Metacognitive processes in human-computer interaction: Self-assessments ofknowledge as predictors of computer expertise

Metacognition, a person's knowledge of her or his own cognitive processes, is a concept that can be applied to many areas of human-computer interactio n. This article reviews the state of contemporary knowledge regarding metac ognition and describes implications for the domain of human-computer intera ction. A conceptual framework is presented that distinguishes monitoring an d regulation processes of metacognition. One aspect of metacognition, self- evaluation of knowledge, was investigated for a word processing application as an illustration. An experiment was conducted to evaluate which of four methods of self-assessment of expertise was the best predictor of declarati ve knowledge (accuracy and completeness of descriptions on how to perform a task). In addition, the experiment examined whether classifying users base d on self-reported estimates of expertise would produce differences in thei r declarative descriptions. Results showed that individuals' ratings of the ir overall knowledge were better predictors than were estimations of freque ncy of use, as would be expected from the literature on metacognitive monit oring. In addition, classifying; users based on their self-ratings of exper tise showed differences in accuracy of declarative knowledge and strategy c hosen to perform a task. Experts were more accurate in their descriptions o f how to complete a task compared to novices and used more complex strategi es to complete hard tasks.