Fig. 1

Reliable individual differences in the Unstable Towers task. a Left, examples of unstable tower stimuli used in the task. Right, each tower was displayed in a 6 s video at 30 frames per second that circled the tower in 360°. After viewing each video, participants reported which side of the platform the majority of blocks would land on - gray or white. b Performance on the towers task was reliable across independent sets of stimuli. We split the set of towers into two halves (see “Methods”) and computed performance in the two stimulus sets (each plotted point is one participant). Performance across the split halves was significantly correlated (r (98) = 0.50, p < .001) and spanned the range from near chance to near perfect, indicating that our version of the towers task is a reliable and sensitive measure of individual differences. c The difficulty of assessing each tower was reliable across independent sets of observers. We split the participants into two groups and computed performance for each tower stimulus within each group (plotted points are individual towers). Accuracy was highly significantly correlated across split halves (r (98) = .79, p < .001) and spanned a broad range of difficulty