Title: Using crowd sourced similarity ratings and machine vision to examine the structure of children’s drawings
Legend: Examples of nearest neighbors determined by crowd sourcing and machine vision approaches (top two figures). Bottom two figures show the correlation between the predicted age and actual age of the individual producing the drawing based on crowd sourcing and machine vision. Red dots represent drawings produced by girls, blue dots represent drawings produced by boys.
Citation: Rosengren, K. S., Caceres, P.A., Jensen, C., Kirkorian, H., Rogers, T., Travers, B. G., Rosengren, K. S., (In preparation). Using crowd sourced similarity ratings and machine vision to examine the structure of children’s drawings.
Abstract: Children’s drawings have long been used to assess aspects of general cognitive functioning, intelligence, perceptual motor development, and even socio-emotional development. The goal of the current study was to examine the structure of children’s drawings using crowd-sourced human similarity judgments and machine vision approaches. The goal is to determine whether either or both of these techniques can be used to quickly and efficiently estimate drawing quality or characteristics of the person who produced the drawing. Both approaches effectively capture the similarity structure of different drawings, but crowd sourcing provides a better estimate of age related changes in children’s drawings.
About the Lab: Rosengren’s Cognition in Action Laboratory in the Waisman Center focuses on children’s cognitive and motor development, with a particular emphasis on how cognitive and motor factors interact in everyday tasks. Drawing behavior is one example of behaviors examined in the lab that involve a dynamic interaction between young children’s planning and executive skills on the one hand and children’s motor control on the other. Other behaviors investigated include children’s action errors, behaviors where young children attempt to pick up objects in a photograph or attempt to perform an action on object that is too small to accommodate the action (scale errors). By studying these behaviors we hope to gain insight into factors that influence the development of young children’s representations of symbols, objects and themselves.