Broadly, I am interested in understanding how people learn concepts and categories. What kinds of representations and processes best capture how people learn categories, and what factors influence the acquisition of category knowledge? One line of my research has involved understanding how people solve problems such as clustering and structure learning (traditionally problems in the area of unsupervised learning), and to what extent providing supervised instances shape people’s learning and representations in these tasks.
A second area of my research has focused on how people can learn categories with many features. Learning in high-dimensional spaces should be theoretically challenging due to the curse of dimensionality, where the number of possible classification rules can grow exponentially large. Yet, humans are able to do so quite easily and part of this research is to discover why and how learning is possible. This research is driven by a combination of computational modelling using ideas from Bayesian statistics and machine learning, as well as behavioural experiments conducted through Mechanical Turk.