Poster #39 - The role of task demands and individual differences in preschooler’s probabilistic reasoning

• In Event: 3-053 - Poster Session 10
  In Poster Session: PS 10 Section - Cognitive Processes

Integrative Statement

Contemporary research has provided an inconsistent picture of preschool children’s probabilistic reasoning. Though preschoolers have successfully used probabilistic information to make decisions on some tasks (Denison et al., 2006; Kushnir et al., 2010), 3- and 4-year-olds have been less successful on tasks based on an infant choice paradigm (Girotto et al., 2016). It was suggested that children’s executive functions (EF) might contribute to their poorer performance, because the preschool paradigm appears to place higher demands on children’s EF compared to the infant version. Thus, preschoolers may need to call upon their EF to succeed on probabilistic reasoning problems. However, another potential explanation for preschooler’s poor performance involves the possibility that the preschool choice paradigm, adapted from the infant paradigm, may underestimate the abilities of preschool children. Although modeling tasks after the infant paradigm facilitates comparison, some aspects integral to the infant design may be inappropriate for preschoolers. Preschoolers and infants may find different stimuli rewarding, and, because the infant paradigm seeks to minimize verbal demands, preschoolers may benefit from answering a more direct question.
We attempted to tease apart these possibilities in two experiments. We examined 3- and 4-year-olds’ probabilistic reasoning abilities using a novel choice task designed for preschoolers on a tablet. Children were presented with two gumball machines filled with blue and black gumballs and won a sticker if they selected a machine that gave them a blue gumball. On each problem, the experimenter asked the child, “which machine do you want for a chance at a blue gumball?” Following the probability task, children completed a battery of EF measures (i.e., cognitive flexibility, working memory, inhibitory control) administered on the tablet.
In Experiment 1, children (N=50) completed eight problems (see Figure 1). To create a global estimate of children’s probabilistic reasoning abilities, we varied the difficulty of the problems. Some problems could be easily solved by using a shortcut that approximated true probabilistic inference (i.e., simply selecting the population that contained more blue objects, without considering the proportion of blue to black), while more difficult problems pitted a salient incorrect response against a less salient correct response (i.e., based on a true comparison of proportions). Children performed above chance on all problems (see Table 1 for means and significance tests), suggesting that they were able to engage in true proportional reasoning. To further explore this, we presented a new set of children (N=76) with nine problems in Experiment 2. These problems were more difficult than those in Experiment 1, because they contained fewer shortcuts that could lead to the correct answer. Children performed above chance on all problems (see Table 1).
We observed an inconsistent relationship between EF and children’s probability performance. Though the younger children in our sample benefitted from better-developed EF on the probability problems in Experiment 1, we observed no relationship between EF and probability performance in Experiment 2. These findings suggest that adapting the method to better suit preschoolers improved their performance and that the relationship between probabilistic reasoning and EF requires further inquiry.

Authors

• Samantha Gualtieri, University of Waterloo
  Presenting Author

• Tara McAuley, University of Waterloo
  Non-Presenting Author

• Stephanie Denison, University of Waterloo
  Non-Presenting Author