The failure to exploit collective wisdom is evident in the conspicuous difficulty to solve hidden-profile tasks. While previous accounts focus on group-dynamics and motivational biases, the present research applies a metacognitive perspective to an ordinary learning approach. Assuming that evaluative learning is sensitive to the frequency with which targets are paired with positive versus negative attributes, selective repetition of targets’ assets and deficits will inevitably bias the resulting evaluations. As selective repetition effects are ubiquitous, metacognitive monitoring and control functions are required to correct for repetition biases. However, three experiments show that metacognitive myopia prevents judges from correction, even when explicitly warned to ignore selective repetition (Experiment 1), when same-speaker repetitions rule out social validation (Experiment 2) and when blatant debriefing enforces superficial corrections (Experiment 3). For a comprehensive understanding of collective judgments and decisions, it is essential to take metacognitive monitoring and control into account.