Results from Loftus and Palmer's (1974) study showing how the verb used in the initial description of a car accident affected recall of the incident after one week.

One way in which eyewitness testimony can be distorted is via confirmation bias. This occurs when what is remembered of an event is influenced by the observer's expectations. For example, students from two universities in the United States (Princeton and Dartmouth) were shown a film of a football game involving both universities. The students showed a strong tendency to report that their opponents had committed many more fouls than their own team.

Does it make any difference to the memory of an eyewitness whether the crime observed by him or her is violent? A study by Loftus and Burns (1982) suggests that the answer is "yes". Participants saw two filmed versions of a crime. In the violent version, a young boy was shot in the face near the end of the film as the robbers were making their getaway. Inclusion of the violent incident caused impaired memory for details presented up to two minutes earlier. Presumably the memory-impairing effects of violence would be even greater in the case of a real-life crime, because the presence of violent criminals might endanger the life of any eyewitness.

Post-event information

Elizabeth Loftus has shown very clearly that the memory of an incident can be systematically distorted by the questioning that occurs subsequently. To illustrate this point, we will discuss a study by Loftus and Palmer (1974). Participants were shown a film of a multiple car accident. After viewing the film, the participants described what had happened, and then answered specific questions. Some were asked, "About how fast were the cars going when they smashed into each other?", whereas for other participants the verb "hit" was substituted for "smashed into". Control participants were not asked a question about car speed. The estimated speed was affected by the verb used in the question, averaging 41 mph when the verb "smashed" was used versus 34 mph when "hit" was used. Thus, the information implicit in the question affected the way in which the accident was remembered.

One week later, all the participants were asked, "Did you see any broken glass?". There was actually no broken glass in the accident, but 32% of the participants who had been asked previously about speed using the verb "smashed" said they had seen broken glass. In contrast, only 14% of the participants asked using the verb "hit" said they had seen broken glass, and the figure was 12% for the control participants who had not been asked a question about speed (see Figure 8.5). Thus, our memory for events is fragile and susceptible to distortion.

Even apparently trivial differences in the way in which a question is asked can have a marked effect on the answers elicited. Loftus and Zanni (1975) showed people a short film of a car accident, and then asked them various questions. Some eyewitnesses were asked, "Did you see a broken headlight?", whereas others were asked, "Did you see the broken headlight?". In fact, there was no broken headlight in the film, but the latter question implied that there was. Only 7% of those asked about a broken headlight said they had seen it, compared to 17% of those asked about the broken headlight.

The tendency for post-event information to distort memory presumably depends in part on individual differences in susceptibility to misinformation. This issue was studied by Tomes and Katz (1997). Those who habitually accepted misinformation possessed the following characteristics:

• Poor general memory for the event for items of information not associated with misinformation.

• High scores on imagery vividness.

• High empathy scores, indicating that they were good at identifying with the moods and thoughts of others.

More research is needed to clarify the role of individual differences in susceptibility to misinformation.

The notions that eyewitness memory is fragile and easily distorted were shown strikingly by Schooler and Engstler-Schooler (1990). They presented their participants with a film of a crime. After that, some participants provided a detailed verbal report of the criminal's appearance, whereas others did an unrelated task. Finally, all the participants tried to select the criminal's face on a recognition test. Those who had provided the detailed verbal report performed worse than the other participants on this test. This phenomenon (termed verbal overshadowing of visual memories) presumably occurred because the verbal reports interfered with recollection of the purely visual information about the criminal's face.

Theoretical views

How does misleading post-event information distort what eyewitnesses report? According to Loftus (1979), information from the misleading questions permanently alters the memory representation of the incident: the previously formed memory is "overwritten" and destroyed. In support of this position, Loftus showed that it can be very difficult to retrieve the original memory. In one study, Loftus (1979) offered her participants $25 if their recall of an incident was accurate. This incentive totally failed to prevent their recollections being distorted by the misleading information they had heard.

The notion that the original memory of an event is destroyed by post-event information is not generally accepted, because there is evidence that the original information remains in long-term memory. For example, Dodson and Reisberg (1991) used an implicit memory test to show that misinformation had not destroyed the original memories of an event. They concluded that misinformation simply makes these memories inaccessible.

Loftus (1992) argued for a less extreme position than the one she had adopted previously. She emphasised the notion of misinformation acceptance: the participants "accept" misleading information presented to them after an event, and subsequently they regard it as forming part of their memory of that event. There is a greater tendency to accept post-event information in this way as the time since the event increases.

Zaragoza and McCloskey (1989) argued for a simpler explanation. According to them, participants do what they think is expected. Suppose, for example, they see slides of an accident involving a man using a hammer, but then read an account of the incident in which the instrument is a screwdriver. They are then asked to decide whether the instrument used was a hammer or a screwdriver. Participants who cannot recollect the instrument used by the man may remember that it was described in the subsequent account as a screwdriver. They may feel they will please the experimenter (and show they were paying attention to the slides) by selecting the screwdriver. Thus, the participants are simply playing along with what they think is expected of them; this is known as responding to the demand characteristics of the situation.

Evidence inconsistent with this view was reported by Lindsay (1990). He presented mis leading information in a narrative account after showing slides in which a maintenance man stole money and a calculator from an office. After that, the eyewitnesses were told truthfully that any information in the narrative account relating to the subsequent memory test was wrong. These instructions should have prevented distorted memory performance if demand characteristics were operating. In fact, memory for the incident by the misled participants was distorted by the post-event information, suggesting that this information had genuinely affected memory.

The effects of post-event misinformation on eyewitness memory can also be understood within the source monitoring framework (Johnson, Hashtroudi, & Lindsay, 1993). A memory probe (e.g., question) activates memory traces having informational overlap with it; this memory probe may activate memories from various sources. The individual decides on the source of any activated memory on the basis of the information it contains. What is of relevance here is the possibility of source misattribution. If the memories from one source resemble those from another source, this will increase the chances of source misattribution. If eyewitnesses falsely attribute the source of misinformation to the original event, then misinformation will form part of their recall of the event. In essence, it is assumed that separate memories are stored of the original event and the misinformation, with potential memory problems occurring at the time of retrieval.

A key prediction from the source monitoring framework is as follows: any manipulation that increases the extent to which memories from one source resemble those from another source increases the likelihood of source misattribution. Support for this prediction was reported by Allen and Lindsay (1998). They presented two narrative slide shows describing two different events with different people in different settings. Thus, the participants knew that the post-event information contained in the second slide show was not relevant to the event described in the first slide show. However, some of the details in the two events were rather similar (e.g., a can of Pepsi vs. a can of Coca-Cola). This caused source misattribution, and led the participants to substitute details from the post-event information for details of the event itself. These findings were obtained with an interval of 48 hours between the two events, but not when there was no time gap. Presumably the participants in the latter condition noticed the resemblances in the details incorporated in the two events, and this helped to reduce source misattribution.

Much research in this area can be interpreted within Bartlett's (1932) theory (see Chapter 12). According to Bartlett, retrieval involves a process of reconstruction, in which all of the available information about an event is used to reconstruct the details of that event on the basis of "what must have been true". On that account, new information relevant to a previously experienced event can affect recollection of that event by providing a different basis for reconstruction. Such reconstructive processes may be involved in eyewitness studies on post-event information.

In sum, most of the distorting effects of misleading post-event probably reflect real effects on memory. These effects may involve difficulties of gaining access to the original memory (e.g., because of interference) as was proposed by Loftus (1992), or they may depend on source misattribution. Many distortions may well occur as a consequence of the reconstructive processes emphasised by Bartlett (1932).

An important limitation of most research is its focus on memory for peripheral details of events (e.g., presence or absence of broken glass). As Fruzzetti et al. (1992) pointed out, it is harder to use post-event information to distort witnesses' memory for key details (e.g., the murder weapon) than for minor details.

Eyewitness identification

Eyewitness identification from identification parades or line-ups is often very fallible (see Wells, 1993, for a review). Shapiro and Penrod (1986) argued that eyewitness identification studies typically produce inferior memory performance to more traditional face recognition studies. One key difference is that the same stimuli (e.g., photographs) are used at acquisition and at test in traditional studies of face recognition, whereas the facial appearance of someone may differ substantially between a staged incident and the subsequent identification parade.

One factor influencing the likelihood of an incorrect identification is the functional size of the line-up. This is the number of people in the line-up matching the eyewitness's description of the culprit. If, for example, the eyewitness recalled only that the culprit was a man, then the functional size of a line-up consisting of three men and two women would be three rather than five. When the actual culprit is absent, low functional size of line-up is associated with a greater probability of mistaken identification (Lindsay & Wells, 1980).

The probability of mistaken identification is also influenced by whether or not the eyewitness is warned that the culprit may not be in the line-up (Wells, 1993). This is probably especially important with real-life line-ups, because eyewitnesses may feel the police would not have set up an identification parade unless they were fairly certain the actual culprit was present.

Wells (1993, p. 560) argued that a small functional line-up and lack of warning that the culprit may be absent produce mistaken identifications because eyewitnesses tend to use relative judgements: "The eyewitness chooses the line-up member who most resembles the culprit relative to the other members of the line-up." How can we reduce eyewitnesses' reliance on the relative judgement strategy? One approach is sequential line-ups, in which members of the line-up or identification parade are presented one at a time. Sequential line-ups reduce the effects of functional size and failure to warn of possible culprit absence on mistaken identification (Lindsay et al., 1991).

Other factors in eyewitness testimony

There has been much research on eyewitness testimony. Those factors that deserve special mention are those that are regarded by eyewitness experts as generally reliable and valid, and which do not correspond to common sense. Kassin, Ellsworth, and Smith (1989) compiled a list of such factors (with percentages of experts believing each statement to be commonsensical in brackets):

• An eyewitness's confidence is not a good predictor of his or her identification accuracy (3%).

• Eyewitnesses tend to overestimate the duration of events (5%).

• Eyewitness testimony about an event often reflects not only what the eyewitness actually saw but information they obtained later on (7.5%).

• There is a conventional forgetting curve for eyewitness memories (24%).

• An eyewitness's testimony about an event can be affected by how the questions are worded (27%).

• The use of a one-person line-up increases the risk of misidentification (29%).

Why is an eyewitness's confidence a poor predictor of identification accuracy? This issue was studied by Perfect and Hollins (1996). The participants were given recognition memory tests for the information contained in a film about a girl who was kidnapped, and for general knowledge questions. Accuracy of memory was not associated with confidence with questions about the film, but it was with the general knowledge questions. Perfect and Hollins (1996, p. 379) explained this difference as follows:

Individuals have insight into their strengths and weaknesses in general knowledge, and tend to modify their use of the confidence scale accordingly .So, for example, individuals will know whether they tend to be better or worse than others at sports questions. However, eyewitnessed events are not amenable to such insight: subjects are unlikely to know whether they are better or worse . . . than others at remembering the hair colour of a participant in an event, for example.

Perfect and Hollins (1996) found that eyewitnesses typically had more confidence in their accurate answers than in their inaccurate ones. Thus, they could discriminate among the quality of their own memories to some extent, even though they did not know whether they were better or worse than others at remembering details of an event.

Psychologists have made a valuable contribution to ensuring that justice is done in criminal cases. For example, John Demjanjuk was convicted of being "Ivan the Terrible", the person who operated the gas chambers at Treblinka concentration camp. The main evidence consisted of eyewitness testimony given by survivors of the camp 40 years after the war. Psychologists warned of the fallibility of eyewitness testimony over such long periods, and their warnings seem to have been justified by the subsequent overturning of the conviction.

Cognitive interview

The questions asked during a police interview may unwittingly distort an eyewitness's memory and so reduce its reliability. What used to happen was that an eyewitness's account of what had happened was often interrupted repeatedly by police, with the question-answer format being used excessively. The interruptions made it hard for the eyewitness to concentrate, thus reducing recall. In response to psychological research, the Home Office issued guidelines a few years ago recommending that police interviews should proceed from free recall to general open-ended questions, concluding with more specific questions.

According to Fisher and Geiselman (e.g., Geiselman, Fisher, MacKinnnon, & Holland, 1985), interview techniques should be based on the following notions:

• Memory traces are usually complex and contain various kinds of information.

• The effectiveness of a retrieval cue depends on its informational overlap with information stored in the memory trace; this is the encoding specificity principle (see Chapter 6).

• Various retrieval cues may permit access to any given memory trace; if one retrieval cue is ineffective, find another one. For example, if you cannot think of someone's name, form an image of that person, or think of the first letter of their name.

Geiselman et al. (1985) used these notions to develop the basic cognitive interview:


Cognitive Standard Hypnosis interview interview

Stop Anxiety Attacks

Stop Anxiety Attacks

Here's How You Could End Anxiety and Panic Attacks For Good Prevent Anxiety in Your Golden Years Without Harmful Prescription Drugs. If You Give Me 15 minutes, I Will Show You a Breakthrough That Will Change The Way You Think About Anxiety and Panic Attacks Forever! If you are still suffering because your doctor can't help you, here's some great news...!

Get My Free Ebook

Post a comment