Register now!

act.standupforscience.net/event_campai...

@chelseaboccagno.bsky.social
@dominicmdenning.bsky.social

Ahh, didn't know you were over here on bluesky! Sorry for not tagging ya :). It was awesome getting to work with & learn from you through this work. Cheers to it finally being in print -- & to hopefully many more chats re: perception & beauty in the future!

& see here for a great summary from Hobart & William Smith (Dan Graham's institution), written by Andrew Wickenden: www.hws.edu/news/2025/se...

Huge shout-out to Colin for the incredible computational feat he led!

Since then, I've also learned a ton about computational aesthetics from my fellow co-authors, who are rigorous experts in this space & just wonderful people in general: Dan Graham & Ed Vessel. Great early birthday present for me -- & for Colin, who shares my birthday!

This paper is particularly meaningful for me b/c Colin & I have been chatting about affect, emotions, & aesthetics since year 1 of grad school--w/ me coming at it from an affective science angle, & Colin from a computational aesthetics angle. We sought to merge these lits here.

We looked @ 3 aspects of affective experience: valence, arousal, & beauty. Broadly, we found that machines that got 0 training on emotion can predict considerable variance in human ratings of affect (53%, on average). What does it mean?! See the paper for a whole lot more :)

We used 180 deep neural networks trained only on vision tasks to identify whether these models--despite never having been trained on emotion--can predict how humans report feeling when seeing different images (art, landscapes, etc.).

I aim to do this every time 😅 for me it’s solely my effort to control the stress of airport procedures. One time when I was rapidly stacking bin after bin, a TSA agent told me I should work there. Highlight of my airport experiences 🙌

thanks so much!

thanks for making this! Could I be added if there's room?