Trial-and-Error Modifications as Cognitive Compensation: Analyzing the Concept Mapping Process in Reconstruction Tasks

Abstract

The importance of trial-and-error modifications during concept mapping has been widely acknowledged, especially as an opportunity for learners to engage in metacognitive and higher-order thinking. However, it has been difficult to systematically analyze such modification activities due to the freedom learners have in creating and labeling nodes and links. This study investigates modification processes using the framework of the reconstruction-based concept map, in which a teacher-constructed map is deconstructed into its component nodes and links first, and then learners reconstruct the map using only these provided components. This setting enables the diagnosis of misconceptions—based on deviations from the teacher-constructed map—and consistent, comparable analysis of the mapping process across learners. From the learner's perspective, prior studies have shown its effectiveness in promoting meaningful learning and higher-order thinking. We analyzed three datasets collected from university students enrolled in an object-oriented programming course. Each dataset corresponds to a different topic, with learners individually reconstructing maps using a Web-based application. The analysis focused on two process indicators: the number of false propositions (propositions not matching the teacher-constructed map) and the number of proposition changes, defined as modifying an existing proposition by reconnecting its link to a node. Regression analyses showed that while proposition changes alone had little correlation with map scores (an indicator of map quality), combining proposition changes with false propositions yielded a strong and significant correlation. Further comparison between the high-scoring group and the low-scoring group revealed no significant differences in the frequency or precision of proposition changes. However, high-scoring learners showed significantly higher rates of successful corrections (changes from a false proposition to a correct proposition) and a better recall rate of false propositions. These findings suggest that a form of cognitive compensation may play a key role in concept mapping performance—namely, the ability to correct earlier errors despite incomplete understanding. For example, high-scoring learners still created false propositions, but were more effective at correcting them. The study emphasizes the importance of supporting proposition modification activities, particularly for low-scoring learners, and provides insights into the design of scaffolding mechanisms in digital concept mapping environments.