Training Primary School Children's Memory Monitoring With Performance and Monitoring Feedback

• In Event: Metacognition

Abstract

Children's ability to accurately monitor their memory is crucial for self-regulated learning and academic achievement (Schneider & Löffler, 2016; Roebers, 2017). Memory monitoring is the ability to introspect and evaluate one's memory and to experience higher confidence for correct than incorrect memories (Nelson & Narens, 1990). However, primary school children discriminate only marginally between correct and incorrect memories and are generally overconfident when evaluating their memory performance. This negatively affects their learning and task performance (Destan et al., 2014; Finn & Metcalfe, 2014). Consequently, training memory monitoring might benefit primary school children's learning. Previous studies tried to improve children's monitoring with feedback, but the benefits of feedback for children's monitoring are inconsistent across studies (e.g., Geurten & Meulemans, 2017; Lipko et al., 2009, 2012; van Loon & Roebers, 2017, 2020). However, none of these studies systematically trained monitoring over an extended period of time, which is considered crucial for sustained improvements.
We evaluated a six-times, computer-based intervention program for first graders (M = 7.45 years old) to improve metacognitive monitoring through feedback. We assigned the subjects (N = 127) to either a monitoring feedback group (MF), a performance feedback group (PF), or an active control group (AC). The intervention sessions for children in the MF and PF group consisted of a paired-associates task. Children learned twelve pairs of pictures, did a recognition test with four alternatives, indicated their confidence (on a scale 4 point Lickert scale), and immediately received feedback. The PF group received feedback solely on their recognition. The MF group received feedback on their recognition and the correspondence between recognition and their confidence. The AC solved an inhibition task different from the paired-associates task. We assessed children's memory monitoring pre- and post-intervention with a paired-associates task (different from the intervention task) without feedback. We expected that monitoring accuracy would increase more in the MF and PF groups compared to the AC group.
We preregistered our analyses (https://osf.io/mwnsy). We run a 3 (MF, PF, AC) X 2 (pre- and post-test) mixed design ANCOVA with feedback type varied between participants and pre-test and post-test repeated within participants. The dependent variable was monitoring accuracy operationalized as a difference score between children's average confidence for correct and incorrect trials, with a greater difference score indicating better monitoring. We included age, SES, grammar, and non-verbal IQ as covariates to account for group differences. We found a significant time x condition interaction F(2, 121) = 2.97, p = .05, "η" ^2= .01. Simple effect analyses revealed that participants monitoring increased from pre- to posttest in the MF group F(1, 50) = 4.35, p = .04, "η" ^2= .03, but not in the PF F(1, 31) = 0.35, p = .56, and AC groups F(1, 40) = 1.87, p = .18. Our results confirm our hypothesis that MF is beneficial for children's monitoring. However, we did not find an effect of PF on memory monitoring (see Figure 1).

Authors

• Florian Jonas Buehler, Universität Bern
  Presenting Author

• Simona Ghetti, University of California - Davis

• Claudia M. Roebers, University of Bern