I am an incoming assistant professor at Brown University in the CoPsy↗︎ department. I am recruiting graduate students, postdocs, and a lab manager for the 2027 academic year. I am currently a postdoc at Stanford with the Cognitive Tools Lab↗︎ , and work closely with the Computational Cognitive Science↗︎ and LINGO↗︎ labs at MIT.
My research asks how human minds pull off the computational feat of using language. People seem to learn language from very little data, all things considered. Our brains run on less energy than a laptop. How do we learn and use language so efficiently? How do we figure out what's useful for understanding someone else at any given time? How can we learn totally new things from language? Many of these questions intersect with broader questions about how people decide what they believe, or what is useful to think about, in any given situation. Language is so rich, and can express so much, that it begs the question of how we can possibly use it effectively. I am also interested in how minds give rise to more subjective aspects of language. How do we decide how to tell a fictional story? How does language make us feel emotions?
I look for approaches that can scale up our theoretical and empirical picture of how people use language. This includes human experiments and a wide range of methods from computational cognitive science, including probabilistic and planning models, program synthesis, and machine learning approaches. I’m also a writer. I love a heady and intimate sentence, and would love to explain even a sliver of what we get out of ones as rich and unruly as these ↗︎.
liowong@stanford.edu  /  Google Scholar  /  Github  /  writing and clocks  /  signal hill
Research Areas
How do people learn to understand language with so little language experience?
- From words to worlds: bridging language and thought. [Thesis, 2024][SPP, 2023]
[Glushko Dissertation Talk @ CogSci] [Simons Talk @ Berkeley] - How do people understand language about someone's knowledge and beliefs? [NAACL, 2025]
- How do people understand language about physical scenes and events? [CogSci, 2023]
- How do people understand language about actions and goals? [ICML Theory of Mind, 2023]
- How do people understand vague adjectives in context? [CogSci, 2023]
How do people reason about general problems (like language) with so few computational resources?
- Modeling open-world cognition as on-demand synthesis of probabilistic models.
[CogSci, 2025][Talk @ CogSci] - How do people reason about new games? [Preprint, under review]
- How do people build ad-hoc theories of another agent's mental states? [ACL Findings, 2025]
- How can agents build ad-hoc environment models for planning and action? [ICLR, 2024]
How do we build more robust and interpretable AI systems that talk to and work with people?
- How should we design AI systems for long-term collaboration? [CDPS, 2026]
- Do current AI systems figure out when to redirect dangerous medical questions? [ACL Findings, 2026]
- Are current AI systems good at communicating medical information? [ICML, 2025]
- How should we design AI systems that think with and alongside us? [Nature HB, 2024]
- How do people use AI systems to solve college-level math problems? [PNAS, 2024]
- How do people communicate with each other about abstract reasoning tasks? [NeurIPS, 2022]
How do people learn new concepts from language? How do people learn new concepts at all?
- How might we learn new concepts through the process of learning new words? [ICML, 2021]
- How might we use language to predict which abstractions will be useful? [ICLR, 2024]
- Do people change the words they use to reflect concepts they have in mind? [CogSci, 2023]
- How do writing systems change to reflect the structure of a language?
[CogSci, 2024][CogSci, 2025] - How might we learn new abstract concepts over simpler mental primitives? How can we learn a more complex programming language on top of a simpler one?
[PLDI, 2021] [arxiv-extended-version] - How can we learn new abstract concepts — much more efficiently? [POPL, 2023]