Three studies were conducted to investigate whether individuals whose performance on a learning task fell short of their previous overconfident self‐assessment would apply more effort on a subsequent task to resolve their dissatisfaction and thereby achieve better subsequent performance than individuals who made accurate or underconfident self‐assessments. Specifically, Study 1 and Study 2 used overestimation, and Study 3 used overplacement to predict subsequent performance by measuring students' self‐assessments before the first task, their level of dissatisfaction with their actual performance on that task, the effort they applied in learning, and their performance on the subsequent task. Furthermore, Study 3 divided the participants randomly into a false feedback group (the control group) and a real feedback group (the experimental group). The results showed that when controlling for prior performance, participants who were more overconfident tended to express greater dissatisfaction and increase more effort to achieve their desired outcomes when they perceived a gap between their desired performance and their actual performance. Notably, they achieved better subsequent performance, whereas those in the control group who were overconfident neither applied more effort in subsequent learning nor increased their subsequent performance when they received “unbiased feedback.” The implications of these findings for education are discussed. Copyright © 2016 John Wiley & Sons, Ltd.