Poster #16 - Learning to Think or Thinking to Learn: The Role of Metacognitive Monitoring in Arithmetic.

• In Event: Poster Session 18

Abstract

There is a longstanding tradition in developmental research devoted to uncovering the cognitive correlates of arithmetic performance and development, with the ultimate goal to develop ways to intervene in and stimulate this development (De Smedt, 2022). One cognitive factor that has received less attention in research on individual differences in arithmetic, yet, has been intensively studied in the field of education, is metacognition. Already in 1979, in his pivotal paper, Flavell (1979) theoretically posed that metacognition has an important effect on performance. Recently, we empirically demonstrated that metacognitive monitoring, a specific aspect of metacognition (i.e. the subjective self-assessment of how well a (cognitive) task will be/is/has been performed; Nelson & Narens, 1990) is a powerful unique predictor of concurrent and later arithmetic performance in primary school children (Bellon et al., 2019, 2020, 2021). Although these correlational data help us understand the association between metacognitive monitoring and arithmetic performance, they do not answer the question of the causality of this association. Such causal evidence is however critical to ground educational interventions to improve arithmetic performance. To obtain this causal evidence, carefully designed learning experiments are needed to understand if and how metacognitive monitoring affects arithmetic. This was the aim of the current study.
We investigated this outstanding issue in primary school children in Grade 2 (i.e. 7-8 year-olds) and Grade 3 (i.e. 8-9 year-olds). Our sample included about 200 children (as determined by a priori power analyses allowing for 10% data loss). We used a randomized controlled learning experiment. In this learning experiment, we carefully manipulated metacognitive monitoring during a learning process. We sought to determine whether metacognitive monitoring had a causal impact on subsequent arithmetic learning and whether the effect generalized to other academic domains (spelling). We used a randomized pretest-posttest design with three conditions. The 3 conditions all involved 4 training sessions of approximately 30 minutes Pilot data earlier demonstrated that this amount of training was effective to improve monitoring. In each training session, children were asked to solve single- and double-digit arithmetic items. After each item, children had to answer one additional question on which the conditions differed (i.e., a control question, a metacognitive question without feedback, and a metacognitive question with feedback). Both in pre- and in post-test, children’s arithmetic performance, spelling performance and monitoring performance in both domains were assessed. Additionally, control measures (i.e. intellectual ability and processing speed) were administered at pretest. Given the predicted causal influence of monitoring on performance, we expected children in the metacognitive conditions to have a steeper learning curve during training, and better post-test performance compared to children in the control condition. We expected these effects to occur in arithmetic for all children. We further predicted, in view of the development of monitoring from a specific to a more general process (Bellon et al., 2020) that in the 8-9-year-olds transfer effects to spelling would occur, while they would be absent in the 7-8-year-olds. To analyze this, we will use preregistered ANOVA models (one variable, between groups with two levels).

Authors

• Elien Bellon, Katholieke Universiteit Leuven (KU Leuven)
  Presenting Author

• Wim Fias

• Bert De Smedt, University of Leuven