Honors College

Document Type

Honors Thesis

Publication Date

Spring 2014


There have been 314 post-conviction DNA exonerations, and about 70% of these wrongful convictions were due to witness misidentification. Many factors affect the accuracy of a witness’s testimony, including the concept of “verbal overshadowing,” in which the verbal reporting of a visual memory interferes with the subsequent recognition of the visual stimuli. The present study seeks to replicate Jonathan Schooler’s original findings with regard to this phenomenon (Schooler & Engstler-Schooler, 1990). The first experiment is focused on verbalizing a visual memory that is particularly hard to put into words: the memory of a face. The hypothesis for both the original study as well as the replication is that the verbal recoding will overshadow the visual memory.

Data from 122 participants was used in a replication of Schooler’s experiment on verbal overshadowing. First, the participants watched a video of a robbery. Participants in the Description condition then described the robber, and participants in the Control condition listed countries and capitals. After completing a filler task, all participants were asked to pick the robber out of a lineup consisting of the perpetrator and seven distractors and rate their confidence on their decision. The results showed a similar trend as the original study: the participants in the Description condition identified the robber incorrectly more often than the Control condition.

For the second analysis, 104 participants completed both the replication study and a social phobia scale (Mattick & Clarke, 1998). Within a normal population, people with higher anxiety are expected to perform less well on facial recognition tasks, and those with lower anxiety are expected to perform better. Results from the second portion of the study showed that participants considered to be lower in social anxiety in the control condition identified the robber correctly much more often than did those in the description condition, demonstrating a verbal overshadowing effect. Participants higher in social anxiety were all around worse at identifying the robber in both the description and control conditions. These results show that social anxiety can be a moderator of verbal overshadowing. Within the control condition alone, participants with higher social anxiety were significantly less accurate at identifying the robber than participants with lower anxiety.