Sign in

Robert Hawkins

@rdhawkins.bsky.social
4.2K followers 3.5K following 313 posts

asst prof @Stanford linguistics | director of social interaction lab 🌱 | bluskies about computational cognitive science & language

PostsRepliesMedia
Robert Hawkins @rdhawkins.bsky.social · 8h
www.wired.com/story/experi...
wired.com
Experience What It’s Like to Travel in the Occupied West Bank
Traverse barriers and checkpoints in the occupied West Bank—your choices reflect the reality of nearly 3.5 million Palestinians who live there.
030
Reposted by Robert Hawkins
Jialin Zhou @jialin-lin-zhou.bsky.social · 02/10/2026
🆕 New preprint: Is “person = man” inevitable? In the matrilineal Mosuo, men and women imagined someone like themselves. In mainstream China, younger participants did too—but girls and women increasingly favored men with age. [osf.io/preprints/psyarxiv/ejd76_v1] #NewPreprint
0162
Reposted by Robert Hawkins
DurstewitzLab @durstewitzlab.bsky.social · 02/10/2026
In our #NeurIPS2026 paper “Topological Out-of-Domain Generalization in Dynamical Systems (DS) Reconstruction” (arxiv.org/abs/2606.22969) we aim to infer the DS generating observed TS jointly with control parameters, in order to extrapolate to different dynamical regimes, e.g. across tipping points.
092
Reposted by Robert Hawkins
Akhila Yerukola @akhilayerukola.bsky.social · 01/10/2026
Did you know gifting four koi 🐟 is offensive in Japan (4 = "shi" = death 💀), or that chopsticks 🥢upright in rice is inappropriate in China (like funeral incense 🕯️)? Top VLMs don't! Our #COLM2026 paper introduces 🌏NormViz to measure visual norm understanding in 16 countries. Best model: 25.3% 🚨 🧵👇
Overview of NORMVIZ-BENCH for visual norm understanding, consisting of 3,272 contrastive image pairs across 16 countries that differ only in the culturally relevant behavior. (a) A contrastive pair example from Japan; group accuracy requires a model to correctly classify both images in a pair. (b) Images are sourced via Text-to-Image generation,
inpainting, and Retrieval. (c) Benchmark statistics and pair type distribution.
182
Reposted by Robert Hawkins
Josh de Leeuw @joshdeleeuw.bsky.social · 01/10/2026
I've been working on improving webcam eye tracking for 3+ years and it's nearly ready to go. After a lot of ups and downs I'm in the excited phase again after the latest validation study!
0527
Reposted by Robert Hawkins
James Michaelov @jamichaelov.bsky.social · 28/09/2026
Looking forward to #SNL2026! I’ll be presenting the work published in our recent JML paper about language model scaling and the N400. Find me (or email/message) if you want to chat about predictive coding and NLP in the study of human language comprehension! Paper link: doi.org/10.1016/j.jm...
Title: Better language models better model the N400, but not reading time

Abstract: The probability of a word in context, as captured by large language models, is predictive of both behavioral and neural measures of human language processing. Intuitively, language models that are better at next-word prediction might better model predictability effects in human language comprehension. Yet recent work suggests that language models can become too good at next-word prediction to model reading time, implying that the aspects of human comprehension indexed by reading time do not track perfectly with predictability from language statistics alone. However, it is unknown whether this decoupling is true of reading time only, or whether it is intrinsic to online measures of comprehension more generally. To address this question, we turn to another robust and well-studied measure of online processing, the N400 component of the event-related brain potential. We compare how a language model’s size, number of training tokens, and performance on natural language benchmarks correlate with its ability to predict both reading time and N400 amplitude. Based on an analysis of 4 reading time datasets and 9 N400 datasets, we replicate past results for reading time, but find that larger language models, models that are trained on more data, and models that perform better at next-word prediction and other more complex natural language tasks are better able to predict N400 amplitude. We interpret this difference between the N400 and reading time measures as potentially revealing the comparative importance of semantic prediction in the neurocognitive processes indexed by the N400.
092
Reposted by Robert Hawkins
ingmarvisser.bsky.social @ingmarvisser.bsky.social · 28/09/2026
New paper 🍼📄 ManyBabies 3 is out (open access, registered report in Developmental Science). 30 labs, 33 languages, 839 infants set out to replicate a textbook finding: 7-month-olds learn abstract rules from speech (@garymarcus.bsky.social et al., 1999). We didn’t find it. 🧵
37647
Reposted by Robert Hawkins
Vivian Paulun @vivianpaulun.bsky.social · 25/09/2026
🚨 Preprint alert! 👶🧠🌊 "The Emergence of a Distinction between Things and Stuff" osf.io/preprints/ps... So excited to share my first infant study, with the wonderful Sanghee Song and Liz Spelke.🎉 Do babies expect sand to behave differently from solid objects? 🧵1/5
411833
Reposted by Robert Hawkins
Félicie Dhellemmes @felicie-dhellemmes.bsky.social · 25/09/2026
Proud to present our latest work "Tracking human foragers and their prey reveals adaptive predator-prey dynamics", now hosted on BioRxiv! 🧵 doi.org/10.64898/202... @ralfkurvers.bsky.social @scioi.bsky.social @arc-mpib.bsky.social @alexschakowski.bsky.social @dominikdeffner.bsky.social
26527
Reposted by Robert Hawkins
Dr. Zed Sehyr @zedsehyr.bsky.social · 25/09/2026
🧠🤟 How is meaning organized in ASL? "It's a small world of signs: Semantic network structure in American Sign Language based on free associations" We asked deaf ASL signers to give three meaning-related signs for each sign in ASL-LEX that immediately came to mind.
osf.io
OSF
1114
Reposted by Robert Hawkins
Institute for Replication @i4replication.bsky.social · 25/09/2026
Over the past two years, I4R partnered with Psychological Science (@psychscience.bsky.social) on a large reproducibility project. 67 independent teams reproduced 64 articles published in 2024–2025, about 44% of eligible papers. Paper: www.econstor.eu/bitstream/10... Here's what we learned 🧵
18958
Reposted by Robert Hawkins
Martin Hebart @martinhebart.bsky.social · 25/09/2026
What is the nature of universal representations in AI models, and what determines whether they emerge? Our paper accepted at #neurips2026 led by @florianmahner.bsky.social & @rothj.bsky.social addressed these questions comparing 162 vision models, with intriguing results. arxiv.org/abs/2605.13675 🧵
arxiv.org
Characterizing Universal Object Representations Across Vision Models
Deep neural networks trained with different architectures, objectives, and datasets have been reported to converge on similar visual representations. However, what remains unknown is which visual prop...
34115
Reposted by Robert Hawkins
Sonja Vernes @sonjavernes.bsky.social · 23/09/2026
🦇 Bat1K's biggest paper yet is out. Largest bat genome + fossil study ever: 103 genomes, 44 fossils, 137 researchers. We rewrote the bat family tree! Bats likely originated in Europe ~65 million years ago, not Asia, Africa or Americas. @bat1kgenomes.bsky.social doi.org/10.1038/s41586-026-11007-3
16429
Reposted by Robert Hawkins
Ann Kennedy @antihebbiann.bsky.social · 22/09/2026
Preprint time! This is a cool one. Complex systems, be they neural nets, interacting genes, or power grids, can produce diverse dynamics-- things like point attractors, limit cycles, and chaos. The dynamic landscape you get depends on how elements interact. But those interactions can also change!
28125
Reposted by Robert Hawkins
Tomer Ullman @tomerullman.bsky.social · 22/09/2026
new preprint: "Directing large language models to follow the letter or spirit of the law" arxiv.org/pdf/2609.23083 (by Qian , Li, Chen, Murthy @soniakmurthy.bsky.social , Belinkov, and me) this is particularly cool/important, and I'm allowed to say it because it was headed by @pqian.bsky.social
17323
Reposted by Robert Hawkins
Yu Kanazawa @knzw783.bsky.social · 21/09/2026
👶🧠 Pequay, E., & Dautriche, I. (2026). The development of compositionality in language and thought. Annual Review of Developmental Psychology, 8. Advance online publication. doi.org/10.1146/annu...
doi.org
The Development of Compositionality in Language and Thought
Compositionality is the property of a representational system whereby complex meanings arise from the combination of simpler meaningful elements. Here we take a developmental perspective to…
0195
Reposted by Robert Hawkins
Evan Westra @evanwestra.bsky.social · 21/09/2026
Me and @dryan149.bsky.social's commentary on @julianje.bsky.social and @yarrowdunham.bsky.social's excellent BBS target article "The Institutional Stance" just went live. JJ-E and YD argue that sometimes, we rely on role-based representations instead of mentalizing... philpapers.org/rec/WESWDW-2
philpapers.org
1199
Reposted by Robert Hawkins
Morten H. Christiansen @mh-christiansen.bsky.social · 20/09/2026
👉 New paper with Cris Rivera in Cognitive Science providing preliminary evidence that although chunking across multiple levels of linguistic representation is important to overcoming the Now-or-Never bottleneck, there is no unitary chunking ability 1/2 onlinelibrary.wiley.com/doi/abs/10.1...
onlinelibrary.wiley.com
Is Chunking Ability Unitary Across Levels of Linguistic Representation?
Much like with statistical learning, there is a tendency in the cognitive and language science literature to treat chunking as a unitary, domain-general ability that operates fairly similarly across ...
3193
Reposted by Robert Hawkins
Claudio Tennie @ctennie.bsky.social · 20/09/2026
The role of know-how copying in and beyond [Chater & Christiansen's] Social Tinkering - BBS, just out doi.org/10.1017/S014...
doi.org
The role of know-how copying in and beyond social tinkering | Behavioral and Brain Sciences | Cambridge Core
The role of know-how copying in and beyond social tinkering - Volume 49
1169
Reposted by Robert Hawkins
Gaia Molinaro @gaiamolinaro.bsky.social · 19/09/2026
The ability to pursue self-determined goals is key to human intelligence. Could foundation models unlock similar capabilities in artificial agents? We explore this possibility in a short review — fresh off the press! @ccolas.bsky.social @pyoudeyer.bsky.social @annecollins.bsky.social
1143
Robert Hawkins @rdhawkins.bsky.social · 18/09/2026
new NTS show just dropped
020
Reposted by Robert Hawkins
Iyad Rahwan | إياد رهوان @iyadrahwan.bsky.social · 16/09/2026
Introducing: Time Machine Experiments? 🚀 ⏳ in which participants 'travel' to the past, and interact with a mind from the year 1930 (simulated by an AI with knowledge cut-off). Preprint: arxiv.org/pdf/2609.15468
2157
Reposted by Robert Hawkins
Thomas Talhelm @thomastalhelm.bsky.social · 16/09/2026
🚨 It's published! 🚨 This is the big one. 100 cultures, 12,000 people, 108 researchers. Why people (and our most-used surveys) get collectivism wrong and how to fix it. www.nature.com/articles/s41... @natureportfolio.nature.com
312462
Reposted by Robert Hawkins
Eric Brandom @ebrandom.bsky.social · 16/09/2026
Pirate crews, charted
Chart of connections between pirate crews. From Rediker's early article, "Under the Banner of King Death" tho reproduced in Villains of All Nations.
519132
Reposted by Robert Hawkins
Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 15/09/2026
"LLMs are just n-dimensional classifiers" has a very similar flavor to "everything is just quantum mechanical interactions." www.science.org/doi/10.1126/...
science.org
More Is Different
410512
Reposted by Robert Hawkins
Mariam Aly @mariamaly.bsky.social · 15/09/2026
How do keep a narrative active in our mind despite interruptions? The default mode network can maintain a representation of an interrupted story over a delay, even when the delay is filled with a demanding task. This representation is strikingly transformed from the initial experience!
biorxiv.org
Cortical and Hippocampal Pathways to Mental Continuity
How do people preserve mental continuity when tasks are interrupted? One established pathway runs through the hippocampus, which stores snapshots of our mental context for later reinstatement. We prop...
17824
Reposted by Robert Hawkins
Sukaina @sukaina.bsky.social · 15/09/2026
Thrilled to tell you that my paper “Protests as World Making Projects” with bestie Lidal Dror is now forthcoming and available open access on early online view at Nous 🧵 onlinelibrary.wiley.com/doi/10.1111/...
onlinelibrary.wiley.com
Protests as World‐Making Projects
On one common way of distinguishing forms of protest, civil disobedience is a communicative act that aims to achieve its goals through moral persuasion, whereas direct action has as its goal to direc...
15920
Reposted by Robert Hawkins
Steve Fleming @smfleming.bsky.social · 14/09/2026
Our paper “Confidence is detection-like in high-dimensional spaces” is now published in Open Mind, led by @wiktoriakozyra.bsky.social and @kevingoneill.github.io direct.mit.edu/opmi/article...
direct.mit.edu
Confidence Is Detection-Like in High-Dimensional Spaces
Abstract. Confidence estimates are often “detection-like”—driven by positive evidence in favour of a decision. This empirical observation has been interpreted as showing that human metacognition is li...
15214
Reposted by Robert Hawkins
Sakana AI @sakanaai.bsky.social · 14/09/2026
Introducing PC-ALM: a local-learning alternative to backprop that trains 1000-layer neural nets using only local dynamics. Blog: pub.sakana.ai/pc-alm/
18714
Reposted by Robert Hawkins
Alison Gopnik @alisongopnik.bsky.social · 14/09/2026
A very interesting and clear (and reassuring?) piece relating the Farrell et al (www.science.org/doi/full/10....) idea of LLM's as cultural technology to the recent developments in AI Mathematics. terrytao.wordpress.com/2026/09/13/h...
terrytao.wordpress.com
Happy, those able to know the causes of things
[This is a guest post by Nestor Guillen, crossposted from his blog. This blog post was initially written in a different file format and converted using AI. — T.] Keywords: LLMs, cultural tech…
1238
Reposted by Robert Hawkins
Neil Cohn @neilcohn.bsky.social · 14/09/2026
Our new paper may be one of coolest studies I've ever done. We show that the structures of visual storytelling in comics aligns with the word order of the languages of the comic authors. Here's what we did... www.sciencedirect.com/science/arti...
sciencedirect.com
Syntactic headedness aligns with visual narrative structure
Recent research has highlighted how visual narrative sequences involve similar structures and processing as syntax in language, yet no work has asked …
48734
Reposted by Robert Hawkins
Sam McDougle @actlab.bsky.social · 14/09/2026
New preprint! In this study, the fantastic Apoorva Sharma demonstrates that the amount of time that passes during motor preparation can "tag" for retrieval otherwise competing motor memories. Modeling work links these fx to putative temporal basis sets in the cerebellar cortex. Feedback welcome!
14516
Reposted by Robert Hawkins
NBER @nber.org · 13/09/2026
Foundational expertise may be a prerequisite for extracting durable skill from AI-assisted practice, from David Autor, Tanya Rodchenko, Josh Martin, Zanna Iscenko, Scott Strand, David Pearl, and Melissa Ferere www.nber.org/papers/w35720
02616
Reposted by Robert Hawkins
Tom Silver @tomssilver.bsky.social · 13/09/2026
This week's #PaperILike is "Latent Programming Horizons in Coding Agents" (Silva et al., 2026). Now seems like a good time to better understand what coding agents are doing. Here's a nice example of the kind of analysis one can do. PDF: arxiv.org/abs/2607.05188
arxiv.org
Latent Programming Horizons in Coding Agents
A coding agent solving a software-engineering task spends dozens of steps reasoning, editing code, and running tests, yet little is known about what the underlying language model internally represents...
084
Reposted by Robert Hawkins
NBER @nber.org · 12/09/2026
When experts know more than their evidence proves, verifiability trades credibility for flexibility: it aids communication under conflict but hinders it under aligned preferences, from Alessandro Lizzeri, Yichuan Lou, and Jacopo Perego www.nber.org/papers/w35712
053
Reposted by Robert Hawkins
Alex Holcombe @alexh.bsky.social · 12/09/2026
It's Peer Review Week, and I have pledged to review at least one article per year for a diamond open access journal; one that charges nothing to publish and nothing to read, and is run by and for the scholarly community, using the new Diamond Reviewer Pledge (forrt.org/diamond-revi...).
forrt.org
Diamond Reviewer Pledge
Signatories of the Diamond Reviewer Pledge commit to reviewing at least one article in the next 12 months for a diamond open access journal.
23110
Reposted by Robert Hawkins
Lisa Bylinina @bylinina.bsky.social · 11/09/2026
look at my new babylm paper! arxiv.org/abs/2609.11870 basically, i initialize token embeddings with representations from an image encoder rather than randomly and then train text-only as usual. kind of a visual demonstration to start off the word learning process
1155
Reposted by Robert Hawkins
Michael Clemens @mclem.org · 13/09/2026
This will devastate our research+innovation capacity, bedrock of the 21st century economy. 35% of *all* employed people in the US with STEM PhD degrees are both foreign-born *and* graduated from a US university. —> nap.nationalacademies.org/resource/292... Shutting them out is extremist folly.
1829223
Reposted by Robert Hawkins
Chris Krupenye @chriskrupenye.bsky.social · 10/09/2026
🚨 PhD position in human, nonhuman primate, and/or dog cognition! I am thrilled to be reviewing applicants this year for a new PhD student to join the Social & Cognitive Origins Group in the Department of Psychological & Brain Sciences at Johns Hopkins @jhu.edu @jhuartssciences.bsky.social 1/
18272
Reposted by Robert Hawkins
Max Puelma Touzel @mptouzel.bsky.social · 11/09/2026
Complex Systems Science of Agents is heating up! -Ising: arxiv.org/abs/2608.16578 -EGT: www.pnas.org/doi/pdf/10.1... and ours: arxiv.org/abs/2609.05442
arxiv.org
Physics of Agents: Statistical Mechanics Predicts Collective Behavior of AI Agents
AI agents increasingly operate as part of interacting systems rather than in isolation. As agents exchange information and jointly make decisions, their interactions can improve collective reasoning b...
2184
Reposted by Robert Hawkins
Kunal Jha @kjha02.bsky.social · 11/09/2026
Can self-interested, self-improving, self-replicating agents learn to cooperate? Our new paper, Tapes Together Strong, shows they can: when social behavior, computation, and reproduction share one energy budget, cooperation evolves from scratch. arxiv.org/abs/2609.10817 🧵
48418
Reposted by Robert Hawkins
Maria Antoniak @mariaa.bsky.social · 10/09/2026
New from our lab! #COLM2026 When people generate stories, they don't just write one prompt. Instead, they explore narrative space via branching edits 🌱 We reconstruct 24k of these edit trees 🌳 from chat logs and map edit types, story formats, how they relate to tree depth, and more!
The Garden of Forking Prompts: How Users Explore Narrative
Space in Story Generation
Advait Deshmukh♣ Nora Benedict♠ Melanie Walsh♡ Maria Antoniak♣
♣University of Colorado Boulder ♠University of Georgia ♡University of Washington

Abstract

Large language models (LLMs) have changed the way people engage
with stories. Drawing on public chatbot logs, we can see that when users
generate stories, they iteratively edit their prompts to explore narrative
possibilities, adjusting characters, redirecting plots, and swapping fictional universes. As aggregated data, these prompts represent rich traces of creative preference at scale. Yet story generation evaluation benchmarks rely on static, one-shot prompts that cannot capture this exploratory behavior. In this work, we study how users revise consecutive story prompts in the wild. Using a dataset of naturally occurring user-chatbot conversations, we construct WildStories, a sample of 275,635 story generation prompts (labeled with story format, prompt components, and explicitness), and WildEdits, a collection of 24,291 edit trees that model how users iteratively edit base story prompts and explore branching story possibilities. From these trees we develop a framework of edit types crossing four directions (adding, removing, changing, and extending) with fourteen targets (e.g., plot, character, genre). We then use our datasets and this framework to
analyze user behavior in navigating narrative space via LLMs. Finally, we
show how automated permutations based on the framework can be used
for story generation benchmarking. Content Warning: This paper works with “wild” chatbot logs, which often include toxic and sexually explicit themes.
29629
Reposted by Robert Hawkins
samuel mehr @mehr.nz · 10/09/2026
today we're launching Tone Twist, a new listening game by Sorour Zekrati 🧠🧪👂🏻🔬 in English a tone is "high" or "low" but this is not true of all languages. eg, in Farsi it's "thin" or "thick" Tone Twist asks whether seeing visual metaphors like those interferes with auditory perception. try it here:
themusiclab.org
Tone Twist
Keep your eyes on the motion and your ears on the pitch. Can you track both at once?
21812
Reposted by Robert Hawkins
Jin Ke @jinke.bsky.social · 18/05/2026
Preprint alert! 🧠 Using movie-watching fMRI and NLP, we show that prior social impressions shape how new social information is processed and updated over time, while moments of sudden insight ("aha"!) accompany transient shifts in brain activity that predict impression updates. (1/10)
biorxiv.org
15217
Reposted by Robert Hawkins
Communications Psychology @commspsychol.nature.com · 10/09/2026
A new research Article in the journal by Aaron Baker and colleagues: www.nature.com/articles/s44... (in Press) @aajbaker.bsky.social @yarrowdunham.bsky.social @julianje.bsky.social
nature.com
People use roles to quickly predict how others will act, what they know, and when they can be replaced - Communications Psychology
Using a self-paced reading paradigm, we show that, from just the mention of a role, people build rapid expectations about how other people will act, what they know, and whether they can be interchange...
0157
Reposted by Robert Hawkins
Irina Giurgea @dirinagiurgea.bsky.social · 10/09/2026
🚨Preprint📃🧠 I'm excited to share a preprint for the second paper from my PhD exploring the role of social experience and motivated cognition in concept processing. With @veronicadiveica.bsky.social @pennypexman.bsky.social @rjbinney.bsky.social
2114
Reposted by Robert Hawkins
Marlene Cohen @marlenecohen.bsky.social · 10/09/2026
Now out in Nature Neuroscience: fantastic work by @cxue.bsky.social, Sol Markman et al who discovered a neural reason that performance suffers under cognitive load: “Feature interference underlies a neuronal basis for the behavioral cost of task uncertainty”. www.nature.com/articles/s41... 🧪🧵1/
nature.com
Feature interference underlies a neuronal basis for the behavioral cost of task uncertainty - Nature Neuroscience
Using a combined approach of monkey electrophysiology, artificial neural networks and human psychophysics, Xue et al. show that neuronal interference from irrelevant information underlies behavioral e...
312040
Reposted by Robert Hawkins
Jay Van Bavel, PhD @jayvanbavel.bsky.social · 10/09/2026
The US is facing a mass exodus of scientists The funding cuts & attacks on science are demoralizing a generation of brilliant minds One-third of young scientists who were planning careers in academic research changed their minds in 2025 and many now plan to leave www.nber.org/system/files...
13718
Reposted by Robert Hawkins
Riccardo Fusaroli @fusaroli.eurosky.social · 10/09/2026
How does the architecture of a city shape the way we think, talk, and draw space? And how do spatial conventions travel across cultural generations? New paper in Open Mind, spearheaded by Z Alinam (w J Nölle, C Vesper, K Tylén & me): doi.org/10.1162/OPMI.a.380 🧵 1/
doi.org
When the City Speaks: How Urban Form Shapes the Cultural Transmission of Spatial Conceptualization
Abstract. How do urban environments shape the way people conceptualize and communicate space? Using a cultural transmission experiment in virtual reality, we asked participants to navigate one of two ...
25022
Reposted by Robert Hawkins
William Gilpin @wgilpin.bsky.social · 08/09/2026
Our group discovered that reasoning models produce fractals when asked to solve hard problems. We can use nonlinear dynamics to probe the thinking processes of recurrent depth models on Sudoku, mathematics, and even ARC-AGI (1/N) arxiv.org/abs/2609.04963
24010