Preschoolers’ information search in a question-asking and spatial exploration game.

• In Event: 2-193 - Information Gain as a Guide for Decision-Making in Early Exploratory Learning

Integrative Statement

The informativeness of children’s information search strategies has traditionally been investigated using variants of the 20-questions game, in which the goal is to identify a target object among a given set (e.g., “What is Albert’s favourite animal?”) by asking as few yes-no questions as possible. This research consistently shows that the questions children generate tend to be uninformative until about age ten (Herwig, 1982; Mosher & Hornsby, 1966; Ruggeri & Feufel, 2015; Ruggeri et al., 2015, 2016). Understanding children’s question-asking abilities is crucial because questions are an incredibly powerful tool to learn about the world (Chouinard, Harris, & Maratsos, 2007). However, because questions’ informativeness heavily depends on children’s emerging verbal and categorisation skills, relying exclusively on question-asking paradigms to study the effectiveness of children’s information search strategies is likely to paint an incomplete picture, especially when studying younger children.

In this project, we aim to extend the investigation of children’s information search strategies beyond the traditional 20-questions game. We therefore developed a novel maze task, structurally and computationally identical to the hierarchical version of the 20-questions game (see Ruggeri et al., 2016), to investigate whether spatially representing a task’s hypothesis space can boost young children’s information search strategies. Children (4- to 8-year-olds, N = 111) completed a 20-questions game and a maze task (order counterbalanced). In the 20-questions game, children had to identify which of eight candidate monsters differing across three hierarchically organised features (shape, pattern, colour) could activate a special machine. In the maze task, children had to discover the path leading through a partially blocked maze organised onto three levels by sequentially removing the masks concealing the maze’s passages, thereby revealing which ones were open (see Figure 1).

Preliminary results support our prediction that children’s information search strategies would be more informative in the maze task than in the 20-questions game: across age groups, children required fewer queries in the maze than in the 20-questions game to reach the solution (χ2(1) = 5.70, p = .02), and their queries in the maze had higher information gain (IG, χ2(1) = 42.73, p < .001). However, this effect only emerged for 4-, 5-, and 6-year-olds. For 7- and 8- year-olds, informativeness of questions and maze queries reached the same level (see Figure 2). We also found an age effect on information gain in the question-asking game (χ2(3) = 15.82, p =.001), but not in the maze task (p =.43). Thus, it appears that all children performed better in the maze, yet that this effect diminished with increasing effectiveness of children’s questions.

Possible explanations for the beneficial effect of the maze are the spatial representation of the task’s hypothesis space, making it easier for children to grasp which queries were more informative, as well as the reduced need to generate questions from scratch. We will therefore also describe a follow-up study, which will disentangle these alternative explanations by comparing performance in the maze with a question-selection, instead of a question-generation task.

Authors

• Nora Swaboda, Max-Planck-Institute for Human Development, Berlin
  Presenting Author

• Björn Meder, Max-Planck-Institute for Human Development Berlin
  Non-Presenting Author

• Azzurra Ruggeri, Max-Planck-Institute for Human Development Berlin; Technical University Munich (School of Education)
  Non-Presenting Author