Past research has proven that verbal descriptions of an event or person can negatively affect the accuracy of later recognition. This effect is known as verbal overshadowing. The effects of verbal overshadowing are crucial for eyewitness accounts since eyewitnesses are usually required to give a detailed account of the event and a description of a perpetrator.

3 Processes that contribute to Verbal Overshadowing according to Schooler et. al (1997)

1. Modality mismatch- When giving a verbal description of a perpetrator, featural descriptions are listed (long hair, blue eyes), however when you visualize a target, configural representations are used (how the features fit in relation to one another). Thus, verbal and visual representations compete against one another on recognition.
2. Availability – The visual representation of a target should remain accessible/available even though a verbal description is given, however when retrieving the visual representation, it may not be immediately accessible for a recognition test.
3. Recoding Interference – When giving a verbal description of the target, the verbal description might not be accurate. Thus upon retrieval/recognition, the incorrect, self-generated information is more likely to be retrieved then the actual visual representation that they were first presented with.

Past studies have found inconsistent results concerning the effects of verbal overshadowing. Some studies have shown that describing the target before a recognition test might actually improve accurate identification (Chance & Goldstein, 1976; Mauldin & Laughery, 1981; Mc Kelvic, 1976; Read, 1979; Wogalter, 1991, 1996). Schooler et. al believed the cause of these differing results were due to 1). the encoding & retrieval processes participants used and 2). The degree of interference generated by the description task. Meissner et. al however believe that the reason contradicting results occur is due to the various levels of criteria used by participants upon retrieval.

Experiment 1

Experiment one tests 1). how different criteria levels will affect accuracy and 2). whether or not a delay will result in a "release" of the verbal overshadowing effect, and 3). whether description accuracy predicts identification accuracy.

Participants

240 Introductory Psychology students participated for credit. 28 experimental psychology lab students provided data lineup fairness analyses.

Design & Procedure. Participants were told that they would briefly be shown a slide and that they would later be asked what they had seen. The target face was presented for 10 seconds. After viewing the target face, participants were involved in a 5-min. filler task and were then assigned to 1 of 4 conditions.

Participants in the description conditions were given response sheets that had lines numbered from 1 to 25. Participants in each separate description condition were administered the following directions.

***Participants in the standard recall condition were told to describe the face they saw in the slide with as much detail as possible. They were told to describe the target so that based on their description someone else could identify the target. They were also told to concentrate on the task.

***Participants in the warning recall condition were given the previous directions. In addition to the standard recall directions, participants were told that it was important to report only what the participants were sure of. They were informed they did not have to fill out all of the lines, and not to guess at any features.

***Participants in the forced recall condition were given the standard recall directions and also were told that it was important to report everything that they could remember, even if it felt like they were guessing. They were told to fill all the 25 lines with descriptions of the target, even if they weren’t positive about certain features.

***Participants in the control condition did not give a description of the target face.

After the description task, participants in the delay group answered an attitude questionnaire for 30-minutes and were unaware of the identification test. Participants who were not in the delay condition were immediately presented with a photo lineup id. test.

On the identification task, participants were told they would be presented with a slide of 6 faces and to choose the face they had previously studied. Participants were informed that the target may or may not be present.

Results and Discussion

Identification accuracy. In accordance to Meissner et. al’s hypothesis, the different levels of criterion affected later recognition. Participants in the forced recall condition performed significantly worse on the identification accuracy than participants in any other conditions. Participants in the warning condition did significantly better than participants in the forced recall condition and control condition. The delay had no effect on identification accuracy. See Table 1.

Description Quality. Meissner et. al analyzed the description data to determine if description accuracy and description criteria influenced the id. of the target. First, they used 2 independent coders and these coders divided the description info. from each criterion condition into one of 3 categories:

1). Correct details – facial features that correctly matched the target

2). Incorrect details – features that didn’t match the target

3). Subjective details – ambiguous qualities of the face

Coders trained to criterion of 85% agreement.

Results: The response criteria significantly effected participant’s description quality – participants in the forced condition and standard condition generated significantly more inaccurate details than the warning condition. (See Table 2). There was a significant correlation between identification accuracy and description accuracy. See Table 3.

Experiment 2

Experiment 2 examines 1). whether participants in the forced recall condition using self-generated descriptions would differ on id. accuracy from participants who are given misinformation from the experimenter and 2). whether a warning (source monitoring) can improve the accuracy of identification.

Experiment 2 uses a forced recall condition, experimenter provided description condition(inaccurate descriptions provided by the experimenter that the participants read), and a control condition. A warning was applied to some participants in the forced recall condition and in the experimenter provided information condition.

Method

Participants. 150 intro. Psych. Students who received class credit for their participation.

Design & Procedure. Participants in the forced recall condition were given the same instructions as in Experiment 1.

***Participants in the experimenter-provided misinformation were given an inaccurate description of the target. They were told to read over the description and that they would later be asked questions about it.

***Participants in the control condition did not generate nor read a description.

After 5 minutes, participants were immediately presented with a lineup and given the same directions as in Experiment 1. Half of the participants in the forced recall condition and experimenter provided description were given a warning. The warning told the participants to only consider the slide for identification, and not the self-generated or experimenter-provided descriptions.

Results & Discussion.

Identification Accuracy.

No Warning Condition - Participants in the forced recall condition performed significantly worse than participants in the control condition when no warning was given. Participants in the experimenter provided information also did worse than the control group when there was no warning, but this was not significant.

Warning Condition – Participants in the forced recall condition still did significantly worse on identification accuracy than the control group even though a warning was given. However, participants in the experimenter provided information improved on identification accuracy so that they did not differ from the control condition.

Description Quality. A discriminant analysis was used following the same procedure as that in Experiment 1. Results illustrated, as in Experiment 1, that participants who made less inaccurate descriptions made less identification errors (see table 5), thus the descriptions did predict id. accuracy.

Self-Generated Misinformation (Forced Recall Condition).

***When participant’s descriptions were self-generated, identification was significantly less accurate than the control condition, even after a 30-minute delay.

***Participants couldn’t use source monitoring to improve accuracy on identification when they generated the descriptions themselves but they could use the warning to increase identification accuracy when the experimenter provided the misinformation.

Transfer Inappropriate Retrieval (TIR) or Retrieval-Based Interference?

Schooler et. al (1997) state that verbal overshadowing is caused by the difference in verbal and configural representations (modality mismatch). When giving a verbal description, the person lists featural aspects of the target, however when presented with a recognition test, the person uses configural processes to identify the target. Schooler et. al propose that these different processes cause verbal overshadowing.

The results of Experiment 1 & 2 support Meissner et. al’s alternative theory that verbal overshadowing results when participants use different levels of criteria.