General Intelligence (g), ACT Scores, and Theory of Mind: (ACT)g Predicts Limited Variance Among Theory of Mind Tests


This study is the first to examine relations between general intelligence (g), non-g factors, and theory of mind (ToM) using structural equation modeling with multiple indicators of g and ToM. g was based on the subtests of the ACT, a college admissions test that is strongly g loaded, and ToM was based on the Reading the Mind in the Eyes Test and the Short Story Test (SST). g correlated strongly with a latent ToM factor ($β$ = .65) and moderately with the two ToM tests ($β$ ≈ .34), which correlated modestly with each other ($β$ = .27). The modest correlation between the ToM tests indicates that g predicted a small amount of variance among the ToM tests (7%) and suggests that the ToM tests had little in common. In addition, non-g residuals of the ACT subtests, obtained after removing g, correlated negligibly with the ToM factor and the ToM tests (|$β$| textless 0.06). Similar results were obtained for the ToM residuals, which correlated trivially with the ACT subtests. The trivial non-g effects suggest that g-ToM relations were attributable to “not much more than g.” The results replicated with different combinations of ACT subtests, controls for possible confounds (reading comprehension on the SST), and another college admissions test (the SAT). The use of a convenience sample (college students) and the limited measures of g and ToM are discussed as limitations. Future research should examine the robustness of effects using different measures of g and ToM and also examine possible mediators of g-ToM relations (e.g., executive functions).