The effect of self-correction on implicit and explicit representational change in children’s problem-solving

Abstract

Children’s understanding of problems becomes more abstract and sophisticated as they gain experience with solving them. This thesis uses the framework of the Representational Redescription Model (RRM; Karmiloff-Smith, 1992) to capture how children’s representations progress from implicit to explicit through a process of representational redescription, which progressively re-represents information in the mind to make it more conscious, manipulable, and accessible to verbal and gestural report. In particular, I focus on the RRM’s claim that redescription is triggered through a period of behavioural success without explicit understanding. The lack of opportunities to achieve such implicit, procedural success may explain why young children fail to solve causal problems that they might otherwise understand (e.g., trap-tasks). The RRM suggests that modifying such tasks to facilitate behavioural mastery, such as by incorporating correction of one’s mistakes, might improve performance and understanding amongst young children.

The empirical studies of this thesis investigated the role of self-correction on problem-solving in 2.5- and 3-year-old children. Using a physical trap-task, Study 1 showed that self-corrective practice (compared to non-corrective practice) led to significant improvements in performance when the role of the trap was reversed. Study 2 extended these findings to a digital trap-task, finding that such improvement was not unique to self-correction but could be elicited through extra remedial trials as well. In Study 1, self-correction benefited all ages, while in Study 2, only older children benefited. However, in both studies, children could explain the task appropriately for their age regardless of their performance or their practice condition. I discuss my interpretation of these findings as possible evidence of partial representational redescription, and as evidence of a more complex relationship between behaviour and explanation than predicted by the RRM, with implications for the model’s future use as well as for our understanding of representational change during problem-solving.