Intellectual Humility Questionnaires Fail to Predict Metacognitive Skill: Implications for Theory and Measurement

Intellectual humility (IH) is commonly defined metacognitively: as recognizing the limits of one’s knowledge and abilities. But in the vast and growing space of IH research, the enigmatic construct is commonly measured incommensurately with its theoretical basis (via self-reports questionnaires that reflect one’s perceived, rather than objective, metacognitive abilities). We set out to clarify and measure the theoretical roots of IH, identifying two metacognitive statistical-behavioral properties that seem to be necessary to be intellectually humble (i.e., calibration and resolution) and conducting critical tests of the correspondence between these statistical-behavioral properties and self-reported IH. Using variations of overconfidence paradigms alongside a novel application of generalized additive mixed models capable of detecting individual differences in metacognitive abilities, we find that most IH measures are associated with inflated confidence regardless of accuracy (i.e., a null relation with metacognitive sensitivity), or are specifically related to confidence when participants were correct but not when they were incorrect (Nobservations per model = 7,223-12,706)—leading self-reported IH to be unrelated to epistemic calibration (i.e., metacognitive bias). These findings, which we corroborate in a pre-registered replication, suggest that foundational elements of either the theory or measurement of IH merit substantial revision.