Tomer Ullman

someone talking loudly with a friend at the coffee shop: lovely, the fun of being human, the joy of a coffee shop, good for them

someone talking loudly over a zoom conversation at the coffee shop: you are the worst of us, how dare you, this is not what we Do Here

embedded video paused at the right time

It's been 15 years since Edna Ullmann-Margalit passed away, and I keep going back to stuff she's written.

I highly recommend 'Normal Rationality', which collects her essays.

If you're looking to start, maybe look here:

bit.ly/4qk2GZS

bit.ly/46XudIV

bit.ly/4nkfLQc

bit.ly/3KUIiOy

the best things in life are freeequently around 20$

I find the model in your next linked paper quite relevant, as it suggests such a separation :)

Thanks! we make no strong hypothesis about what processes lead to not noticing stuff in perception, call it M[real image], the q is then whether non-commitment is due to M[mental images], and it seems not.

Halloween stores sections are organized by "decorations", "props", and "costumes" when they should be organized by "spooky", "scary", and "creepy"

There's a LOT more to say, and the discussion gets into a bunch of interesting related ideas in neuroscience in beyond, but you can read the paper for that :)

++

Kudos to the co-authors, especially Yichen Li, who led the project from start to finish

(think of "dropping something onto something" - what do you see?)

Li found properties like type and location were more likely to to appear or occur earlier

(this builds and expands on a 'physics vs. graphics' proposal of scene construction www.sciencedirect.com/science/arti...)

we could just stop there, but Yichen Li went further. If non-commitment is driven by lazy evaluation is hierarchical scene construction, maybe we can probe this process further?

Li designed clever studies to do that, asking people to report what they imagined first in an abstract scene

if non-commitment is similar to perceptual inattention, these properties should strongly correlate.

...But they don't. Across scenes and two experiments, there's no real relationship between what people don't notice in real images and what they report not committing to

we created a bunch of real images & scenes, validated to match prompts used previously for studying non-commitment.

we briefly showed people the real images, then examined what properties people noticed, then compared it to non commitment.

when ppl say 'I didn't think about the color of the ball', did they

(1) create a full, perceptual-like mental image and then forget (or encode) the color, or

(2) really just not think of the color to begin with?

(these options showed up decades ago, but weren't studied empirically)

Non-commitment in mental imagery We examine non-commitment in the imagination. Across 5 studies (N > 1, 800), we find that most people are non-committal about basic aspects of their m…

the empirical work found ppl don't commit to basic details in the imagination + non-commitment is non-uniform (more here: www.sciencedirect.com/science/arti...)

BUT, we forget (or don't notice stuff) in real images. So maybe imagery is super rich, and we just forget details like in real images?

the story so far, briefly:

when people imagine ("a person knocks a ball off a table") they seem to leave basic details out (e.g. the color of the ball). We call this 'non-commitment in imagery'

This was noticed years ago, but only studied empirically recently; it matters for the imagery debate

New preprint!

"Non-commitment in mental imagery is distinct from perceptual inattention, and supports hierarchical scene construction"

(by Li, Hammond, & me)

link: doi.org/10.31234/osf...

-- the title's a bit of a mouthful, but the nice thing is that it's a pretty decent summary

This is separate from the massive pressure on Harvard (and other unis) in the form of freezing federal money until demands are met.

Basically there are several different shakedowns, threats, and extortion attempts going on in parallel

The administration recently offered a specific compact to 9 unis (MIT, Brown, U Tex, etc) that said 'accept these terms and we will give you favorable funding treatment', Harvard wasn't one of them.

I agree that MIT is institutionally cooler, but as a minor aside Harvard wasn't offered this compact, and you can't reject what you aren't offered.

(though I'm guessing in the counterfactual world where Harvard was offered this MIT would still reject it before them)

"we present Work(Around)-Bench, a benchmark for assessing whether LLMs are actually doing a task or simply finding silly short-cuts around it. So far every model is at 99.6%"

are you at #COLM2025? Check out this spotlight talk by Sonia Murthy! @soniakmurthy.bsky.social

(room 520B @ 11:30am)

are you NOT at #COLM2025? Check out this paper by Sonia Murthy!

paper: arxiv.org/abs/2506.20666
Kempner Deeper Learning blog feature: kempnerinstitute.harvard.edu/research/dee...

(Daughter, 4 years old, crying her heart out)

Me: "What's wrong, tutu?"

Daughter (moving her hands on the sofa): "If my fingers were markers they would ruin the sofa!"

Me: "But your fingers...are not...markers?"

Daughter (peak distress): "I said IF!"

I had a great time visiting the Northwestern Institute on Complex Systems today (much better than that time at the Southeastern Collective in Simple Entities)

Buffering vocalizations?

is there an accepted technical name for the noises people make while doing some task in the presence of other people, e.g. pulling up some code and meanwhile going 'do do doo dee doo"?

Using Cognitive Models to Reveal Value Trade-offs in Language Models - Kempner Institute People’s actions and words are the result of a balance of different goals. The authors use a leading cognitive model of this value trade-off in polite speech to systematically examine […]

NEW on our #DeeperLearning blog

People balance being kind vs. being honest — and #LLMs should too.

New research shows training choices often favor informativeness over kindness, but prompting can induce sycophancy.

Read more: bit.ly/3Wqrtxl

😉

also the Modified Julesz Conjecture :)

"well, hang on," says Other, "I just have to jimmy with the sail a bit", taking out the sail, "and these planks need to be augmented with the latest Wheel-Method", and so on.

Over time, the ship becomes a motorcycle.

"See?" Other chuckles as they drive away, "the ship can roll on land!"

there's a kind of argument I've taken to calling Theseus' Motorcycle:

You point to the ship and say "this ship cannot roll on land"

"It can too", says the other side.

repeated tests show the ship cannot roll on land,