Do people learn from failure or do they mentally “tune-out” upon failure feedback, which in turn undermines learning? Recent research (Eskreis-Winkler & Fishbach, 2019) has suggested the latter, whereas research in educational and work settings indicates that failure can lead to more learning than can success and error-free performance. We conducted two preregistered experiments to replicate the tune-out effect and to test two potential boundary conditions (N = 520). The tune-out effect fully replicated in those experimental conditions that represented close replications of the original study, underscoring the reliability of the original effect. However, the effect disappeared when the same monetary incentives for participation were expressed in terms of a loss (i.e., losing money for each wrong answer) rather than a gain (i.e., earning money for each correct answer; Experiment 1). The effect also disappeared when additional corrective feedback was given (Experiment 2). It seems that switching from gain to loss framing or giving corrective feedback (vs. no corrective feedback) are substantial and meaningful variations of the original paradigm that constitute boundary conditions of the tune-out effect. These results help explain the conflicting findings on learning from failure and suggest that in many applied settings, tuning out upon failure might not be an option