Competitions, particularly in professional contexts, convey information about participants’ relative ability, yet they also reveal their self-beliefs. They transform personal judgements into common knowledge that others can observe and evaluate. This paper studies how exposing individuals’ competitive information affects how they state their rank beliefs, and whether these effects differ by gender. I propose a framework in which individuals incur psychological costs when inaccuracies in their self-beliefs are revealed, either privately or publicly, and test its predictions in a repeated laboratory experiment (N = 544). Without feedback or observability of belief accuracy, women assess themselves more moderately than men, while introducing public observability reduces overconfidence for both genders. In contrast, private feedback generates gender-specific learning patterns: women adjust more after overestimation, men after underestimation. Together, the findings suggest that belief exposure as a distinct social mechanism in competition helps explain how gender differences in self-assessment emerge and persist in labour markets.