Sam Nastase

nature.comBehaviorally-relevant features of observed actions dominate cortical representational geometry in natural vision - Nature CommunicationsHan et al. show that behavioral judgments of action meaning better capture brain activity than models of semantic and visual content across much of cortex, including the action observation network and...

New perspective piece out in @cp-neuron.bsky.social with Zaid Zada, @adelegoldberg.bsky.social, and Uri Hasson! We try to articulate some of our excitement about LLMs and discuss what kinds of insights they might provide into the neural computations supporting natural language in the human brain.

Sam Nastase reposted

Kathy Garcia

@gkathy.bsky.social

What dimensions organize how we see social interactions? We learn them from 53,000+ human judgments of naturalistic social clips. They explain 2× the variance of pre-labeled annotations & better predict visual cortex, EBA, and pSTS New preprint w @lisik.bsky.social 🧵 osf.io/preprints/ps...

Sam Nastase reposted

Jane Han

@jane-han.bsky.social

Finally the ‘new’ paper in Nature Communications! 🧠🎬 @samnastase.bsky.social @haxbylab.bsky.social Action purpose & social content explain cortical representational geometry better than visual & word-embedding models, even in ventral temporal cortex. www.nature.com/articles/s41... 1/6 🧵👇

Sam Nastase reposted

Elliot Murphy

@elliot-murphy.bsky.social

4 Sept

Out today - a brief commentary on this paper from @samnastase.bsky.social, @adelegoldberg.bsky.social and others. 🧵[1/8] arxiv.org/abs/2609.03160

arxiv.orgNo country for old linguists: LLM-brain alignment underdetermines neural computationNastase et al. (2026) argue that large language models (LLMs) may illuminate language processing because both rely on distributed, context-sensitive representations shaped by statistical learning. The...

Sam Nastase@samnastase.bsky.social

Sam Nastase reposted

Harrison Ritz

@hritz.bsky.social

27 Aug

Our task-switching paper is now out at Current Biology! www.cell.com/current-biol... We find that that our brains reset to a task-neutral state between trials, providing flexibility when the upcoming task is uncertain. RNNs also learn this strategy, but only when trained to switch tasks.

We’re recruiting a full-time lab manager to join the Shared Minds Lab at USC! This will be a great opportunity for someone who wants to get hands-on experience with research before starting a PhD program in psychology or neuroscience. More below:

This will be an fMRI-forward position. We need someone who can help lead the data collection effort on intensive, naturalistic fMRI paradigms for natural language, interactive conversations, and social-narrative videogame play.

usccareers.usc.eduLab Assistant (Research Lab Tech I) at USCLearn more about applying for Lab Assistant (Research Lab Tech I) at USC

Applications are being reviewed on a rolling basis and will be reviewed until the position is filled. To apply, please provide: 1. CV/Resume 2. Cover Letter Apply here: usccareers.usc.edu/job/los-ange...

shared-minds.github.ioShared Minds LabHow do individual brains converge on a shared understanding of the world? Explore our research at USC.

The Shared Minds Lab led by PI Sam Nastase at the University of Southern California (shared-minds.github.io) is accepting applications for a full-time lab manager / research assistant to begin in Fall 2026.

Sam Nastase reposted

Neuron

@cp-neuron.bsky.social

14 Aug

dlvr.itUnifying the structures of language in a neural population codeHow does language emerge from neural computations in the human brain? In stark contrast to symbolic theories of language, large language models demonstrate how the structures of language can be unified in a high-dimensional neural population code.

Finally, we argue that the apparent complexity of LLMs should not discourage us from trying to understand them. They are not black boxes! They are fully transparent—and we hope the explosion of interpretability work continues to cross-pollinate with computational neuroscience.

These models are also deceptively elegant in important ways. Their complexity largely reflects the complexity of their environment. The fact that they can pull this off with “just a bunch of matrix multiplications” is precisely what makes them so relevant for human neuroscience.

For example, years of neuroimaging work tried to disentangle the cortical bases of syntax and semantics based on their separability according to linguistic theory—but more recent work indicates that syntax and semantics are highly coextensive throughout the language network. Why?

In LLMs, syntax and semantics are processed through the same neural mechanisms and both are encoded geometrically in a neural population code. If the brain follows a similar strategy, we should not be surprised that syntax/semantics are co-localized and functionally entangled.

To the LLM, these are all just patterns in context—it will learn them insofar as they are *useful* for producing natural language. This context-first approach to language renders different types of linguistic structure learnable by a simple, general-purpose learning algorithm.

Later in the paper, we walk through some examples of empirical work in speech and language neuroscience where LLMs might provide some insights 🧠

The model learns that certain verbs take a regular past tense while others have irregular forms, but it doesn’t process these two classes in a fundamentally different way; it doesn’t turn on a “rule” for one type of input and memorize others.

Certain patterns are highly regular; these are the kinds of patterns that are more easily detected and have been richly described by linguists. Other patterns are more abstract, graded, or multidimensional, and therefore harder to describe.

This dense, geometric representational format also yields a surprisingly powerful form of generalization: LLMs can interpret (and produce) novel linguistic contexts by interpolating the meaning of new locations in this geometric landscape (within the bounds of their prior experience).

Principle 2 also has important implications: LLMs do not rely on any language-specific learning machinery; rather, LLMs reframe the problem of language learning as a problem of learning patterns in context.

Principle 2: They replace rule-based learning with a generic, self-supervised, context-driven learning algorithm for reproducing the statistical structure of real-world language.

Principle 1 has important implications: for example, it helps us understand how many seemingly different structures of language (e.g., syntax trees, semantic categories) can be unified in a geometric lingua franca that supports neural computation.

We argue that neural network models for language—like LLMs—have ultimately outstripped symbolic, rule-based models of language processing by appealing to two basic principles of biological learning:

Principle 1: They encode all of the structures of natural language (syntax, semantics, pragmatics, etc.) in the geometry of a continuous, high-dimensional embedding space—i.e., a neural population code.