Fig. 1: Example data.

WM data from three participants were selected specifically to visualize certain problems with typical aggregation-based measures. Panel A shows the participants’ similar overall average accuracy. An analysis relating overall WM performance to Gf would expect similar levels of Gf in these participants. Yet, when examining the participants’ trajectories of performance on the WM task (B), they evince “same overall performance” in quite different ways. They differ markedly in initial accuracy (C), rate of change (D), and in final accuracy (E). As such, more detailed theories linking WM and Gf would not make identical predictions for the three participants.