This is one of the most outstanding examples of circuit understanding I've seen in a long time. The unification of theory and experiment is beautiful.

When Malcolm presented this in my lab, the audience was cheering at the end, and one person shouted (non-ironically) "You did it!"

Now out in Cognition, work with the great @gershbrain.bsky.social @tobigerstenberg.bsky.social on formalizing self-handicapping as rational signaling!
📃 authors.elsevier.com/a/1lo8f2Hx2-...

Interesting! We’re trying to figure out _why_ LLMs don’t quite rely on counterfactual reasoning when judging responsibility. It could be—as you suggested—that they’re worse at counterfactual simulations, or that they simply don’t think counterfactuals are relevant here. Excited to dig further 🙂

Come by our poster at CogSci (Poster Session 2, P2-X-215), Friday 8/1 at 10:30am!

Our results shed light on how we can make LLMs more human-like and how to study the mechanisms underlying complex behavior in LLMs. Co-led by me and @ebig.bsky.social, with the great @tobigerstenberg.bsky.social @tomerullman.bsky.social @gershbrain.bsky.social (4/4)

LLM and human data are highly correlated, BUT they are best explained by different factors! LLMs evaluate collaborators based on force (how much output they contribute), whereas humans evaluate collaborators based on their actual and counterfactual effort. (3/4)

We adapted materials from human studies on responsibility and reward attributions and compared LLMs’ responses to human data and seven cognitive models. (2/4)

Our latest on the cognitive science of LLMs! To be presented @CogSci‬2025 🎉

LLMs are increasingly involved in human collaborations. How do LLMs assign responsibility and reward to collaborators? Is it similar to how humans do it? 🤖🧑

📃 gershmanlab.com/pubs/XiangBi... (1/4)

🎈 Out now: 🎈

"The capacity limits of moving objects in the imagination"

(by Balaban & me)

of interest to people thinking about the imagination, intuitive physics, mental simulation, capacity limits, and more

www.nature.com/articles/s41...

@gershbrain.bsky.social, @yangxiang.bsky.social, and I have a new project out in preprint form!

osf.io/preprints/ps...

Here are the main takeaways: (1/6)

By offering a systematic explanation of self-handicapping, we hope to lay the groundwork for developing effective interventions that target academic self-handicapping, helping people to realize their full potential. A preprint of the paper is available on PsyArxiv: osf.io/preprints/ps... (5/5)

We tested the theory's predictions in two experiments, showing that self-handicapping occurs more often when it’s unlikely to affect the outcome and when it increases a naive observer's perceived competence. With sophisticated observers, it’s less effective when followed by failure. (4/5)

We developed a signaling theory of self-handicapping, involving a naive observer who evaluates the actor’s competence, an actor who seeks to impress the naive observer through strategic self-handicapping, and a sophisticated observer who considers the actor’s decision whether to self-handicap. (3/5)

Self-handicapping is a strategy where people deliberately impede their performance to protect perceived competence in case of failure, or enhance it in case of success. Despite much prior research, it is unclear why, when, and how self-handicapping occurs. (2/5)

Really excited about this project, and thanks so much to my wonderful collaborators @gershbrain.bsky.social @tobigerstenberg.bsky.social for making this happen! Some main takeaways in thread 🧵 (1/5)