like us! No one is doing computer science and math with their naked brains. Pen & paper, math notation, computer language—these are our neurosymbolic add ons! @lampinen.bsky.social , @blaiseaguera.bsky.social , @kmahowald.bsky.social all get it. Not sure why this isn’t more obvious to folks
Andrea Lathrop@cabernet.bsky.social · 4d agoThis is a very good point. It's an LLM-system, at this point, not a pure LLM, so theoretical critiques about what an LLM can or cannot do should not be taken to be predictions of what an LLM-plus-a-bunch-of-stuff can or cannot do.
What should we make of recent language model advancements in mathematics? In my new post, I reflect on long-standing cognitive debates about symbols and neural networks in light of this progress. infinitefaculty.substack.com/p/symbols-ne...
Symbols, neural networks, and mathematical intelligence
Some reflections on connectionism and the basis of higher-level cognition
infinitefaculty.substack.com
hello CogSci 2026 people! We (@adedhe.bsky.social @rbanda.bsky.social) are excited to virtually discuss and present a sneak peek at our ongoing work comparing 🐦⬛, 🖥️, 🐒, 🧒, and 🧓 (very casual about the sneak there). If this piques your curiosity, stop by! Virtual Flash Talks 46-48 @ Rio CogSci Con
That’s right! Are LLMs just stochastic parrots doing linear pattern matching? How do they compare to natural intelligences that do hierarchical reasoning? 🧵🪡
mukund choudhary@mukundc2k.bsky.social · 2w agohello CogSci 2026 people! We (@adedhe.bsky.social @rbanda.bsky.social) are excited to virtually discuss and present a sneak peek at our ongoing work comparing 🐦⬛, 🖥️, 🐒, 🧒, and 🧓 (very casual about the sneak there). If this piques your curiosity, stop by! Virtual Flash Talks 46-48 @ Rio CogSci Con
"There won’t be more discoveries, and we’ll just stop at what we already know." @cantlonlab.bsky.social www.nytimes.com/2026/07/10/c...
She Studied Ways to Make People Smarter. Then Her Grant Was Cut.
Jessica Cantlon had a grant with the U.S. Navy to explore whether people could improve spatial problem-solving skills through training. Then the program was cut.
nytimes.com
God bless America! 🇺🇸🦅 and all the wonderful scientists in it
🐸 New position paper on compositionality! 🐸 I synthesize my thoughts about the proper role/interpretation of behavior and mechanism in asking the question "Is this system (mind or machine) exhibiting compositionality?"
And written by an agent 1 generation older than the one that wrote the document sent to you in the first place.
So excited to have you here @rbanda.bsky.social and for all the cool work we're going to do together!
Ridhi Bandaru@rbanda.bsky.social · 2mo agoBtw late news but I’m now a predoc with @lexidecker.bsky.social @ the How We Learn Lab!!! 📊📑🎊🧠
If human editors can’t control who reviews science, it’s no longer peer review — it’s a rubber-stamp machine designed for volume and profit, not quality. I have no intention of attaching my name to it. So I’m out.
I wrote a piece for the Yale Review on AI and "jagged intelligence". (Note: headline was not written by me.) yalereview.org/article/mela...
Melanie Mitchell: What We Get Wrong About AI
Melanie Mitchell probes the jagged landscape of AI and its uncertain future.
yalereview.org
Btw late news but I’m now a predoc with @lexidecker.bsky.social @ the How We Learn Lab!!! 📊📑🎊🧠
New results! The prefrontal cortex uses a compact, low-dimensional “coding space”. Information can be captured in just 3–6 dimensions. The brain compresses what we remember. Low-dimensional prefrontal representations of objects during working memory doi.org/10.64898/202... #neuroscience
doi.org
Hi language science bluesky! If you have 20 minutes, PhD student Dhanya Charan (dhanya.humspace.ucla.edu) and I are recruiting volunteer participants for a 20-minute speech production study! Link here: www.labvanced.com/player.html?... Please re-skeet / share widely! Thank you!
DHANYA CHARAN – [d̪an.jaː t͡ʃa.ɾan]
dhanya.humspace.ucla.edu
Tomorrow (6/2) @ the DI Summit, find ME talk about how we might estimate the "price" of hierarchical reasoning! Pull up if you're at Princeton — 9:20 am, Robertson 016 Also l'll be here til the 3rd so hmu if you want to hang out!
I think it's important for neuroscientists to reinforce this too. It's the first thing I teach in my cog neuro class. A thing isn't more real just because it's found in the brain. Every psychological thing must be in the brain at some level
New opinion piece on the interface between research on concepts and categories in minds vs. in neural network LMs! I take the position that there is much to be learned from this interface (e.g., learning about concepts from language alone) and outline some directions for future.
Our *Monster paper* is finally out. Begun instead of data collection during covid for Melis's PhD, expanded during Jenni's PhD, & finally now published. Some papers are worth the wait onlinelibrary.wiley.com/doi/10.1111/...
Capturing Joint Attention: Theoretical Foundations and Their Implications for the Link Between Joint Attention and Vocabulary Growth
Despite decades of research, we still know less than we would like about the association between joint attention (JA) and language acquisition. One reason for this is that we still have not agreed ho...
onlinelibrary.wiley.com
"Codex grew programmatic policies with no neural nets: max score on Breakout, and SOTA-level scores on MuJoCo. Maybe heuristics were not too weak. Maybe they were just too expensive to maintain. Maybe it's the next paradigm." trinkle23897.github.io/learning-bey...
Learning Beyond Gradients
Learning Beyond Gradients
trinkle23897.github.io
A very interesting piece by @henryconklin.bsky.social et al. on how brains and LLMs acquire abstract, meaningful internal representations by compressing the information they receive. Similar ideas in my book "How we learn"! henryconkl.in/posts/i-am-a... Full paper: henryconkl.in/posts/llms-a...
i am, myself, a lossy compression – h
a companion piece to our ICLR 2026 Paper: Learning is Forgetting: LLM Training As Lossy Compression
henryconkl.in
A unique neural signature of long-term memory encoding from EEG inter-electrode correlation doi.org/10.1162/IMAG... #neuroscience
A unique neural signature of long-term memory encoding from EEG inter-electrode correlation
Abstract. Classic memory models proposed that the encoding process involved in visual working memory (VWM) controls the bandwidth of encoding in long-term memory (LTM). Behaviorally, VWM and LTM accur...
doi.org
Excellent thread "We risk all of science if we rush to build 'AI Scientists' before we understand the value of human science." ⬇️
M.J. Crockett@mjcrockett.bsky.social · 3mo agoLast weekend I received the Troland Research Award from the @nationalacademies.org. I’m so grateful to the communities who made this work possible. It's strange to receive this award at a time when much of the work being recognized is not eligible for federal funding. My remarks 👇 and a 🧵>>
Entering the grand halls of the NAS I felt a potent mixture of awe and fury. Science is so amazing. Not just for its material discoveries, but also for what it reveals about human values and our ability to know the world *together*. Shame on everyone who is complicit in this crisis. >>
Our conclusion: it is! We think that innateness can be a useful "orienting principle" for discussions of language learning. BUT, we don't think that classic syntactic content nativism is still a viable proposal.
I want to do the starters but I have deadliiiines
