Elizabeth Spelke (Harvard): "What infants know (and don’t know)"
Abstract: To build systems that develop commonsense knowledge as fast, flexibly, and effectively as children do, it’s useful to consider both what infants know from the beginning, and what gaps in their knowledge get filled by later learning. Developmental psychologists have long addressed these questions, and we have a lot to say about what young children know when. To create machines that learn like children, however, we need a computational understanding of the origins and growth of human knowledge: an understanding of how children learn and why they learn in the ways they do. Developmental cognitive science hasn’t achieved this understanding, but I think we can get closer, through coordinated research on minds and machines. To flesh out this suggestion, I’ll talk about the origins and development of one early-emerging system of knowledge that has been studied extensively in humans and other animals, as well as in machines: the knowledge that provides our sense of place within the large-scale, navigable environment, and that supports our learning of the locations of significant objects and events.
Speaker bio: Elizabeth Spelke is the Marshall L. Berkman Professor of Psychology at Harvard University and an investigator at the NSF-MIT Center for Brains, Minds and Machines. Her laboratory focuses on the sources of uniquely human cognitive capacities, including capacities for formal mathematics, for constructing and using symbols, and for developing comprehensive taxonomies of objects. She probes the sources of these capacities primarily through behavioral research on human infants and preschool children, focusing on the origins and development of their understanding of objects, actions, people, places, number, and geometry. In collaboration with computational cognitive scientists, she aims to test computational models of infants’ cognitive capacities. In collaboration with economists, she has begun to take her research from the laboratory to the field, where randomized controlled experiments can serve to evaluate interventions, guided by research in cognitive science, that seek to enhance young children's learning.
CVPR 2020 Workshop, June 15
Minds vs. Machines: How far are we from the common sense of a toddler?
[ Ссылка ]
