What We Do

Research in our lab examines a range of questions in cognitive science, from language to concepts to decision-making. Most recently we’ve been interested in addressing these question from a social perspective: how does the fact that people learn from other people — and transmit ideas through others — shape what is learned? On a methodological front, we use a combination of experimentation and computational modelling, often but not always within the Bayesian framework. You can get a good sense of what we do by checking out the papers linked to in our publication list. But if you just want a quick overview, read on...



Our research in language centres on questions of representation, evolution, and acquisition. What biases explain people's language learning, and to what extent are they domain general? What drives the difference in language acquisition abilities between adults and children? What structure do our mental linguistic representations have, and why? How is language, or any cultural product, shaped by the organisation of the world, the nature of communication, and the structure of the human mind? How is language and language learning shaped by the fact that it is a fundamentally social construct shared by social agents?

Current projects focus mostly on the dynamics of information transmission, social assumptions in language learning, statistical learning, the distributional structure of language, and semantic network structure. Previous work (to which we may return) centres on phoneme learning, grammatical representation and learning, and regularisation.


Understanding human conceptual knowledge is one of the central problems in cognitive science. We focus on a number of key questions. How is knowledge organised in the mind? How do the assumptions people make about how data is generated, especially the social assumptions about the motivations of the providers, drive their conceptual knowledge and what they do with it? What kind of inductive inferences are licenced by our conceptual knowledge, and what are not? What must be "built in" in order to explain human category learning?


Current projects involve understanding how people learn high-dimensional categories, how people identify relevant features, what kind of conceptual inferences people make based on different types of statistical sampling, and how concepts (especially social concepts) change and why. We also study concepts and meaning by looking how words are connected in our mental dictionary. We have been building a semantic network of most of the words a person knows using a word association game. If you're curious and want to help us with a few minutes of your time, go to https://smallworldofwords.org. Besides this, previous work (to which we may return) focuses on the role of labels in concept learning and learning categories with different kinds of structure.


Decision Making

Learning about the world is only one half of the problem facing an intelligent agent. The other half is taking actions within that world. How do people take action based on limited information, when operating under time pressure, or when faced with uncertainty? How do people search for information on which to make that action? How do strategies change if the world changes? We are particularly interested in how people reason in a social context or based on social assumptions, including in applied areas like medical decision-making.

Current projects involve understanding how people make decisions in ambiguous or uncertain situations with rare or extreme payoffs, how people take into account the informational value of data when searching, and how trust in information providers affects what decisions people make and what info they pass on. Previous work (to which we may return) investigates how strategies are affected by evidentiary value, number of options, and time pressure.

Other Stuff

There is a long tail of “other stuff” which usually stems out of these three areas but doesn’t properly fit into any of them. This usually includes thinking about the role and interpretation of computational and mathematical models in cognitive science, the development of statistical approaches to applied problems, and various other bits and pieces. There is no coherent plan to this: we just write about things when we have good ideas about them!

other stuff