Mitchell, K.J., & Zargoza, M.S. (2001). Contextual overlap and eyewitness suggestibility. Memory & Cognition, 29 (4), 616-626.

This study examines whether featural overlap of narrative content, narrative structure, and environmental context contribute to people’s tendency to confuse suggested details for the details they actually witness. In typical eyewitness suggestibility studies, the witnessed event and the postevent questioning occur close in time, in the same room, with the same experimenter, and with the same co-participants, all of which tend to maximize the narrative similarity between the original event and the postevent interview. To better understand the effects of these various types of featural overlaps, which lead to misatrribution errors (i.e. source monitoring perspective), the authors manipulated these factors since they will vary in real world situations.

Experiment 1A

• Phase 1 - Participants viewed a 5-minute video of a house burglary.
• Phase 2 - Afterwards, they answered a postevent questionnaire containing misleading suggestions, some of which were repeated. There were 36 items containing 3 subsets of 12 items, which basically reviewed the original events 3 times in succession.
• Across groups, the suggestions were the same. The manipulation was the amount of narrative overlap (ranging from high to low) between the video and the postevent questions.
• High narrative overlap – Included a substantial amount of narrative overlap between the content of the video and the questions.
• Moderate narrative overlap – Removed the temporal cues that served as links between scenes to reduce the cohesiveness of the story; also reduced the number of narrative details; reduced, but did not eliminate the narrative overlap
• Moderate overlap plus irrelevant information – To control for the number of words in the postevent questions so that any reduction in errors in the moderate overlap group was not simply the result of participants’ being exposed to fewer verbal materials
• Low narrative overlap – To further reduce overlap, the moderate overlap questions were randomly ordered within each subset to drastically reduce the cohesiveness of the story line.
• Phase 3 – A surprise source recognition test was given. Participants were instructed that some of the statements they would hear during the test phase contained information only from the video, or not in the video and only in the questions they answered, or from both the video and questions, or finally from neither the video nor the questions.

• Reducing narrative overlap (i.e. true details, cohesiveness of the story, chronology of events, adding extraneous information) had no significant effect on participants’ suggestibility even with repeated exposure.
• Results were consistent with past research demonstrating that repetition increases misremembering. However, suggestibility did not differ as a function of the narrative overlap manipulation.

Experiment 1B

• Same as experiment 1A except for:
• 24-hour delay between the video and the postevent questions.
• Different room, tested with different co-participants, different experimenter
• Only had "high overlap" and "low overlap" groups
• The reason for doing experiment 1B was to boost the effects of reducing narrative overlap.

• Replicated the findings for 1A
• Even with these substantial changes, the overall incidence of misattribution errors increased, but again did not affect suggestibility as a function of narrative overlap.

In summary, experiments 1A and 1B demonstrate that false memory for suggested events was unaffected by substantial changes in content, structure, and environmental context for the postevent interview.

Experiment 2

• Assessed whether narrative overlap would influence participants’ tendency to recall the suggested items when asked about their memory for the video 2 days after the postevent interview (used a delayed cued recall test)
• The rationale behind this experiment was that the authors thought that experiments 1A and 1B caused participants to rely too much on familiarity and that 1A and 1B were insensitive to group differences in confusion regarding the source of the suggested items.

• Same procedures for experiment 1A for the high narrative overlap and low narrative overlap groups except for:
1. Day 1 – After the video, participants did an unrelated filler task for 5 minutes and then completed 12 postevent questions.
2. Participants returned 2 days later to complete the source memory test (a printed cued recall test that asked them about their memory of the video and questionnaire)
3. Again, participants were explicitly instructed about some of the items only being in the video, some from only the questions, etc.

Results

• Used 2 measures:
1. Assent to the initial question about the general category of information (i.e. "In the video, do you remember seeing the thief carrying a weapon?") Participants responded "yes," "no," or "not sure."
2. The proportion of initial assents that were followed by recall of a suggestion (i.e. writing in "gun" for the weapon)
• Contrary to the hypothesis (narrative overlap increases source misattribution), the high overlap participants were less likely to recall seeing critical items in the video compared to the low overlap group.
• As expected, both groups were more likely to recall seeing the critical items in the video if they had been exposed to them in the questions than if they had not.
• Memory that the suggestions were only read in the questions was significantly more accurate for the high overlap group than the low overlap group.
• So, there was clear evidence that the narrative overlap manipulation did affect participants’ memory for the source of suggested items, but in the opposite direction.

Discussion

The authors discussed several possibilities for their findings which are as follows:

1. Once the overlap between 2 sources reaches a certain threshold, further increases in the amount of overlap do not increase confusion appreciation.
2. The high level of relatedness as a result of sharing a common referent produces high levels of source confusion
3. Participants reflect back to the original event and fill-in gaps using their own ideas (especially for the low narrative overlap groups)
4. Narrative coherence of the high overlap questionnaire served to enhance memory.

In summary, the narrative overlap manipulations produced no effects on the recognition test but did on the delayed cued recall test. However, high overlap produced better source discrimination than low overlap.