Abstract
Visual orienting by human infants to a singleton bar oscillating in a field of similar static bars is influenced by non-motion related characteristics of the static bars such as their colors and contrast polarities implying natural variation across trials in the stimulus dimension controlling orienting (Dannemiller, 1998; 2001). With a sample of 207 infants in six experiments (7- to 11- OR 17- to 21-week-olds), two possible models for understanding how different stimulus dimensions (e.g., color, motion) interact to determine a directional orienting response were compared. Model 1: all stimulus dimensions are reduced to a single salience dimension with the location of the highest salience element controlling orienting (winner-take-all, WTA). Model 2: on a given trial one dimension (e.g., motion) dominates the orienting response while others (e.g., color) have no influence, with the dominant dimension switching across trials (dimensional switching, DS). A set size manipulation (range 2 to 28 bars) in which static bars are added equally to both sides of the visual field can distinguish the models. The WTA model predicts a decrease in orienting to the singleton moving bar because with internal noise, the likelihood that one of the static bars will produce the maximum response increases as more bars are added to the field. The DS model predicts no set size effect because trials on which orienting is dominated by non-motion related stimulus dimensions (e.g., color) will lead to chance orienting with respect to the location of the oscillating bar independently of the number of static bars added to the visual field. The DS model captured the behavior of older infants better than the WTA model with motion pop-out evident at this older age, while the WTA model better captured the behavior of the younger infants. A quantitative version of the DS model that captures this variation in stimulus-related orienting behavior will be presented. [Supported by NICHD R01 HD32927.]