Many real-world decisions involve uncertainty, yet some are governed by logical constraints that make certain outcomes objectively more or less probable. Despite this, people often commit reasoning errors such as the conjunction fallacy, i.e. judging a conjunction of events as more likely than a single constituent event. This fallacy has proven robust across populations and contexts, and difficult to fully mitigate. While previous research has shown that reformulating problems or providing explicit instruction can reduce such errors, these approaches often yield limited success and may not generalize to everyday decision-making, where task structures are ambiguous and feedback is sparse. We investigated whether individuals could reduce conjunction fallacies through repeated exposure to decision tasks paired with minimal outcome feedback. Participants (N = 56) were randomly assigned to a training group, which completed a series of conjunction judgment tasks with feedback, or a control group. Results showed that the training group improved on both trained and untrained conjunction tasks, including those based on real-world and clinical scenarios, while the control group showed no such improvement. No transfer effects were observed for unrelated base-rate tasks. Performance gains were gradual, suggesting that participants developed judgment strategies over time rather than immediately adopting normative rules. These findings demonstrate that conjunction fallacies can be mitigated through self-guided learning with minimal instruction, offering a promising approach to improving probabilistic reasoning.