Medidas de eficiência metacognitiva na aprendizagem de domínios em física

Abstract

Metacognition is the ability that the individual has to recognize his own learning process and the capacity to improve it, whether in aspects of simple stimulus discrimination, tasks that involve memory, decision making, or problem solving. Studies have evaluated and concluded that the regulatory brain systems - which are the neurophysiological bases of human cognitive abilities - and behavioral systems have a direct relationship with metacognitive processes. Such processes involve the knowledge and regulation of cognition, which implies being aware of the factors that influence learning, knowing strategies to foster it, choosing for each specific situation the appropriate strategies, monitoring and controlling the process, and evaluating learning by discriminating and quantifying it, through measures of metacognitive bias (the overall level of confidence expressed regardless of whether the test is correct or incorrect), metacognitive sensitivity (the accuracy in distinguishing the ability to discriminate correct from incorrect judgments), and metacognitive efficiency (a subject's level of metacognition given a given basic task performance). Signal Detection Theory (SDT) is an important methodology for discriminating between sensitivity and response deviation in probability judgment activities as well as assessing the confidence between the judgment and its correspondence with judgment correctness. Therefore, the present study - of exploratory nature - aims to investigate the possibility of tracing metacognitive profiles from metacognitive parameters based on Signal Detection Theory. To this end, we sought to measure the metacognitive sensitivity and metacognitive efficiency of students in learning physics domains, calculating the through d'or D-prime measure and the joint probability distributions in performance judgments in which there is accuracy of hits conditioned to high levels of confidence in the judgment (meta-d-prime or meta-d'), in order to evaluate the relevance of TDS models for measuring metacognition. For this study, we used data from 16 first year civil engineering students of the Institute of Technology, Federal University of Pará, which were collected during the course Physics I, at the Richard Feynman Laboratory for Teaching Innovation in Physics. Over the course of 1 month, the students answered 8 tests, each with 10 questions (40 items) assessed as RIGHT and WRONG the physics domains related to the subject taught. All test judgments were made according to the Typology and Error Mapping (TME) methodology, which measures the confidence level (certainty/uncertainty) with which the student judges certain learning domains. At first, they were instructed on how to use the TME methodology in the tests, and then the students took the physics tests that dealt with mechanics topics. Based on the analysis of the Signal Detection Theory, the results showed that there is a great deficit in terms of sensitivity and metacognitive efficiency, since there was among the metacognitive classes a predominance of the metacognitive profile III, which shows little ability to make judgments about the learning domains themselves. This shows a large discrepancy between what the individual actually knows and what he/she thinks he/she knows, which implies a low metacognitive sensitivity and, consequently, a lowering of the metacognitive efficiency of this individual. Finally, the results corroborate the importance of creating teaching methodologies that aim to develop metacognition in the learning process.