Kanishka Misra 🌊

I will be giving a short talk on this work at the COLM Interplay workshop on Friday (also to appear at EMNLP)!

Will be in Montreal all week and excited to chat about LM interpretability + its interaction with human cognition and ling theory.

On my way to #COLM2025 🍁

Check out jessyli.com/colm2025

QUDsim: Discourse templates in LLM stories arxiv.org/abs/2504.09373

EvalAgent: retrieval-based eval targeting implicit criteria arxiv.org/abs/2504.15219

RoboInstruct: code generation for robotics with simulators arxiv.org/abs/2405.20179

Language Models Fail to Introspect About Their Knowledge of Language There has been recent interest in whether large language models (LLMs) can introspect about their own internal states. Such abilities would make LLMs more interpretable, and also validate the use of s...

I’m at #COLM2025 from Wed with:

@siyuansong.bsky.social Tue am introspection arxiv.org/abs/2503.07513

@qyao.bsky.social Wed am controlled rearing: arxiv.org/abs/2503.20850

@sashaboguraev.bsky.social INTERPLAY ling interp: arxiv.org/abs/2505.16002

I’ll talk at INTERPLAY too. Come say hi!

Heading to #COLM2025 to present my first paper w/ @jennhu.bsky.social @kmahowald.bsky.social !

When: Tuesday, 11 AM – 1 PM
Where: Poster #75

Happy to chat about my work and topics in computational linguistics & cogsci!

Also, I'm on the PhD application journey this cycle!

Paper info 👇:

Also, I’m on the look out for my first PhD student! If you’d like to be the one, please reach out to me (dms/email open) and we can chat!!

@jessyjli.bsky.social and @kmahowald.bsky.social are also hiring students, and we’re all eager to co-advise!

I’ll also be moderating a roundtable at the INTERPLAY workshop on Oct 10 — excited to discuss behavior, representations, and a third secret thing with folks!

On Language Models' Sensitivity to Suspicious Coincidences Humans are sensitive to suspicious coincidences when generalizing inductively over data, as they make assumptions as to how the data was sampled. This results in smaller, more specific hypotheses bein...

2. @sriramp05.bsky.social on LMs and Suspicious Coincidences at the PragLM workshop poster session on Friday, (work w/ me, @kmahowald.bsky.social, and @eunsol.bsky.social)

Paper: arxiv.org/abs/2504.09387

Both Direct and Indirect Evidence Contribute to Dative Alternation Preferences in Language Models Language models (LMs) tend to show human-like preferences on a number of syntactic phenomena, but the extent to which these are attributable to direct exposure to the phenomena or more general propert...

Traveling to my first @colmweb.org🍁

Not presenting anything but here are two posters you should visit:

1. @qyao.bsky.social on Controlled rearing for direct and indirect evidence for datives (w/ me, @weissweiler.bsky.social and @kmahowald.bsky.social), W morning

Paper: arxiv.org/abs/2503.20850

All of us (@kanishka.bsky.social @kmahowald.bsky.social and me) are looking for PhD students this cycle! If computational linguistics/NLP is your passion, join us at UT Austin!

For my areas see jessyli.com

We'll all be attending #COLM2025 -- come say hi if you are interested in working with us!!

Separate tweet incoming for COLM papers!

Characterizing the Role of Similarity in the Property Inferences of Language Models Property inheritance -- a phenomenon where novel properties are projected from higher level categories (e.g., birds) to lower level ones (e.g., sparrows) -- provides a unique window into how humans or...

Understanding how internal representations drive conceptual behavior in LMs

E.g., arxiv.org/abs/2410.22590

Vision-and-Language Training Helps Deploy Taxonomic Knowledge but Does Not Fundamentally Alter It Does vision-and-language (VL) training change the linguistic representations of language models in meaningful ways? Most results in the literature have shown inconsistent or marginal differences, both...

Role of language (vs. other modalities) in learning meaning-sensitivities

E.g., arxiv.org/abs/2507.13328

Generating novel experimental hypotheses from language models: A case study on cross-dative generalization Neural network language models (LMs) have been shown to successfully capture complex linguistic knowledge. However, their utility for understanding language acquisition is still debated. We contribute...

Using neural networks to generate experimental hypotheses about language acquisition in scenarios where hypothesis-spaces are intractable

E.g., arxiv.org/abs/2408.05086

Language Models Learn Rare Phenomena from Less Rare Phenomena: The Case of the Missing AANNs Language models learn rare syntactic phenomena, but the extent to which this is attributable to generalization vs. memorization is a major open question. To that end, we iteratively trained transforme...

"Controlled Rearing" of LMs to understand the role of input in acquiring linguistic generalization

E.g., arxiv.org/abs/2403.19827, arxiv.org/abs/2503.20850

Picture of the UT Tower with "UT Austin Computational Linguistics" written in bigger font, and "Humans processing computers processing human processing language" in smaller font

The compling group at UT Austin (sites.utexas.edu/compling/) is looking for PhD students!

Come join me, @kmahowald.bsky.social, and @jessyjli.bsky.social as we tackle interesting research questions at the intersection of ling, cogsci, and ai!

Some topics I am particularly interested in:

Sigmoid function. Non-linearities in neural network allow it to behave in distributed and near-symbolic fashions.

New paper! 🚨 I argue that LLMs represent a synthesis between distributed and symbolic approaches to language, because, when exposed to language, they develop highly symbolic representations and processing mechanisms in addition to distributed ones.
arxiv.org/abs/2502.11856

Hehe but really — unifying all of them + easy access = yes plssss

Friendship ended with minicons, glazing is my new fav package!

God’s work 🙏

GitHub - aaronstevenwhite/glazing: Unified data models and interfaces for syntactic and semantic frame ontologies. Unified data models and interfaces for syntactic and semantic frame ontologies. - aaronstevenwhite/glazing

I've found it kind of a pain to work with resources like VerbNet, FrameNet, PropBank (frame files), and WordNet using existing tools. Maybe you have too. Here's a little package that handles data management, loading, and cross-referencing via either a CLI or a python API.

Would you say it’s a dead area right now? (Ignoring the podcasts)

Love to start the day by mistakenly stumbling onto hate speech against south-asians ty internet

Accepted at #NeurIPS2025! So proud of Yulu and Dheeraj for leading this! Be on the lookout for more "nuanced yes/no" work from them in the future 👀

Abstract deadline changed to *December 1, 2025*

📣@futrell.bsky.social and I have a BBS target article with an optimistic take on LLMs + linguistics. Commentary proposals (just need a few hundred words) are OPEN until Oct 8. If we are too optimistic for you (or not optimistic enough!) or you have anything to say: www.cambridge.org/core/journal...

Happy and proud to see @rjantonello.bsky.social’s work awarded by SNL!

Density plot with X axis being probability of text being synthetic from an AI detector model. Plots show that GPT4.1 outputs are assigned high probability of being AI text, but GPT5 outputs are assigned low probability of being AI text.

Exhibit N on how synthetic text/AI detectors just don't work reliably. Generating some (long) sentences from GPT4.1 and GPT5 with the same prompt, the top open-source model on the RAID benchmark classifies most GPT4.1 outputs as synthetic and most GPT5 as not synthetic.

Title: The Cats of Cogsci. Two cats, Coco and Loki, with Northwestern Cognitive Science logo in the background. Coco is sitting on some books and Loki is holding an apple. Both are wearing glasses b/c they are academics.

Loki the cat has his paw on a laptop; text "Remember to add Cog Sci 110 to you shopping cart now in Caesar so that you're ready to enroll come September 12th!"

Super happy with Cogsci program assistant Chris Kent's work for our college Instagram feed. Glad I could get our Loki featured to advertise my class.

Excited to speak alongside such an illustrious set of speakers!

Composition with Grid A few months into the pandemic, my wife and I adopted a new pastime: we would complete the New York Times crossword puzzle every day.

Lovely write up by @ksetiya.bsky.social on @rkubala.bsky.social’s
piece on the art of crosswords! Come for Robbie, stay for Sondheim crossword quotes. ksetiya.substack.com/p/compositio...