Confounding the effects of delay and interference on memory distortion: commentary on Schmolck, Buffalo, and Squire.

In a recent article, Schmolck, Buffalo, and Squire (2000) investigated memory distortions in college students’ recollections of how they heard the verdict in the O.J. Simpson murder trial. Students were questioned either 15 months or 32 months after the verdict was rendered. Their responses were compared with their own reports 3 days after the verdict. Responses 15 months after the verdict were more likely than those given after 32 months to be in concordance with participants’ initial reports. Whereas only 11% of 15-month reports contained major distortions, 42.9% of 32-month responses were in error. In addition, Schmolck et al. reported that 77.8% of the major distortions involved incorrectly reporting the source of the news (e.g., television or radio), and that these participants, “described an entirely different situation than the one they had described earlier” (p. 43). Despite their apparently poor memories, 61% of the participants whose responses were characterized as “grossly inaccurate” (p. 43) were highly confident of their accuracy. The breakdown of confidence ratings for these inaccurate participants by delay condition was not reported. The authors concluded that the discrepancy between the 15and 32-month conditions was due to the difference in the time that had elapsed after the verdict. In drawing this conclusion, they failed to consider a possible confound. Specifically, 16 months after the murder-trial verdict, the month after the 15-month data were collected, the verdict of Simpson’s civil trial was announced. This event, although perhaps not as surprising or emotionally charged as the first verdict, was certainly a memorable event in its own right. Therefore, participants in the 32-month condition were confronted not only with a longer delay, but also with the possibility of interference from their memories of the second verdict. Previous studies that compared immediate and delayed memories of consequential public events included memories of the Challenger Space Shuttle disaster (Bohannon & Symons, 1992; McCloskey, Wible, & Cohen, 1988; Neisser & Harsch, 1992), the resignation of Prime Minister Margaret Thatcher (Conway et al., 1994), and the Loma Prieta earthquake (Neisser et al., 1996). One distinction that has emerged from these studies is that memories approximately 1 year after the event tend to be accurate, whereas those reported after a 3-year delay are more likely to be distorted. However, as these studies differed in various ways, it is still an open question as to whether memory distortion is a function of delay. One of the goals of Schmolck et al. was to determine the effect of delay duration on memories of a single event. However, one cannot be certain that the 32-month participants in this study, unlike participants in earlier studies, were attempting to recall the target event and not the more recent similar event. The remarkably high proportion of errors that involved the report of an “entirely different” (p. 43) event may be a result of the interference of memories of the civil-trial verdict. This confound makes it difficult to tease apart the effects of time and interference. Moreover, the fact that participants in the 15-month condition were more likely to respond “don’t remember” than were those in the 32-month condition (21.4% and 5.7%, respectively) is consistent with the 32-month participants having two memories of verdicts and the 15-month participants having only one. One possible strategy for assessing the relative contributions of memories of the civil-trial verdict to the reports of the 32-month participants would be to classify the responses based on distinctions between the two verdicts. Whereas the criminal-trial verdict was delivered in the morning, when participants would likely be at school, the civil verdict was reported in the evening (interestingly, during President Clinton’s State of the Union Address), when participants would be more likely to be away from school. The fact that the two verdicts were at odds with one another (the criminal verdict was in Simpson’s favor, the civil verdict was not), and that public opinion was split along racial lines (Newport & Saad, 1997), may also allow for the classification of recollections. If the “major distortions” at 32 months are found to have details that are consistent with the events surrounding the civil-trial verdict (e.g., a participant said that the verdict was reported in the evening, or that a mostly Caucasian group of observers cheered), it is likely that the participants were victims of interference. Although Schmolck et al. presented compelling data, care should be taken in interpreting their results. The effects of the confounding factor of a second, similar event on participants in the 32-month condition must be considered. The question of whether flashbulb memories decay over time is still open.