ABS 074: The Impact of Task Structure on Metacognitive Judgment

Tanvi Palsamudram ¹ , Haley Keglovits ¹ , David Badre ¹

¹ Brown University, RI, USA

Van Wickle (2025) Volume 1, ABS 074

Introduction: Metacognition is the knowledge and regulation of one’s own cognition. An open question concerns how people judge their own performance on a task, even before performing it, and what aspects of task structure affect those judgments. Prior research suggests that people perceive cognitive control tasks requiring higher degrees of policy abstraction, or the number of steps in a decision tree between seeing a stimulus and making a response, as being more effortful, and thus avoid them. Given that people use this aspect of task structure to predict effort independent from their actual performance, we asked whether this feature of task structure could also influence metacognitive judgments. We hypothesize that as a task’s policy abstraction increases, metacognitive accuracy decreases, while as experience with a task increases, metacognitive accuracy increases. Thirty participants completed established stimulus-response tasks, which manipulate policy abstraction while controlling for other difficulty factors like load. Participant judgments of their performance and their confidence in these judgments were measured before and after each task. As policy abstraction increased, performance was more error-prone and slow, matching the findings of previous studies. We show evidence that abstraction does impact metacognition, as shown by an interaction between abstraction and confidence in participants’ prediction of their performance, such that for tasks with mid-level abstraction, but not high- or low-level, greater confidence is associated with faster response times. Unexpectedly, task experience worsened participants’ ability to judge their performance in relation to others, even though their objective performance estimates improved independent of policy abstraction or task difficulty. Overall, the results suggest that metacognition is influenced by task structure, such as policy abstraction and task experience, which supports our hypotheses. These features contribute to human estimates of their own task performance, even before performing it.