Children's coding of human action: cognitive factors influencing imitation in 3-year-olds

We used imitation as a tool for investigating how young children code action. The study was designed to examine the errors children make in re-enacting manual gestures they see. Thirty-two 3-year-old children served as subjects. Each child was shown 24 gestures, generated by systematically crossing four factors: visual monitoring, spatial endpoint, movement path, and number of hands. The results showed no difference as a function of whether the children could visually monitor their own responses. Interestingly, children made significantly more errors when the adult's action terminated on a body part than they did when the same movement terminated near the body part. There were also significantly more errors when the demonstrated act involved crossing midline than when it did not, and more errors when it involved one hand rather than two hands. Our hypothesis is that human acts are coded in terms of goals. The goals are hierarchically organized, and because young children have difficulty simultaneously integrating multiple goals into one act they often re-enact the goals that are ranked higher, which leads to the errors observed. We argue that imitation is an active reconstruction of perceived events and taps cognitive processing. We suggest that the goal-based imitation in 3-year-olds is a natural developmental outgrowth of the perceptual-motor mapping and goal-directed coding of human acts found in infancy.