Within-individual and across-individual variation in a large corpus of conflict tasks.

Are patterns of within-individual variation the same as across-individual variation in cognitive control conflict tasks? We study whether conflict effects are additive or multiplicative on response-time distributions, both within and across individuals, using a hierarchical shift-scale model and quantile regression on a corpus of 64 conflict tasks spanning over 10 million trials. Results show that mean RT and conflict effect size are only weakly related across individuals because they are driven by largely different parameters. For Stroop effects, both within- and across-individual variation is primarily in scale, indicating a multiplicative locus; flanker and Simon effects also have an additive, within-individual component indicating that conflict in these tasks may involve motor components in addition to more central cognitive components. Hence, Stroop, flanker, and Simon tasks are not interchangeable measures of control. In contrast to within-individual variation, across-individual variation has a more consistent signature. Within-individual variation may be in both shift and scale, but across-individual variation is largely in scale and not as much in shift. Hence, perhaps what differentiates individuals is more central and less related to motor control.