Aha-related surprise tracks violated expectations at multiple levels in perceptual insight

Insight is often accompanied by an Aha experience, but why some solutions feel especially insightful remains unclear. The insight-as-prediction-error (IPE) account proposes that this experience arises when a solution violates expectations formed before recognition. Here, we tested whether this account generalizes to perceptual insight and whether Aha-related surprise reflects violated expectations at multiple levels. Across two online experiments (Experiment 1: N = 50; Experiment 2: N = 43), participants briefly previewed Mooney images, rated expected solvability, attempted to identify the hidden object, and then rated the Aha dimensions surprise, suddenness, pleasure, and certainty. In both experiments, meta-cognitive prediction error (PE) about solvability reliably predicted greater surprise on correctly solved trials. Longer preview durations increased expected solvability (from 250 to 650 ms), suggesting that these expectations form rapidly from minimal perceptual input but are not simply accurate forecasts of later identification success. Critically, content-based PE-defined as mismatch between anticipated and actual solution content-explained additional variance in surprise beyond meta-cognitive PE and showed clearer selectivity for surprise than the other Aha dimensions. Together, these findings extend the IPE account to perceptual insight and support a multi-level account in which insight-related surprise is associated with trial-wise violations of expectations about both whether a solution will occur and what that solution will be.