Sign in

Sev Harootonian

@harootonian.bsky.social
90 followers 80 following 9 posts

Postdoc at NYU studying how and when people trust and defer to AI. PhD from Princeton. Sometimes I teach machines to teach. sharootonian.github.io

PostsRepliesMedia
Sev Harootonian @harootonian.bsky.social · 26/08/2026
BONUS: We also looked at if LLMs teach the same way and whether the same nudges work on them. arxiv.org/abs/2604.01594 6/5 :)
arxiv.org
Do Large Language Models Mentalize When They Teach?
How do LLMs decide what to teach next: by reasoning about a learner's knowledge, or by using simpler rules of thumb? We test this in a controlled task previously used to study human teaching strategie...
060
Sev Harootonian @harootonian.bsky.social · 26/08/2026
Of course, there's also a bunch of cool modeling in the paper: the mentalizing teacher is modeled with Bayesian inverse reinforcement learning, inferring what the learner knows from their behavior alone. 5/5 Paper: www.nature.com/articles/s41...
nature.com
Mentalizing and heuristics as distinct cognitive strategies in human teaching - Nature Human Behaviour
Harootonian et al. find that people switch between effortful, tailored teaching and easier, simple shortcuts, favouring shortcuts even when unhelpful. Prompting participants to reflect on learners’ li...
230
Sev Harootonian @harootonian.bsky.social · 26/08/2026
In Exp 3 we pushed people toward the effortful strategy just by probing teachers about the learner's knowledge before they taught. And they kept teaching by mentalizing even for new learners we never probed them about. We can scaffold better teaching. 4/5
120
Sev Harootonian @harootonian.bsky.social · 26/08/2026
Mentalizing is way more effortful: you have to infer what caused the learner's mistakes from their behavior alone. So a lot of teachers fall back on the cheaper strategy and teach what seems best in general, without asking whether this learner can actually use it. 3/5
110
Sev Harootonian @harootonian.bsky.social · 26/08/2026
People mostly fall into 2 camps. Some teach by mentalizing: inferring the learner's mental state, including their knowledge, to pick the most useful thing to teach. Others teach with heuristics: ignore the learner, teach whatever tends to be best in that situation. 2/5
140
Sev Harootonian @harootonian.bsky.social · 26/08/2026
My main PhD paper is finally out in Nature Human Behaviour! The question: what cognitive strategies do people use when they teach, and can we nudge them toward a better one? Work with my PhD mentors @yaelniv.bsky.social, @markkho.bsky.social and @cocoscilab.bsky.social. 1/5
18619
Sev Harootonian @harootonian.bsky.social · 19/05/2025
Our work frames teaching as a resource-rational trade-off and why teachers rely on simple but suboptimal heuristics. Incorporating these insights could inform how think about machines that teach people or help teachers teach people more effectively. 3/3
010
Sev Harootonian @harootonian.bsky.social · 19/05/2025
Across three experiments we found variability in people’s use of mentalizing vs heuristics, show how people systematically choose to use one or the other strategy, and that brief interventions can reliably shift people toward more or less effortful strategy. 2/3
130
Sev Harootonian @harootonian.bsky.social · 19/05/2025
🚨 New preprint alert! 🚨 Thrilled to share new research on teaching! Work supervised by @cocoscilab.bsky.social, @yaelniv.bsky.social, and @markkho.bsky.social. This project asks: When do people teach by mentalizing vs with heuristics? 1/3 osf.io/preprints/os...
23314