Average proportion of correct answers in Experiment 2, separated by condition and trial type (see Fig. 1 for an example of each trial type). Trials differ in whether the faster object in a pair was heavy (heavy = fast), small (small = fast), small and heavy (small/heavy = fast), or big and heavy (big/heavy = fast). A mixed-design condition-by-trial analysis of variance (ANOVA) revealed a main effect of trial type. Specifically, we found worse performance on big/heavy = fast trials (M = .88) than on all other trials (Ms > .97), F(3,153) = 62.54, p < .001, η
2 = 0.55, 1-β = .99. There was also a significant condition-by-trial interaction, F(3,153) = 13.87, p < .001, η
2 = 0.21, 1-β = .99: In the static-images condition, only performance on big/heavy = fast trials (M = .74) was lower than all other trials (Ms = .99). In contrast, in the real-objects condition, performance on both big/heavy = fast trials (M = .88) and small = fast trials (M = .95) was lower than the performance on all other trials (Ms = .99). No other effects were significant, F(3, 153) < 0.86; p > .46.