Andrea de Varda
Postdoctoral fellow, Department of Brain and Cognitive Sciences & McGovern Institute for Brain Research, MIT — 📧 Email: devar_ag ✶ mit ✹ edu
K. Lisa Yang ICoN Fellow · Ev Fedorenko's Language Lab & Computational Psycholinguistics Lab
About
I am a postdoctoral fellow at MIT, working with Evelina Fedorenko and Roger Levy. I study language processing, and increasingly higher-level cognition more broadly, in humans and large language models. Some broad questions I am interested in are: what are the defining features of language as a cognitive system—properties that hold across diverse languages, and between artificial and biological language systems? What makes some reasoning domains and problems hard? How is higher-level cognition organized in minds and models, and are there any privileged features that tend to emerge across systems? My work combines computational modeling with behavioral experiments and brain imaging, and I use language models both as tools for cognitive science and as scientific objects in their own right.
I received my Ph.D. from the University of Milano-Bicocca, where I worked with Marco Marelli on multilingual neural language models and their relevance for cognitive science. Before that, I did a M.Sc. in Cognitive Science at the University of Trento (CIMeC).
- Language and thought in humans and LLMs
- Language in the brain
- Multilingual models
- Interpretability
- Psycholinguistics
- Iconicity & non-arbitrariness
Selected papers
Behavioral and brain responses to language reflect different levels of linguistic representation
Multilingual computational models capture a shared meaning component in brain responses across 21 languages
The cost of thinking is similar between large reasoning models and humans
A distributional model of concepts grounded in the spatial organization of objects
A Cross-Modal and Cross-lingual Study of Iconicity in Language: Insights From Deep Learning
Full list on the Publications page.
News
- Visiting Cory Shain's CLiMB lab at Stanford (September–October 2026).
- Two new preprints: Modular cognitive architecture emerges in large language models (led by Pengrui Han, with Jacob Andreas and Ev Fedorenko) and Behavioral and brain responses to language reflect different levels of linguistic representation (with Yevgeni Berzak, Ev Fedorenko, and Roger Levy).
- Invited talks at the CogSci 2026 workshop Laying the foundations for foundation models of cognition (Rio de Janeiro), the Deep Linguistic Modeling colloquium at Heinrich-Heine-Universität Düsseldorf, and the MilaNLP lab at Bocconi University.
- Daria Kryvosheieva's paper Different types of syntactic agreement recruit the same units within large language models is out at ACL 2026.
- I was awarded a Gemini Academic Program Award from Google DeepMind.
- I gave a keynote at the Multilingual Minds & Machines Meeting (Radboud University, Nijmegen); talks at the University of Amsterdam and the University of Milano-Bicocca.
- My dissertation received the Glushko Dissertation Prize from the Cognitive Science Society.
- Large-scale collaboration on AI-assisted vs. human-only teams in assessing research reproducibility (Brodeur et al.) published in PNAS.
Older news
- Three letters on The cost of thinking and our replies appeared in PNAS. See the exchange on the publications page.
- I gave a talk on language and reasoning in models and humans at the Language and Cognition series, Harvard.
- Patrick J. McGovern Travel & Technology Award (MIT); talk at INRIA Lille.
- Talk on compositionality and reasoning at the Workshop on Neural-Symbolic Reasoning, University of Warsaw.
- IconicITA, iconicity ratings for the Italian affective lexicon, published in PLOS One.
- The cost of thinking is similar between large reasoning models and humans published in PNAS. Covered by MIT News.
- Journal Club piece in Nature Reviews Psychology on learning meaning from latent patterns in language use.
- Correspondence in Nature Human Behaviour on the high variability of LLMs' analogical reasoning.
- Started as a postdoctoral fellow at MIT, working with Ev Fedorenko and Roger Levy.
- Defended my Ph.D. at the University of Milano-Bicocca.
- I was awarded the K. Lisa Yang Integrative Computational Neuroscience (ICoN) Fellowship at MIT.