Processing the Parts and the Whole: Linguistic Influences in Young Children's Visual Object Recognition

• In Event: 2-160 - The More You Know: How Prior Knowledge Scaffolds Cognitive Development

Integrative Statement

Children’s worlds are filled with people and objects, yet recognizing objects and faces are slow developmental achievements, with some components not reaching mature levels until late adolescence. Although early theories suggested that recognizing objects engaged the processing of individual features whereas identifying faces engaged configural processing, more recent evidence from adults suggests that both types of processing might be engaged depending on the task constraints (e.g., Chua et al, 2015). There has been considerable research on the development of face perception, but very little on the development of everyday non-face objects, even though these processes are central to children’s success in many cognitive tasks. Here, we present evidence that the activation of prior knowledge influences how young children recognize non-face objects.

In Experiment 1, 64 typically developing children (26 girls, M=35 months, SD=2.5) completed 30 test trials where they had to find a target in a cluttered display. We used two stimulus conditions, one that should encourage feature processing by displaying individual features in random spatial locations (Figure 1A, Scrambled) and another that should encourage configural processing by displaying the same individual features in their typical spatial locations (Figure 1A, In-Place). We also used two instruction conditions, one in which children were shown the specific visual search target and had to visually remember it (Figure 1B, Visual cue) and one in which they were given the spoken name of the target (Figure 1B, Label cue). As Figure 2A shows, children were equally accurate at finding the target when they heard its name irrespective of the spatial arrangement of the features, a result that suggests recognition by features without regard to configural information. However, when children searched for a visually specified target, they were more likely to be accurate when the features were scrambled; that is, the configural information hurt the processing of the individual features. This finding fits with one account of adult findings (e.g. Murray, 2004) that suggests that configural processing can be disrupted when there is partial configural information (occluded features in place) but not when all configural information is removed (randomly placed features). Experiment 2 (N=20, 6 girls, M=37 months, SD=2.6) confirmed this hypothesis – children were unable to detect a difference in a local feature when the stimuli encouraged configural processing (Figure 2B, Upright presentation), but did so at above chance levels when the configural information was removed (Figure 2B, Upside-down presentation).

The results provide two new insights about visual object recognition in young children. First, instructions using words (that activate internal representations of category information) and those using visual presentations (that must be visually processed and remembered) yield different patterns, with word-activated representations being more feature based and visual representations being more configurally based. Second, in visual matching tasks (Experiment 1 - Visual Cue condition, and Experiment 2), children do process the configural information in non-face objects.

Authors

• Catarina Vales, Carnegie Mellon University
  Presenting Author

• Linda B Smith, Indiana University
  Non-Presenting Author