Mark Thornton

dlvr.itMotor abstraction training generalizes to the refinement of specific movement patternsNature Human Behaviour, Published online: 19 October 2026; doi:10.1038/s41562-026-02597-zPracticing unfamiliar characters in rotated orientations improved upright writing as much as direct practice, showing that the motor system learns by abstracting a movement’s structure, not just its details.

Excited to share this new preprint from @lindseytepfer.bsky.social, @chujunlin.bsky.social, and me, in which we investigate the different ways our brains allow us to change our minds about people! "The neurocomputational mechanisms of naturalistic trait impression updating" osf.io/preprints/ps...

Mark Thornton reposted

Kathy Garcia

@gkathy.bsky.social

23h

What dimensions organize how we see social interactions? We learn them from 53,000+ human judgments of naturalistic social clips. They explain 2× the variance of pre-labeled annotations & better predict visual cortex, EBA, and pSTS New preprint w @lisik.bsky.social 🧵 osf.io/preprints/ps...

Mark Thornton reposted

Sam McDougle

@actlab.bsky.social

Proud to have this paper out today at Nature Human Behaviour @nathumbehav.nature.com ! Heroic, creative work by Zekun Sun (w/ Zhiran Xie and me).

Nature Human Behaviour@nathumbehav.nature.com

Motor abstraction training generalizes to the refinement of specific movement patterns

Mark Thornton reposted

cambridge.orgMetabolic considerations for cognitive modeling | Behavioral and Brain Sciences | Cambridge CoreMetabolic considerations for cognitive modeling - Volume 49

We're excited to announce that the #SANS2027 Presidential Keynote Speaker will be Dr. Rebecca Saxe (@rebeccasaxe.bsky.social)!

Mark Thornton reposted

David Colaço

@davidcolaco.bsky.social

The complete version of @phaueis.eurosky.social and my BBS target article, "Metabolic considerations for cognitive modeling," is now available. It includes 32 commentaries and our response. Please share #philsci #philsky #cogsci

Mark Thornton reposted

Exciting news! SANS has received an R13 from the National Institute for Aging at NIH! It will provide support for trainee travel and programming focused on aging at the SANS Annual Meeting as detailed below:

What would you want to see in a textbook (free, online, jupyter/marimo style) about quantitative/computational approaches for analyzing naturalistic social interactions (e.g., face-to-face conversations)? I'm in the early phases of drafting one and would value external input on its audience's needs.

Mark Thornton reposted

Rob Chavez

@robchavez.bsky.social

Our latest paper led by Taylor Guthrie with former undergrad RA Aussie Frost. We show that multivariate fMRI signals can decode the identity of specific individuals within the DMN and in the absence of visual features or other lower-level sensory information. link.springer.com/article/10.3...

There are still 3 months left in 2026, and I have already reviewed significantly more papers than in any past calendar year. Yet simultaneously I have had to decline more review requests than ever. Not saying it's all AI's fault, but I have encountered some very blatant and counterproductive uses.

Jaclyn A. Siegel, PhD@jacasiegel.bsky.social

If you use AI to do your research, you really cannot complain when reviewers use AI to review your papers, imo. The processes you use to expedite your workflow create more work for the rest of us, who are under the same constraints to maximize our own output. Plus, peer review is invisible labor.

I don't, and won't, use AI to review papers for me - for both principled and practical reasons. But ultimately that means my capacity is limited, and I will have to be more selective in the reviews I agree to do as volume increases.

I think the option to look at the full manuscript before agreeing to review is increasingly important. Some journals already offer this, the rest should start. If I can tell from a glance that a paper is chock full of slop, I'm going to say no. Human reviews should be reserved for human authors.

Mark Thornton reposted

wimpouw

@wimpouw.bsky.social

21 Sept

Here we go! We can finally share what Sharjeel Shaikh @sharjeelshaikh has been leading the work on with us! Consider participating in our detection challenge

sharjeelshaikh.bsky.social@sharjeelshaikh.bsky.social

Gesture Challenge 2026 is open. envisionbox.org/gesture_chal... @babajideowoyele.bsky.social @wimpouw.bsky.social @jamestrujillo.bsky.social @esamghaleb.bsky.social @gdemelo.bsky.social

Mark Thornton reposted

socialaffectiveneuro.orgSymposia Submission - Social Affective Neuroscience Society

31 Aug

Symposium submissions are now open for #SANS2027! Submissions must be made through the online submission portal by 23:59 (Eastern Time) on Tuesday, October 13, 2026. More details here: socialaffectiveneuro.org/symposia-sub...

Mark Thornton reposted

Christopher Honey

@chrishoney.bsky.social

16 Sept

1/ How do we sustain mental continuity in the face of interruption? @xianl-explorer.bsky.social found that, while people listen to a story, posterior medial cortex expresses a slowly evolving "background" representation of the story's context, which survived through silence and interfering tasks.

14 Sept

Yes sorry for the lack of clarity if you're outside the US! I usually specify the data source as the psych job wiki in the image itself (e.g. bsky.app/profile/mark...) but I realize now, while putting this together in a hurry, I accidently bumped that off screen while adjusting the legend this time.

Eiko Fried@eikofried.bsky.social

Quick note about this valuable effort (thanks Mark) that at least the 2026 numbers on the "psych job wiki" are nearly 100% US jobs. So this may say a lot about the US (or not, depending if the jobs had equal selection bias in prior years), but not necessarily about psych jobs elsewhere, eg in Europe

biorxiv.orgNeural tracking of surprisal and semantic distance in naturalistic movie viewingUnderstanding speech requires listeners to integrate incoming input with prior linguistic and thematic knowledge to access meaning, a task greatly aided by prediction. Surprisal and related phenomena ...

14 Sept

I hope these data can still be useful to you - even if you're not from the US or Canada and not on the US/Canadian market - both because I haven't seen data like these complied from other markets, and because the US is a very large market that affects the global market.

Mark Thornton reposted

Jonathan Peelle

@jpeelle.bsky.social

13 Sept

"I like milk and sugar in my _______" Classic work on prediction has used violation paradigms to study context and prediction. New work from my lab, led by Ryan O'Leary, compares semantic distance and GPT-2 surprisal in natural speech. #neuroscienceoflanguage www.biorxiv.org/content/10.6...

11 Sept

The psychology job market may be showing some early signs of regaining strength, relative to last year. A week ago last year, fewer than 200 jobs had been listed - today, it is nearly 300, a figure we didn't hit till early October in '25. #psychjobs #neurojobs #academicjobs

11 Sept

Only time will tell if this is a genuine increase over last year, or if the market is just earlier this year than last year. Last year's market was by all accounts substantially delayed relative to the usual timeline, with many more jobs appearing well into late winter / early spring.

10 Sept

Come join @elisabaek.bsky.social & me on the SANS social media committee!

SANS@sansmeeting.bsky.social

Interested in becoming a social and affective neuroscience influencer? SANS will soon be launching a new media initiative, in which we make short-form video content to reach a wider audience. If you're interested in joining this initative, please DM us here or email sans.media.chair@gmail.com!

Mark Thornton reposted

10 Sept

Mark Thornton reposted

jerrytang.bsky.social

@jerrytang.bsky.social

Our new study on cross-participant cortical mapping (with @alexanderhuth.bsky.social) is out in @imagingneurosci.bsky.social! direct.mit.edu/imag/article... 1/7

Mark Thornton reposted

Context Lab

@contextlab.bsky.social

@lindseytepfer.bsky.social

In our latest study, led by @paxt0n4.bsky.social, we build a geometric framework to ask how our memories change with the passage of time ⏰ Title: Delayed recountings preserve but simplify the semantic geometry of earlier recountings Preprint: osf.io/mhxtd_v1 Code/data: github.com/ContextLab/m... 🧵

osf.ioOSF

Mark Thornton reposted

We don’t just lock on to our first impression of someone and call it a day. Instead, we regularly revise our appraisals as we learn more about them. But *how* do our brains make this process possible? Check out this fresh pre-print to see how we went about answering this question!

Mark Thornton@markthornton.bsky.social

Whether we're forming impressions of politicians, celebrities, or friends, our ability to change our minds about other people, given appropriate new information, is a critical social cognitive and moral capacity. This work brings us closer to understanding how our brains make it possible.

Data & code avaliable here: osf.io/pnyut/overview

osf.ioOSF

They also show that there is more to explain! Because even together the mechanisms we investigate here only partially mediated the relationship between neural reponses and trait ratings.

Neuroscientifically, these results also highlight the importance of the superior temporal sulcus in trait impression updating. This region was implicated by all of the putative mechanisms we found evidence for, and is also the main place where part of the trait-brain assocation remains unexplained.

We also found that all four of sets of individual differences predicted intersubject correlation in brain responses. However, only one of these set of differences - participants' explicit attitudes towards age, race, and gender social groups - statistically mediated the trait-brain relationship.

Together, these results show how multiple mechanism - perceptual, computational, and individual difference - contribute to shape how we change our minds about other people when presented with rich naturalistic stimuli.

We also found evidence that 3/4 of the computational operations we investigated (all but prediction) generated representations that uniquely explained brain activity over and above the perceptual modalities. However only one, compression, uniquely statistically mediated the trait-brain relationship.

At the individual difference level, we found several regions within which differences in intersubject correlation were correlated with individual differences in trait updating.

Consistent with other recent investigations of trait representations/updating that have used naturalistic stimuli, we found activity correlated with traits throughout much of the brain, but particularly concentrated in the superior temporal sulcus.

Unsurprisingly, all three perceptual modalities we investigated explained unique brain activity. However two modalities - face and semantics - went beyond that to unique statistically mediate the relationship between trait ratings and brain activity.

To this end, we performed two forms of representational similarity analysis (RSA): spatiotemporal cross-stimulus (ST-RSA) to model the continuous features, and intersubject (IS-RSA) to model the individual differences.

In both cases, we were looking to see not only where the traits/features/individual differences were reflected in brain activity - we also wanted to see where they overlapped not just in terms of space, but in terms of shared neural variance within the same parcels of the brain.

Using these annotations, we also trained deep neural networks (DNNs) as cognitive models to instatiate differerent algorithmic operations (compression, prediction, multimodal fusion & temporal integration), generating a second set of representations learned to perform each of these operations.

We modeled participants' brain activity while watching the SEND videos using the trait ratings, perceptual modality annotations, DNN features, and individual different measures (demographics, traits, mood, & social group attitudes) collected from the scanner sample.

Building on our earlier behavioral work (osf.io/preprints/ps...), we used short videos of people narrating real life events to camera (from the SEND: github.com/StanfordSoci...) to elicit brain activity from scanner participants, and continuous trait ratings from online participants.