Poster #23 - Training Uncertainty Monitoring in Preschoolers

• In Event: Poster Session 09

Abstract

Uncertainty monitoring refers to the ability to experience higher certainty for correct compared to incorrect decisions and is highly relevant for children's self-regulated learning and academic achievement (Roebers, 2017). Previous research suggests that children begin to show evidence of uncertainty monitoring at around three years of age (Lyons & Ghetti, 2011). However, there is still much room for improvement, and little is known about whether early skills can be promoted through training. Answering this question has the potential to inform scientists and educators on factors promoting learning in early childhood. Given recent evidence that vocabulary predicts later uncertainty monitoring (Gonzales et al., 2021), we sought to determine whether an intervention supporting the connection between uncertainty language and uncertainty states would improve young children's uncertainty monitoring.

Participants. So far, we have collected data from 32 children (M=47.39 months, SD=3.34 months), but the target sample size is 72 children.

Procedure and Design. We assessed children's uncertainty monitoring with a perceptual identification task pre- and post-intervention. Participants completed 20 trials in which they identified objects from fragments and reported their confidence (rated on a scale from 0-2). We computed uncertainty monitoring as the confidence difference between correct trials and incorrect trials, with a greater difference score representing better uncertainty monitoring.
We delivered our intervention via storybooks under three conditions. Storybooks were identical across conditions (e.g., same characters and general theme surrounding a treasure hunt in which each page presented different numbers of the objects being sought). The conditions varied in the specific language used to describe the story. In the uncertainty language (UL) condition, children were exposed to contrastive uses of "know" and "do not know" (i.e., they were asked if they knew where the object was in the scene and asked to explicitly label their knowledge states). In the mental state language (MSL) condition, children were exposed to mental terms concerning their linking of the objects ("like" or "dislike"). Finally, in the counting (C) condition, children counted objects and then stated the total number of objects in the scene. This condition did not reference mental states and focused on the children's rudimentary math skills. The conditions are displayed in Figure 1. We predict that children in the UL condition will increase their uncertainty monitoring from pre- to posttest more than children in the ML or C conditions.

Preliminary Results. We preregistered our analyses (https://osf.io/7wqx6). We will conduct a 3 (condition) X 2 (pre- and posttest) mixed design ANOVA with conditions varied between participants and pre- and posttest repeated within participants. Average perceptual accuracy so far is above chance level (.50) at pretest (M=.62; SD=.12) and posttest (M=.62; SD=.12), indicating appropriate task difficulty. Descriptively, uncertainty monitoring increased in the uncertainty language and counting conditions but decreased in the mental state language condition, as shown in Figure 2. However, our sample size is too small for inferential statistics. We will be able to complete data collection and analysis in order to present the results at the meeting.

Authors

• Christopher Gonzales, University of California - Davis

• Florian Jonas Buehler, Universität Bern
  Presenting Author

• Simona Ghetti, University of California - Davis