Abstract

Human infants acquire various cognitive abilities such as self/other cognition, imitation, and joint attention in the first few years of life. Although developmental studies have revealed the behavioral changes in infants, underlying mechanisms for the development are not fully understood yet.

We hypothesize that contingency learning plays a key role in infant development and have proposed several computational models to verify our hypothesis. Contingency is defined as a conditional probability between an action and perceptual states. For example, the visual state of our own body exhibits perfect contingency because it always produces the same state change in response to a certain motor command. In contrast, the body of other individuals produces lower contingency due to an influence of the context. Among various responses of others, however, social activities such as joint attention produce relatively higher contingency than non-social behaviors, which becomes a clue for the acquisition of social cognition.

My talk will present our robotics studies investigating how infants acquire the ability of self/other cognition, goal-directed actions, and the understanding of others’ goal. The importance of caregivers’ scaffolding for infant development is also demonstrated in our human-robot interaction experiments. Furthermore, potentials of our computational approach to understand the mechanism of autism spectrum disorders (ASD) will be explained. Our research supports a recent hypothesis that ASD is characterized by a difficulty in learning sensorimotor contingency rather than in social interaction.

(See http://cnr.ams.eng.osaka-u.ac.jp/~yukie/ for more details about my research)