When several causes contributed to an outcome, we often single out one as ‘the’ cause. What explains this causal selection? While most previous work has argued that people select abnormal events as causes, recent work has shown that in some situations, normal events are preferred over abnormal ones. However, all this work has relied on the same methodology to investigate causal selection: vignette studies. Since vignettes can never fully specify what happened, it is possible that the reported causal selection effects result from participants filling in the gaps. We demonstrate, for the first time, causal selection effects for visual displays. Participants view physically realistic animations that feature two candidate causes, ball A and ball B, that either cause ball E to go through a gate, or fail to do so. Our experiment manipulates event normality by including motion blocks in the scene that are more or less likely to let balls through, the causal structure of how the two causes bring about the outcome (and, or, and xor), and whether or not balls A and B go through the motion blocks in the test clip. Even though participants saw that the causal role played by the two balls was identical, their causal selections were still affected by normality and causal structure.
<< Back to list of publications