I am interested in how children, adults, and machines learn to represent the world around them. To explore this question, I run experimental studies and write computational models.
I am currently at Daphna Buchsbaum's CoCoDevLab, where we are working on new experimental methods to uncover children's developing categories.
Before, I worked at the University of Warwick, trying to find the algorithm underlying human approximative inference with Adam Sanborn, Nick Chater, and the rest of the SAMPLING group.
I did my Ph.D. with Chris Lucas at the University of Edinburgh. In my thesis, I explored the representational principles that allow humans to generalize functions. My research showed that when faced with continuous-valued relationships, people learn abstract features of the data, such as the type or the variability of the function. These abstract features can then be transferred to inform their generalizations in subsequent tasks. Finally, my thesis showed that humans are perceptive to, and sometimes generalize according to compositional features inherent in the data.