We show face patches implement the following code through recurrent dynamics: Detect face If (face found) Discriminate face else Continue to detect face IMHO, our paper conclusively resolves a debate that has raged since I was a graduate student, about whether face patches are specialized for processing faces or not. It turns out domain-general folks were right early on, domain-specific folks were right later in the response. So proud of @Yuelin_Shi and the entire team!

We’re hiring a new Lab Manager! I’ve loved working in this lab — super supportive environment and meaningful work in AI cognitive neuroscience Highly recommend for anyone looking to gain research experience before grad school!

Come work in LIT Lab!

Language, Intelligence & Thought lab is looking for a lab manager! This is a 2-year postbac position that will allow you to gain experience in human neuroscience, cognitive science, and AI research prior to applying to PhD programs. Express interest here: forms.gle/289sLgZdJ2bQr1Y48

🚨 Paper alert: To appear in the DBM Neurips Workshop LITcoder: A General-Purpose Library for Building and Comparing Encoding Models 📄 arxiv: arxiv.org/abs/2509.09152v1 🔗 project: litcoder-brain.github.io/

Honored that a piece I wrote made it to NYTimes. It's about how my mom's stroke changed my relationship to time, science, and nature. What a privilege to honor my mom in Modern Love. Below is a gift link. Let me know your thoughts🙏🏼 nytimes.com/2024/12/20/style…

🧩 Why do task vectors exist in pretrained LLMs? Our new research uncovers how transformers form internal abstractions and the mechanisms behind in-context learning(ICL).

🚀At @NeurIPSConf tomorrow? Don't miss @nikolasmcneal and @MainakDeb19 's poster at the @unireps workshop on the adversarial sensitivity of vision encoding models of fMRI responses! A brief teaser about what they find.

Graduate training opportunity! See thread... The Center for Research and Education in Navigation (CRaNE) is seeking a graduate student in our Cognition & Brain Sciences (CBS) Ph.D. Program in the School of Psychology at Georgia Institute of Technology.

Working on a figure and google searched "MLP images" for some inspo...

1/6 I usually don’t comment on these things, but @RylanSchaeffer et al.'s paper contains enough misconceptions that I thought it might be useful to address them. In short, effective dimensionality is not the whole story for model-brain linear regression, for several reasons: 🧵👇 (1) The spectral theory due to @canatar_a @jenellefeather @s_y_chung is misrepresented in this paper, and crucially relies on model alignment with brain data (this is already evident from their equations). In fact, they conclude that effective dimensionality alone doesn’t fully predict neural data. (2) Even in this paper’s own Figure 2 (top left panel), networks with low participation ratios, like SRNN, still achieve high neural predictivity, already indicating that participation ratio is not a unilateral predictor of whether a model will match the brain via linear regression. (3) Beyond MEC, this lack of a trend with effective dim. is also the case when looking at models in their match to macaque IT or human OTC. For example, Colin Conwell @_jacobprince_ @talia_konkle et al.’s very thorough work: biorxiv.org/content/10.1101/… shows that effective dim isn’t related to linear prediction of human OTC (cf. their Figure 5). Moreover, in macaque IT, as @apurvaratan & others have seen, if you include very predictive models of IT responses, the effective dim to linear predictivity trend also doesn’t hold. (@jenellefeather & others don’t see this trend holding up in matches to auditory cortex either.) (4) Furthermore, the models they cite participation ratios for, were also compared against several non-fitted metrics (like RSA, score distributions, simpler-than-linear mappings, etc.), which found similar conclusions that matched linear regression results, across MEC, IT, and auditory cortex.

Ev Fedorenko's Keynote at COLM. youtube.com/watch?v=8xS7tjy9… This talk is quite accessible for computer scientists interested in cognitive and neuro questions. Also touches on many of the shared themes of the two areas.

Just now finding out voxel is a portmanteau from volumetric pixel...what else have I missed?!

*TWO* job searches @EmoryPsychology ‼️ Open rank Neural Mechanisms of Behavior in Small Animal Systems apply.interfolio.com/153594 Assoc/Asst Professor, Clinical Science apply.interfolio.com/152692 And I’m recruiting a PhD student! Several opportunities to join our amazing dept ✨

Excited to have been named one of @techreview's 35 under 35! I am happy that, these days, language & human cognition are topics that the world cares deeply about (thanks to recent developments in AI). Not only are these topics impactful, they are also fun to study!

Have you ever wondered what’s going on in a baby’s brain when she looks at you? I have! So, @rebecca_saxe and I set out to investigate.

🚨New paper!🚨My first (co)first-authored paper is now out in @NatureHumBehav! We show that neural populations in the language network differ in the size of their temporal receptive windows rdcu.be/dR0sz co-led w/ @tamaregev 1/ 🧵

Thanks @NatRevNeurosci for the chance to clarify that1️⃣the language network is not monolithic (we never said it was, but happy to emphasize); and2️⃣language network boundaries don't depend on a specific 'localizer' and can be recovered from task-free data: tinyurl.com/5y3rhfrh

Happy to see this work led by @zaidzada_ now published in @NeuroCellPress! We use LLM embeddings to capture word-by-word linguistic content transmitted from the speaker's brain to the listener's brain in real-time, face-to-face conversations: cell.com/neuron/fulltext/S08…

Some thoughts on how to think about "world models" in language models and beyond: lingo.csail.mit.edu/blog/wor…