Abstract
Background: Open behavioral datasets can be used to ask new questions that were not the primary inferential targets of the source studies. We used two complementary datasets to examine two distinctions relevant to intuitive judgment: whether a self-reported hunch carries more task-relevant information than a self-reported guess, and whether subjective confidence should be treated as equivalent to objective performance or metacognitive efficiency. Methods: This theory-guided, non-preregistered secondary study returned to participant- and trial-level open data rather than relying on published summary statistics. Dataset A was Experiment 2 of Monaghan et al. (22 participants; 6,336 training trials). We recomputed within-participant accuracy for Intuition versus Guess and formally tested adjacent source-attribution contrasts using paired analyses and participant-clustered generalized estimating equations (GEE). Dataset B was the multi_intero dataset of Banellis et al. We directly compared cross-domain correlations in mean confidence with corresponding correlations in metacognitive efficiency (M-ratio) in identical participant sets and tested whether self-reported interoceptive awareness (MAIA) related differentially to cardiac confidence, accuracy, and M-ratio. The datasets were analyzed separately and were not pooled. Results: In Dataset A, Intuition-attributed decisions were more accurate than Guess-attributed decisions (mean difference = 0.146, 95% CI 0.086–0.205; t(21) = 5.12, p < 0.001; dz = 1.09); the adjusted trial-level association was OR = 1.56 (95% CI 1.25–1.96). Exploratory adjacent contrasts were also positive for Recollection versus Intuition and Rule knowledge versus Recollection. In Dataset B, cross-domain confidence correlations exceeded corresponding M-ratio correlations (delta r = 0.448–0.616; all Holm-adjusted p < 0.001). MAIA was associated with cardiac confidence (r = 0.262) but not cardiac accuracy (r = 0.019) or cleaned cardiac M-ratio (r = −0.001), and the correlation differences were significant. Conclusions: These analyses do not validate a unified construct or mechanism of intuition. They provide complementary constraints on its measurement: a reported hunch was not equivalent to a guess in Dataset A, while confidence was not equivalent to objective performance or metacognitive efficiency in Dataset B. Because intuition, confidence, performance, and metacognitive efficiency were not measured jointly in the same task, stronger claims about their within-person architecture require prospective within-task testing.