorxiv.org/content/10.1101/2024.08.11.607487v1

If I might interject … yes there’s a literature on some relevant factors in scams (eg overlap with reflective thinking) doi.org/10.4018/IJCB... doi.org/10.1371/jour... ucilnica.fri.uni-lj.si/pluginfile.p... And increasingly more recent work in this area (too much for one post!)

The UK launched a metascience unit. Will other countries follow suit? Tasked with finding better ways to conduct, distribute and fund research, the unit could set the standard for government-led ‘science of science’ initiatives. www.nature.com/articles/d41...

Has anyone got a large (>=7b) GPT running on a neuromorphic chip? Who’s working on this & how close is this? Inference time scaling laws have me suddenly fascinated by physical computing hardware & where it’s gone since I last tracked that area (longer ago than I care to admit).

In addition to this thoughtful piece in eLife, it’s nice to see a note from the Editor in Chief of Science espousing on the topic today as well: www.science.org/doi/10.1126/... There is incredibly valuable work to be done in a participatory, partnered mode. Stay tuned for more on this from me soon