Abstract
The visual system extracts information about causal structure in the environment. However, people often attach a causal structure to a coincidence, even though there is no causal structure underlying a coincidence. Given this attribution of causal structure to a non-causal event, do we recognize and remember coincidental events similar to causal events? Here, we compare the visual representation of causal versus coincidental events. 150 participants watched eight brief videos (~4 seconds) depicting different real-life scenes. In each video, two objects were directly involved in either a causal event (i.e., a tennis ball knocking over a soda can) or a coincidental event (i.e., office door closing after a chair starts spinning). In addition to these integral objects, two other objects were visible but incidental to the event in each scene. Each participant viewed 4 causal events and 4 coincidental events and the event type in each scene was counterbalanced across participants. After watching these videos, participants took a surprise visual recognition test, in which they viewed objects one at a time and indicated whether they had seen the object in any of the videos or not. Participants saw 64 total objects: 2 objects x 2 object types (integral and incidental) x 2 event types (causal and coincidental) x 4 videos + 32 category-matched foils. We found that visual memory for coincidental events was just as accurate as causal events; in fact, there was a significant interaction between event type and object type, such that participants remembered objects integral to coincidental events better than objects integral to causal events, but had worse memory for incidental objects in coincidental events compared to causal events. These results suggest that the mere attribution of causal structure to a non-causal event is sufficient for coincidences to be represented and remembered like causal events.