Event Detail

Humans, particularly children, are often likened to sponges, soaking up the patterns and regularities in their environment as they learn new concepts and skills. However, unlike a passive sponge, human capacities for curiosity, selective sampling, and intervention often alter our experience in important ways. Any complete account of human learning must explain not only what is learned from the data we experience, but also the capacity for our actions and choices to expose that information. Unfortunately, the vast majority of laboratory studies of category learning emphasize entirely passive learning situations by limiting participants’ control over the information they experience on every trial. In this talk, I will discuss recent work in my lab exploring how people search for and select information to support their own learning processes. The primary aim is to characterize (formally) the information sampling strategy that participants use to reduce their uncertainty about the world, and to examine how the active versus passive learning distinction influences category acquisition/generalization. To foreshadow, we find that participants acquire categories faster when they can select and sequence category items themselves, but that this advantage is only present when it is the learner him/herself that does the selecting. In addition, participant’s tendency to design optimally-useful queries depends, in a dynamic way, on the structure of the to-be learned categories and the space of hypotheses that the learner considers. A series of computational models (inspired by research on active learning/vision in machine learning) are evaluated which try to predict which items people will want to sample/query, and which explain the differential impact that active sampling has on the learning processes. Implications of this work for education, instructional design, as well a contemporary theories of human categorization will be entertained. (This is joint work with Doug Markant at NYU).