Infants learn less from a televised demonstration than from a live demonstration, the video deficit effect. The present study employs a novel approach, using touch-screen technology to examine 15-month-olds' transfer of learning. Infants were randomly assigned either to within-dimension (2D/ 2D or 3D/3D) or cross-dimension (3D/2D or 2D/3D) conditions. For the within-dimension conditions, an experimenter demonstrated an action by pushing a virtual button on a 2D screen or a real button on a 3D object. Infants were then given the opportunity to imitate using the same screen or object. For the 3D/2D condition, an experimenter demonstrated the action on the 3D object, and infants were given the opportunity to reproduce the action on a 2D touch-screen (and vice versa for the 2D/3D condition). Infants produced significantly fewer target actions in the cross-dimension conditions than in the within-dimension conditions. These findings have important implications for infants understanding and learning from 2D images and for their using 2D media as the basis of actions in the real world.

No abstract

Imitation plays a critical role in social and cognitive development, but the social learning mechanisms contributing to the development of imitation are not well understood. We developed a new imitation task designed to examine social learning mechanisms across the early childhood period. The new task involves assembly of abstract-shaped puzzle pieces in an arbitrary sequence on a magnet board. Additionally, we introduce a new scoring system that extends traditional goal-directed imitation scoring to include measures of both children's success at copying gestures (sliding the puzzle pieces) and goals (connecting the puzzle pieces). In Experiment 1, we demonstrated an age-invariant baseline from 1.5 to 3.5 years of age, accompanied by age-related changes in success at copying goals and gestures from a live demonstrator. In Experiment 2, we applied our new task to learning following a video demonstration. Imitation performance in the video demonstration group lagged behind that of the live demonstration group, showing a protracted video deficit effect. Across both experiments, children were more likely to copy gestures at earlier ages, suggesting mimicry, and only later copy both goals and gestures, suggesting imitation. Taken together, the findings suggest that different social learning strategies may predominate in imitation learning dependent upon the degree of object affordance, task novelty, and task complexity.

Research asserting that the visual system instantiates a global closure heuristic in contour integration has been challenged by an argument that behaviorally-detected closure enhancement could be accounted for by low-level local mechanisms driven by collinearity or "good continuation" interacting with proximity. The present study investigated this issue in three experiments. Exp. 1 compared the visibility of closed and open contours using circles and S-contours from low to moderately high angles of path curvature in a temporal alternative-forced choice task. Circles were more detectable than S-contours, an effect that increased with curvature. The closure enhancement observed can, however, be explained by the fact that circles contain more 'contiguity' than S-contours. Additional tests added discontinuities to otherwise closed paths to control for the effects of good continuation and closure independently. Exp. 2 compared the visibility of incomplete circles (C-contours) and S-contours derived from the full circles and S-contours in Exp. 1. Exp. 3a compared the visibility of arc pairs arranged in an enclosed position similar to "()" and a non-enclosed position similar to ")(". Results consistently showed enhanced visibility of contour configurations enclosing a region even after controlling for differences in contiguity and changes of curvature direction. A control test (Exp. 3b) demonstrated that the gap in the contours of Exp. 3a was too large to be bridged by local-level collinearity/proximity alone. The combination of good continuation and proximity alone does not explain the closure effects observed across these tests, as demonstrated through the application of a Bayesian model of collinearity and proximity (Geisler et al., 2001) to the stimuli in Exps. 3a and 3b. These results argue for the presence of a global closure-driven contour enhancing mechanism in human vision.

Visual contour detection is enhanced by grouping principles, such as proximity and collinearity, which appear to rely on horizontal connectivity in visual cortex. Previous experiments suggest that children require greater proximity to detect contours and that, unlike adults, collinearity does not compensate for their proximity limitation. Over two experiments we test whether closure, a global property known to facilitate contour detection, compensates for this limitation. Adults and children (3-9 years old) performed a 2AFC task; one panel contained an illusory contour (closed or open) in visual noise, and one only noise. The experiments were identical except proximity was doubled in Exp. 2, enabling shorter-range spatial integration. Results suggest children are limited by proximity, and that closure did not reliably improve their performance as it did for adults. We conclude that perceptual maturity lags behind anatomy within this system, and suggest that slow statistical learning of long-range orientation correlations controls this disparity.