What Drawings can tell us About Children and Adults’ Object Recognition

• In Event: 1-031 - Children’s Drawings as Representations of Their Spatial Cognition

Integrative Statement

How do we learn what information is needed to recognize and name the world around us? In English, categories of solid objects (e.g., couch) are similar in shape, but vary in color and material; categories of nonsolid substances (e.g., yogurt) are similar in material, but vary in color and shape (Samuelson & Smith, 1999). Different information is needed for recognition of materials than for recognition of shapes. For example, tactile information is beneficial for recognizing materials (Lederman & Klatzky, 1990)—and may even be necessary for children (Perry et al., 2014). Shape, on the other hand, is readily accessibly by static visual information making solid objects easier to recognize by children. Furthermore, although we can teach artificial intelligence systems to recognize solid objects, it is nearly impossible to teach them to recognize nonsolid substances (Adelson, 2001). How do we learn what information is relevant, given that different kinds of things require such different information?

In the current study we examine children and adults’ drawings of familiar solids objects and nonsolid substances from memory. Examining these drawings allows us to assess what visual information is relevant to representations of different kinds of things and how this information changes over development. As such, this study is an important first step in understanding we recognize and name objects and substances.

Four- to six-year-old children and adults drew familiar solid objects (e.g., apple, couch) and nonsolid substances (e.g., yogurt, water). Drawings were coded for inclusion of various visual cues (e.g., shape, color, container). A separate group of adult participants, from Amazon Mechanical Turk, and a separate group of children attempted to guess the identity of each drawing. Not surprisingly, adults’ drawings (M=.39) were more accurately identified than children’s drawings (M=.08), p<.00001. More interesting were the differences in the two groups’ use of visual information in their drawings (see Fig. 1). In particular, adults were more consistent in their use of color information—even for things that could appear in a variety of colors, p=.06. For example, 100% of adults drew brown (i.e., chocolate) pudding and purple grapes while children drew pudding and grapes in a variety of colors. However, when children did use colors consistent with the dominant color in adult drawings, their drawings were much more likely to be recognized by the separate group of guessers, p<.00001. Together, these results suggest 1) that color might be more important to our object representations than previously believed and 2) that as children develop, they become more systematic and prototypical in the colors they associate with objects.

Additionally, the guessers were more accurate in identifying drawings of solid objects than nonsolid substances. Both children and adults tended to include container information for nonsolids. Drawings of nonsolids that depicted very distinct and prototypical containers (e.g., milk carton, coffee mug) led to greater recognition accuracy, p<.01. These results demonstrate that visual recognition of both solids and nonsolids—is aided by color and shape, suggesting that we may conceptualize nonsolids as more object-like than was previously thought.

Author

• Lynn K Perry, University of Miami
  Presenting Author