After several years of work, my lab is starting to put out our first papers on learning in a unicellular organism (Stentor coeruleus).
Here we show evidence for a form of associative learning in Stentor:
www.biorxiv.org/content/10.6...
Functional MRI scans of more than 100 2-month-old infants suggest that they are capable of distinguishing among a variety of different objects. The findings challenge perceptions of cognitive development as a gradual process.
By @helenak.bsky.social
www.thetransmitter.org/cognitive-ne...
New preprint
"Human-Like Coarse Object Representations in Vision Models"
arxiv.org/pdf/2602.12486
Very happy to see "Pretending not to know reveals a capacity for model-based self-simulation", a collaboration with @chazfirestone.bsky.social and @ianbphillips.bsky.social, out in Psych. Science!
journals.sagepub.com/doi/10.1177...
🧵
My lab is looking to recruit 1-2 paid summer interns to do wet lab work on learning in a unicellular organism (Stentor coeruleus). You can apply here:
forms.gle/b47WpSobjFjo...
A new and improved version of TheoryCoder, which learns to play video games in a human-like way by synthesizing both high-level abstractions and a low-level model of game mechanics:
arxiv.org/abs/2602.00929
What's going on in aphantasia? Balaban and I recently argued it might be 'broken rendering'; but then why do normies bother rendering anything?
Monzel, Zeman & co wrote an interesting response:
www.cell.com/trends/cogni...
...and this is our response response:
www.cell.com/trends/cogni...
now accepted at ICLR! 🐺🥳🐺
arxiv.org/abs/2506.20666
Amid the rise of billion-parameter models, I argue that toy models, with just a few neurons, remain essential—and may be all neuroscience needs, writes @marcusghosh.bsky.social.
#neuroskyence
www.thetransmitter.org/theoretical-...
Even as Atte Kadoma and Isaias Ghezae are immersed in demanding PhD programs at Harvard, being resident advisor brings balance to their lives and enriches both their student experience and their scholarly pursuits.
Super great opportunity!
new: Eric Bigelow @ericbigelow.bsky.social suggests the 2 main ways of controlling LLMs (prompting & steering) can be understood as changing model beliefs (as in Bayesian belief updating)
"Belief Dynamics Reveal the Dual Nature of In-Context Learning & Activation Steering"
arxiv.org/pdf/2511.00617
Thanks Alex for the opportunity to work on this project!
Another new paper from the lab: Predictive theories like the SR imply that navigators who navigate differently should have cognitive maps which differ in predictable ways. Here we show that this holds in mouse hippocampal CA1.
www.sciencedirect.com/science/arti...
"abandoned teddy bears" category on wikimedia commons
Another fun project from @yangxiang.bsky.social. She asks the question: do people assign responsibility to personality traits in the same way that they assign reponsibility to people? The answer: sort of!
osf.io/preprints/ps...
Dr. Roosevelt is the overarching antagonist of Naoki Urasawa's 2003 manga thriller series Pluto and its 2023 anime adaptation, both of which are a reimagining adaptation of The Greatest Robot in the World arc of Astro Boy. He is an advanced supercomputer that takes the form of an immovable brown teddy bear that speaks with a child-like voice.
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...
it's not that i ever doubted that being a barista was a skilled position, but my year-long effort to try to produce a half-decent latte at home has convinced me that a good barista is easily more skilled than virtually every c-suite executive on the planet
overheard in coffee-shop convo now between two tech bros.:
"I don't want to think about AI 24/7, you know? I want to think about other stuff! Like...like...stoicism, you know?"
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
Camp:
An aesthetic style in which something appeals because of its irony. Basically it’s so bad that it’s good.
One then ought to attribute bad writing to intention. Then it’s camp, and thus good.
#KempnerInstitute researchers find #ChatGPT’s guardrails are biased against certain users based on contextual information and inferences about the user.
Research by Victoria Li, @yidachen.bsky.social & @nsaphra.bsky.social.
Check out @theregister.com's article: bit.ly/47eaANa
#AI #LLMs
Considering the compensation is good, would “Cereal murder” count?
Wow! This is incredibly cool!
Applications are now open for our #KempnerInstitute Research Fellowships!
Postdocs studying the foundations of #intelligence or applications of #AI are encouraged to apply.
Learn more and apply by Oct. 1:
bit.ly/3UvpSpe
#LLMs #NeuroAI #ML @shamkakade.bsky.social
whelp, they changed the link:
news.harvard.edu/gazette/stor...
