Sholei Croom

You can see the experiments (and even try them yourself) here: perceptionresearch.org/knots/. And check out the final version when it comes out in Open Mind!

But maybe we've underestimated the breadth of physical intuitions and the contexts they arise in. Our task makes that clear: Despite their long history (knots predate the wheel!) and prevalence in daily life (shoelaces/drawstrings/etc), knots are surprisingly hard to reason about

Why does this matter? One reason: There's been a lot of excitement recently—including in my lab!—about the success of intuitive physical reasoning. There's even an idea that a “physics engine” in the brain approximates accurate outputs across a variety of physical scenarios.

In other words, *even when subjects understand exactly how a knot is structured*, they fail to understand how that structure translates into its strength.

Crucially, this is *not* due to a failure to visually discern the topology of the knots. In other experiments, we show that people have no problem matching a photograph of a knot to a diagram of its structure, including subtle details like the exact placement of the threads.

Our setup is simple: We take a series of similar-looking—but mechanically very different—knots, and ask people which looks stronger (i.e. could withstand more pulling force before unraveling). The actual differences in strength are massive; but do people's judgments reflect this?

Across several variations in presentation—including photographs of the knots, digital renders, dynamic videos, and schematic diagrams—the answer was a resounding NO! People’s strength judgments are so bad that they are not only near chance, but often even below chance!

One of these knots is strong and reliable, a staple of scouting + sailing. The other is a "prank" knot; it's so weak that it slips easily when pulled. But can you tell which is which? Our new paper shows how knots expose the limits of physical reasoning. philpapers.org/rec/FIRTPK

To all our scientist friends around the world who feel outraged about what is happening now and what has been happening for so many decades now: one of the things you can do is to join our organization and work for a better future for Palestine: scientists4palestine.com

Palestine is our struggle. SftP stands in solidarity with Palestinian liberation, and calls on members and supporter to join the movement. Official statement below: scienceforthepeople.org/2023/10/16/sft…

(As you can probably tell I'm not the most prolific tweeter or promoter of my own work 😅😅 but this is the most fun project I've worked on so far in grad school! excited to finally share what I've been up to!)

One of these people is trying to figure out the number of objects in the box; the other is trying to figure out their shape. Can you tell who’s who? In a new @PNASNews paper with @HanbeiZ & @chazfirestone, we explore *epistemic action understanding*! 📜: philpapers.org/rec/CROSAU

Congratulations to Bill Warren, who has been awarded the 2023 Ken Nakayama Medal by @VSSMtg! Bill has made foundational contributions to the study of perception and action, locomotion, crowd behavior, and even the philosophical underpinnings of our field. visionsciences.org/2023-ken-nakay…

My TL a few tweets apart. Nobody does firing in retaliation as well as big tech corporations

I'm hoping that with this panel discussion on Monday I can facilitate a critical conversation about the assumptions our models of morality make, and how we as cognitive scientists can be accountable to the impact these models have when packaged to AI. Hope to see you there! (end)

@mutalenkonde @merbroussard, @jovialjoy @TimnitGebru, @NoTech4Tyrants, (and countless others)

🧵Hello Twitter! Long time no tweet :) I am excited to share I’ve been invited to moderate Monday’s panel discussion at the Engineering and Reverse-Engineering Morality workshop at #CogSci2021. You can find more information about the workshop here: sites.google.com/view/engineeri…