Lightnews — Scholar-powered news

Reposted by Mark Thornton

Brandon Woo @brandonwoo.bsky.social · 19h

New paper in Open Mind, with @emmayu23.bsky.social, Megan Richardson, and @ashleyjthomas.bsky.social! We find that by 6 years of age, children think that close friends know the content of each other’s minds.

direct.mit.edu/opmi/article...

@openmindjournal.bsky.social

Developing Intuitions That Close Friends Know the Content of Each Other’s Minds

Abstract. To maintain and develop close relationships, people need to accurately represent the minds of their social partners. Although studies have characterized many aspects of children’s intuitive ...

direct.mit.edu

1 7 26

Mark Thornton @markthornton.bsky.social · 4d

I agree that there are plenty of computational people who are not AI obsessed or bandwagon riders. I hope some of them are getting these AI jobs instead of others who just started "doing AI" by prompting ChatGPT.

1

Mark Thornton @markthornton.bsky.social · 4d

I think you're interpreting my attempt to provide some quantitative estimates on the phenomenon you're commenting on as a disagreement with the sentiment of your post. That's not how I intended it, and I'm sorry that's how it came across.

bsky.app/profile/mark...

Mark Thornton @markthornton.bsky.social · 4d

I didn't comment on the normative desirability of the number of AI jobs. I also think that the AI hiring focus is unwise and obviously hype driven. My hope is that many cases are just hiring people they'd already want, using the AI label as leverage to get the line from uni/donors.

1

Mark Thornton @markthornton.bsky.social · 4d

I didn't comment on the normative desirability of the number of AI jobs. I also think that the AI hiring focus is unwise and obviously hype driven. My hope is that many cases are just hiring people they'd already want, using the AI label as leverage to get the line from uni/donors.

Mark Thornton @markthornton.bsky.social · 4d

I mean, that job *does* say "computational" on the wiki, so it's included in the 9% figure I quoted. I think that illustrates my point that adding those secondary labels can help give us a rough upper, in addition to the lower bound that comes from looking for "AI" specifically.

1

Mark Thornton @markthornton.bsky.social · 4d

That's why I included the computational/data science cases too, as a comparison, since I figured those would be the most likely to contain hidden AI.

1

Mark Thornton @markthornton.bsky.social · 4d

Yeah, no - I haven't scraped the text of each ad because they're all on differently formatted sites which makes that a real pain. So as you say, it's possible that there are some which have that in the ad but not the wiki description.

1

Mark Thornton @markthornton.bsky.social · 4d

A lot of the AI jobs are cross-listed across areas and ranks (more so than other jobs) which may contribute to the feeling that they are particularly prevalent - the estimates above taken cross-listing of the same job into account.

1 1

Mark Thornton @markthornton.bsky.social · 4d

AI jobs currently make up ~4% of posts on the psych job wiki. If you throw in "computational" and "data" (science), that goes up to 9% For comparison, the median size of the content areas (e.g., clinical, social, cognitive, etc.) is typically around 6-8%.

1 4

Mark Thornton @markthornton.bsky.social · 4d

As far as relative market-share among the areas, neuroscience/biopsych lost the most (-4% relative to last year) while clinical has gained the most (+5%). Of course, clinical was already the largest area, and thus it has lost the most in absolute terms from the general collapse of the market.

Mark Thornton @markthornton.bsky.social · 4d

Comparing this year and covid to the years preceding each, this year currently represents a 1.5x larger drop. There are likely to still be some late job postings (more so than in previous years) but at this point we're sitting at 37% of last year's total.

1 1

Mark Thornton @markthornton.bsky.social · 4d

We're a month further into the job market - how are things looking? The good news is that the market does seem to have been delayed: ~150 new listings have appeared since my last post. The bad news is that the total is still substantially lower than what it was during the covid dip.

2 20 47

Reposted by Mark Thornton

Gabriel Fajardo @gabefajardo.bsky.social · 5d

I’m excited to share my 1st first-authored paper, “Distinct portions of superior temporal sulcus combine auditory representations with different visual streams” (with @mtfang.bsky.social and @steanze.bsky.social ), now out in The Journal of Neuroscience!
www.jneurosci.org/content/earl...

Fig. 1. a. Visual and auditory regions of interest (ROIs). b. Responses in a combination of visual (e.g., early dorsal visual stream; Fig. 1a, middle panel) and auditory regions were used to predict responses in the rest of the brain using MVPN. c. In order to identify brain regions that combine responses from auditory and visual regions, we identified voxels where predictions generated using the combined patterns from auditory regions and one set of visual regions jointly (as shown in Fig. 1b) are significantly more accurate than predictions generated using only auditory regions or only that set of visual regions.

1 10 19

Reposted by Mark Thornton

William J. Brady @williambrady.bsky.social · 8d

The computational psych preconference is back @spspnews.bsky.social for a full day! This year's lineup:

👉theory-driven modeling: Hyowon Gweon
👉data-driven discovery: @clemensstachl.bsky.social
👉application: me
👉 panel: @steveread.bsky.social Sandra Matz, @markthornton.bsky.social Wil Cunningham

1 7 18

Mark Thornton @markthornton.bsky.social · 21d

Very excited to share @landrybulls.bsky.social's 1st lead-author preprint in my lab! Using datasets from MySocialBrain.org we measured people's beliefs about how mental states change in intensity over time, the dimensional structure of those beliefs, and their correlates: osf.io/preprints/ps... 🧵👇

4 21

Reposted by Mark Thornton

Landry Bulls @landrybulls.bsky.social · 21d

Excited to share the preprint for my 1st 1st-author manuscript! @markthornton.bsky.social and I show that people hold robust, structured beliefs about how individual mental states unfold in intensity over time. We find that these beliefs are reflected in other domains of mental state understanding.

2 6 33

Mark Thornton @markthornton.bsky.social · 25d

That said, I think departmental areas are often more political formations than intellectual ones. Having already compartmentalized ourselves, breaking down those administrative walls may result in more locally powerful subfields exerting hegemony over weaker ones, rather than interdisciplinarity.

1 1

Mark Thornton @markthornton.bsky.social · 25d

I agree that folks should be pushed to reckon with differences in intellectual traditions across subfields more. Being interdisciplinary really exposes the tedious absurdity of many discipline-specific practices. Seeing how other folks operate can be reassuring or challenging, but always enriching.

1 1

Reposted by Mark Thornton

jamelle @jamellebouie.net · 27d

condemnations of political violence are well taken but it does feel as if some
prominent political commentators are of the view that violence isn’t a part of the history of American politics and that is very much not true.

270 2K 15K

Reposted by Mark Thornton

Helena Miton @helenamiton.bsky.social · 28d

🚨 New paper alert! (slightly belated)

We (with @joshcjackson.bsky.social) suggest that complex technologies, as they need several (often many) people to use them, require innovations that distribute cognition to help regulate cognitive load and coordinate.

www.sciencedirect.com/science/arti...

Complex technology requires cultural innovations for distributing cognition

Over the last decade, new research has shown how human collectives can develop technologies that no single individual could discover on their own. How…

www.sciencedirect.com

8 28

Mark Thornton @markthornton.bsky.social · Sep 6

You may already know it because it won the Hugo a few years ago, but if not, one of my recent favorite sci-fi novels of manners was "A Memory Called Empire" by Arkady Martine.

1 1

Reposted by Mark Thornton

Buddhika Bellana @buddhikabellana.bsky.social · Sep 5

How might stories shed light on brain function? Check out this opinion piece by @alexbarnett.bsky.social and I about the DMN and "situation models" -- our understanding of the current "state of affairs" in a story (or even experience).

www.sciencedirect.com/science/arti...

On the left: an illustration from Brooke's 1904 rendition of Goldilocks and the Three Bears, where Little bear discovers their favourite chair is broken 😲. On the right, a sketch of what a corresponding "situation model" might contain.

3 20 52

Mark Thornton @markthornton.bsky.social · Sep 4

Today, SCRAP Lab returned (right) to the Path of Life Garden in Windsor, VT - the site of our first in-person get-together as a lab 5 years ago (left) - to welcome our newest member, graduate student @gabefajardo.bsky.social!

4 16

Mark Thornton @markthornton.bsky.social · Sep 4

At a glance, it seems like the new wiki is much more extensive, but there are a few jobs which are only on the old wiki. However, the final numbers from 2024 (new wiki) and 2023 (old wiki) are very similar, suggesting that using the new one exclusively is unlikely to result in a major under count.

3

Mark Thornton @markthornton.bsky.social · Sep 4

Unfortunately it would be quite difficult to reliably combine data from the two within the same year. In part because named object recognition is inherently tricky and imperfect, and in part because they have different cross-listing (across area/rank) policies that would be hard to reconcile.

1 2