Discrepancy detection and vulnerability to misleading postevent information

JAMES P. TOUSIGNANT 0 1 DAVID HALL 0 1 ELIZABETH F. LOFTUS 0 1 0 This work was supported by grants from the National Science Foun dation and the National Institute of Mental Health. David Hall. who is at Thiel College. spent a year at the University of Washington with funding from the National Science Foundation and his college. Please University of Washington , Seattle, WA 98195 1 University of Washington. Seattle, Washington When people are exposed to misleading details after a witnessed event, they often claim that they saw the misleading details as part of the event. We refer to this as the misinformation effect. In four experiments, involving 570 subjects, we explored the role that discrepancy detection plays in the misinformation effect. Experiment 1 showed that subjects who naturally read a postevent narrative more slowly were more resistant to the effects of misleading information contained in the narrative. In Experiment 2, subjects who naturally read more slowly were more likely to detect a discrepancy between what they were reading and what was stored in their memory. In Experiment 3, subjects who were instructed to read slowly were more likely to detect a discrepancy than were those who were instructed to read quickly. In Experiment 4, subjects who were instructed to read slowly were more resistant to misleading postevent information. Taken together, these results suggest that longer reading times are associated with a greater scrutiny of postevent information. This leads to an increased likelihood that discrepancies will be detected and that the misinformation will be resisted. - In the last few years, a great deal of research has indi cated that misleading postevent information can alter a person's recollection of an event (e.g., Bekerian & Bowers, 1983; Christiaansen, Sweeney, & Ochalek, 1983; Dodd & Bradshaw, 1980; Hertel, 1982; Loftus, 1979; Shee han & Tilden, 1983; Weinberg, Wadsworth, & Baron, 1983). We refer to this phenomenon as the misinforma tion effect. Although research on the misinformation ef fect is clear in showingthat posteventinformation can in fluence a person's reported recollection, many questions remain as to why this occurs (Bowers & Bekerian, 1984; Loftus, Schooler, & Wagenaar, 1985; McCloskey & Zaragoza, 1985). Our current work was motivated, in part, by the ob servationthat exposure to postevent information does not have a uniforminfluence. That is, not all subjects exposed to misleading information display changes in recollection. For example, in one study (Bekerian & Bowers, 1983), subjects were exposed to a series of slides of an auto pedestrian accident, followed by a single piece of mis leading information. If a stop sign was present in the slides, for example, the misleading information implied the existence of a yield sign. In a subsequent randomly ordered test, 40% of the subjects responded in a way that was commensurate with the misleading information. The remaining 60% responded correctly, despite the misin formation. Similar results were obtained in another study usingdifferentmaterials (Weinberg et al., 1983): approx imately 40% of the subjects who received misinformation responded in accord with that information, whereas the remaining 60% responded correctly. What determines whether a subject who is exposed to misinformation will accept or reject that information? Prior studies have shed little light on this question. In some studies on the misinformation effect, subjects have been exposed to multiple pieces of postevent infor mation (e.g., Greene, Flynn, & Loftus, 1982; Sheehan & Tilden, 1983). Vast individual differences in vulnera bility have been observed, with some subjects showing consistent acceptance of the intervening information (sus ceptible subjects) and others showing relatively little or no acceptance of the intervening information (resistant subjects). Thus, in Greene et al. (1982), in which the ex posed subjects received four pieces of misleading infor mation, some subjects accepted all four pieces, some three, some two, some one, and some none. Again, a question arises as to why some subjects are susceptible and others are not. Because very few data were collected from individual subjects, beyond simplemeasures of their final test performance, we knowless than we mightabout the process by which postevent information exerts its in fluence. In Greene et al.'s (1982) study, one useful additional measure was taken. Some subjects were warned in ad vance about the possibility that they would be exposed to misleading information. Subjects who were warnedim mediately prior to reading the postevent information were moreaccurate about whatthey saw then subjects who were not warned at all. Moreover, because all subjects were timed while they read the postevent information, the ex perimenters observed that the warned subjects read the information more slowly. Greene et al. argued that the warned subjects were reading more slowly and using this extra time to more carefully scrutinize the postevent information, which resulted in a greater resistance to its sug gestive effect. Based upon this prior work, we hypothesized that even in the absence of a warning, there would be a relation ship between the time subjects take to read and compre hend the postevent message and their ultimate test per formance. We expected to find that those who read more quickly would be more influenced by the misleading post event information. Such a finding would be consistent with the idea that rapid reading minimizes the chances that sub jects will detect a discrepancy between what they are read ing and what is already in their memory about the event. Several lines of research on the misinformation effect hint at the important role of discrepancy detection is sub jects' resistance to misinformation. Some evidence comes from studies manipulating the intervals of delay between viewing an initial event, encountering a subsequent mis leading message, and engaging in a final test of recollec tion (Loftus, Miller, & Burns, 1978). These studies show that the number of subjects whose recollection is distorted increases with longer, as opposed to shorter, intervals be tween an event and subsequent misinformation. Put another way, recollection change appears to be enhanced by the fading of original memory with the passage of time (see also Hertel, Cosden, & Johnson, 1980; Spiro, 1977). Another line of work concerns the form of the post event information. These studies show that recollection is more often affected by the misleading object of an aux iliary clause than by the same misinformation presented in the focus of the question (Loftus, 1981). For example, subjects were more likely to claim that they saw a nonex istent moustache if the moustache was suggested to them via the question "Did the intruder who was tall and had a moustache say anything to the professor?" than if it was suggested to them via the question "Was the moustache w (...truncated)