Tomer Ullman
banner
tomerullman.bsky.social
Tomer Ullman
@tomerullman.bsky.social
Associate Professor, Department of Psychology, Harvard University. Computation, cognition, development.
November 20, 2025 at 8:56 PM
November 20, 2025 at 8:56 PM
wondering if youtube comments count as data
November 19, 2025 at 1:53 PM
Officially out in the current issue of Trends in Cognitive Sciences:

"Physics versus graphics as an organizing dichotomy in cognition"

www.cell.com/trends/cogni...
November 17, 2025 at 1:55 PM
I can't claim to understand what it's like to be that guy from Memento, but I do have to pull up my phone and scroll through photos in order to answer the question 'so what did you do this weekend?'
November 16, 2025 at 8:40 PM
November 16, 2025 at 5:51 PM
something i noticed in preparing a class
November 14, 2025 at 2:01 PM
wonder if I can re-arranging this Epstein name dump to the tune of We Didn't Start the Fire
November 14, 2025 at 3:30 AM
dear academic book presses 👋

Don't Do This 🤗
November 12, 2025 at 1:58 PM
oh wowww
November 12, 2025 at 5:01 AM
I love the addition of a label that says "an accurate 1000-sided chiliagon' to something that isn't it, like a reverse 'The Treachery of Images' by Magrittte, like painting something that isn't a pipe and saying 'Ceci une pipe'
November 11, 2025 at 2:09 PM
A chiliagon is a 1000-sided polygon, famous in philosophy for Descartes using it as an example of the diff b/w imagining something visually and conceptualizing it.

Anyway I tried to get Gemini to draw it
November 11, 2025 at 2:09 PM
we examined "Chain of Time" in 2d and 3d domains, including simple motion, fluid pouring, bouncing, and moving under gravity.

Chain of time improved things on most of them, and improved it more the more fine-grain the chop

(but not for everything)
November 10, 2025 at 1:57 PM
there are obvious parallels here to both human mental simulation and LLM in-context reasoning (including Chain of Thought), which are indeed our inspiration.
November 10, 2025 at 1:57 PM
the idea is pretty well captured by this image:

the baseline method is to give a model some physical stimuli as input and ask it 'what will this look like x steps into the future?"

Chain of Time breaks this into smaller steps; simulate, output, use that output as new input, repeat.
November 10, 2025 at 1:57 PM
new pre-print,

"Chain of Time: In-Context Physical Simulation with Image Generation Models"

(by Wang, Bigelow, Li, and me)

arxiv.org/abs/2511.00110
November 10, 2025 at 1:57 PM
the girl (9) made a postcard for Illinois. I'm not exactly sure on the "Why" but full marks for the "What"
November 7, 2025 at 1:48 PM
November 6, 2025 at 1:55 PM
(so to explain how these lines are relevant it turned out I first needed to briefly explain Hamlet)
November 5, 2025 at 2:01 PM
for various reasons I had to briefly explain the plot of Hamlet to my seminar class yesterday (turns out none of the students had read or seen it), and so I got to use the phrase "Hamlet is Big Mad about this"
November 5, 2025 at 2:01 PM
mid-semester shout out to Andrew Shtulman's book "Learning to Imagine"; I'm teaching a seminar on imagination and pretense, and the book has fun and useful, and leads to interesting discussions
November 4, 2025 at 1:44 PM
Nature suggests you use their "Manuscript Adviser" bot to get advice before submitting

I uploaded the classic Watson & Crick paper about DNA structure, and the Adviser had this to say about one of the greatest paper endings of the century:
November 3, 2025 at 1:55 PM
as above, so below
November 2, 2025 at 3:17 PM
מסורת האלווין: בבוקר אחרי, הילדים מוסרים את מה שנשאר תמורת מתנה, וזה מוסדר בחוזה.

(דרך ללמוד על פרצות בחוק. בגרסאות קודמות של החוזה לא היה כתוב איזה מתנה הם יקבלו ומתי, למשל)
November 1, 2025 at 4:08 PM
a relevant False Knees
October 31, 2025 at 2:05 PM