Abstract
Infants are confronted with a visual world containing a wealth of colors, textures and patterns. How do infants segment objects from backgrounds in static displays? Brady and Kersten (2003) showed that adults could recognize a novel object they had viewed repeatedly against changing 3D backgrounds, suggesting shape regularity enabled segmentation. Here we use a habituation paradigm to investigate whether shape regularity alone enables segmentation in 5.5 month-old infants. Unlike previous studies of infant’s segmentation abilities that used moving displays, we used static displays. The novel objects viewed by infants were 3D objects modeled after Brady and Kersten’s digital embryos; they were simplified to accommodate infants developing vision. A priori preferences for the novel objects were determined in a pretest where infants viewed them on a white background. During a habituation phase, infants viewed one of the novel objects presented on eight 2D plaid backgrounds, repeated up to three times. On each trial the habituation object was accompanied by a labeling phrase ("Look at the toma(modi). Do you see the toma(modi)?"). At post-test, infants viewed the habituation object and the other object side by side on a white background while hearing the habituation object’s paired phrase. Infants looked longer at the habituation object compared to the other object at post-test, p <0.05, indicating they had segmented it correctly associated its phrase during habituation. Furthermore, a pre-post-test comparison showed that infants’ object preferences were modulated by the habituation trials, p <0.03. This novel finding indicates that in 5.5 month-old infants shape regularity alone enables object segmentation in static displays, adding substantially to the evidence regarding how infants segment visual scenes. A second study is underway examining whether these results can be replicated with 3D backgrounds.
Meeting abstract presented at VSS 2013