Exploratory play goes beyond rational action and efficient search

• In Event: Perspectives on play: Motivations and constraints across lifespan and species

Abstract

Play is a universal behavior widely held to be critical for learning and development. Recent work suggests children's exploratory play is consistent with formal accounts of learning (Schulz, 2012). This 'play as rational exploration' view suggests that children's play is sensitive to costs, rewards, and expected information gain. By contrast, here we suggest that a defining feature of human play is that children subvert normal utility functions in play, setting up problems where they incur needless costs to achieve arbitrary rewards.
In Experiment 1, three to seven-year-olds (N=24) were asked to identify which of two storybook characters was playing (Figure 1A-C). Both characters were shown completing the same task but one was efficient and one was inefficient. Across all ages, participants selected the inefficient actor an average of 2.5 out of 3 trials (t(23)=6.78, p < .05), with 58% of participants always choosing the inefficient actor (p < .05). Thus, children distinguished play from otherwise comparable behaviors simply by the violation of rational action.
However, a stronger test of the hypothesis that play is characterized by arbitrary utility functions is whether children themselves engage in unnecessarily costly actions in play. In Experiment 2, we recreated the storybook scenarios to examine children’s decisions when retrieving objects for instrumental reasons (e.g. a pencil for writing with or to give themselves a sticker) or merely as part of play. Using a within-subjects design, we found that most four- and five-year-olds (N=37) performed inefficient actions in play (M=78%, 95% CI: 61%-94%) but efficient actions when instrumentally motivated (M=83%, 95% CI: 67-100%).
In Experiment 3, we asked whether these results would extend to search tasks. No previous work has looked directly at preschoolers’ sensitivity to the size of the search space in exploratory play (but see Ruggeri et al, 2017 for children’s sensitivity to effective question-asking). We offered a choice between two search spaces (e.g. looking in 1 vs 12 drawers). In a within-subjects comparison, preschoolers (N=40 4-5-year-olds) rationally preferred to search in smaller spaces when instrumentally motivated (“Help me find a ball”), but not during play (“Let’s play hide and seek and find the ball”) (McNemar’s p < .001). In a replication study, however, younger three-year-olds also preferred more challenging search problems during play but failed to make efficient search decisions even given instrumental goals (Figure 2).
Collectively, the results suggest that play involves specific and deliberate deviations from efficiency and rational action, despite children otherwise understanding and valuing efficiency in instrumental contexts. We might assume that in play children are pursuing not only the stated goal (e.g. retrieval or search) but because they also are pursuing the goal of having fun. However, this begs the question of why it is “fun” to incur unnecessary costs. We propose that to accommodate the richness of play, theories of action and exploration must explain not only how actions are selected given goals, costs, and rewards, but must also incorporate explanations of how goals are selected and constructed on the fly. To this end, we will discuss ongoing work building computational models of play.

Authors

• Junyi Chu, Massachusetts Institute of Technology (MIT)
  Presenting Author

• Laura E Schulz, Massachusetts Institute of Technology (MIT)
  Non-Presenting Author