Abstract
Would you look for the milk in the bathroom? Probably not, because adults are effective scene searchers. When adults are presented with scenes containing an object that does not fit the meaning of the scene (milk instead of shampoo in the bathroom), their gaze dwells longer on this object than if it was consistent, reflecting their knowledge of scene semantics. When does this semantic congruency effect emerge during development? To answer this question we recorded eye movements of three- and four-year old children (n=15) as well as adults (n=9) while they were viewing photographs of daily-life scenes in a paradigm with gaze-contingent stimulus presentation. Children showed longer dwell times for semantically incongruent than congruent objects independent of their age. This semantic congruency effect was driven by a larger number of fixations on the critical object rather than longer fixation durations. The comparison to adults revealed an increase of the semantic congruency effect in dwell times across age groups. Moreover, only for adults the average duration of fixations tended to differ between congruency conditions. As real-world validation we asked children and adults to put 52 objects into a dollhouse, equipped with only one reference object per room. Here, younger children preferred to put more objects of one type of category together (i.e., various cupboards next to the reference wardrobe in the child room) compared to older children and adults, who tended to distribute them across rooms. While the eye tracking data imply that already by the age of three, children show indices of semantic scene knowledge, the results of the dollhouse task suggest differences in developmental processes within this age group. Future studies should include children of a younger age to identify the entire range of important milestones in the development of scene knowledge.
Meeting abstract presented at VSS 2016