Understanding factors affecting cause identification of unexpected events

350603197@@Hisashi@SHIBATA

Abstract

When a system gives outputs that we do not predict, we regard those as unexpected events; and try to identify the causes affecting the unexpected events. Suppose that a complex system consists of multiple subsystems. Cause identification means: first identifying subsystems that perform irregular processing (called odd subsystems), and then find out the irregular processing. In this study, we used a card magic as an experimental material. Magician's card manipulations correspond to subsystems' normal processing and tricks involved in the card magic correspond to the odd subsystem's processing. Participants were required to find out the tricks by viewing the movie repeatedly. We conducted two experiments. In the first experiment, we observed the participants' behavior, and tried to hypothesize the factors that influenced their cause identification. As a result of this experiment, through the comparison of the successful and unsuccessful participants' behavior, the following two tendencies were found: (1) the successful participants carried out their reasoning based on determinate information obtained as the system's output, (2) they were also able to identify the odd subsystems more quickly. In the second experiment, we tested two hypotheses extracted from the first experiment. We designed a two x two between- subject experiment. The first factor concerns the possibility of using the system's output, comprising two levels: Easy and Hard. The second factor concerns the difficulty of the odd system identification, comprising two levels: Easy and Hard. The experimental results showed that the second factor actually affected the performance of cause identification of unexpected events; however the first factor did not. Analysis of the unpredicted result, i.e., no effect of the possibility of using the system's output, implied that the participants may perform the reasoning for cause identification based on hypothetical information not directly observed as system's output.