What Is Eyewitness Memory in Psychology and Why Is It Unreliable?

A common perception is that when we witness a crime, our minds simply record what happened and store it for later.

But memory does not work like a video camera.

Understanding eyewitness memory requires following its journey from perception and encoding to storage, retrieval, identification, and eventually testimony. Because these processes can influence investigations and courtroom decisions, it is crucial to comprehend the process as an application of cognitive psychology within forensic psychology.

What Is Eyewitness Memory?

Eyewitness memory refers to the memory a person forms for an event they personally witnessed. In the context of a crime, this may include information about the people involved, their actions, objects present at the scene, environmental details, and the sequence in which events occurred.

However, eyewitness memory is not a literal recording of what happened. Like other forms of memory, it involves processes through which information is perceived and encoded, retained over time, and later retrieved. What a witness remembers can therefore depend not only on what happened during the event, but also on what happened to that memory afterward.

Modern research has challenged the simple idea that eyewitness memory is either “reliable” or “unreliable.” John Wixted argues that eyewitness memory can be highly reliable under appropriate conditions, while the 2025 review by Mickes, Wilson, and Wixted emphasizes that the major problem is often memory contamination rather than an inherent inability of eyewitnesses to remember accurately.

How Does Eyewitness Memory Work?

To understand why eyewitness accounts sometimes become inaccurate, it helps to follow a memory from the moment a crime occurs to the moment a witness eventually reports or identifies what they remember.

A useful way of thinking about eyewitness memory is:

Errors can potentially occur at almost every stage – however, they are driven by different factors, depending on the stage.

Memory Formation During the Crime

Eyewitness memory begins during the event itself, rather than when the police later ask the witness what happened.

The first stage is perception. A witness must have sufficient access to information about the event for it to be encoded into memory.

Factors that Affect Eyewitness Testimony

Lighting, viewing distance, duration of exposure, obstructions, and the amount of attention available to the witness can all affect the quality of the information that enters memory. The National Academies’ review of eyewitness identification research specifically identifies the following as factors that influence perception and identification:

• Dim illumination,
• Brief viewing times,
• Large viewing distances,
• Emotional arousal, and
• Distracting elements.

Attention is particularly important because people cannot process every aspect of a complex event equally. During a crime, a witness may focus heavily on the perpetrator’s weapon, the perpetrator’s face, an escape route, or their own immediate safety. Information outside that focus may receive less attention and consequently be less well encoded.

Hence, a witness might remember what happened reasonably well while having difficulty remembering who was responsible or what was happening elsewhere in the scene.

Stress and Eyewitness Memory

Stress is often treated as though it automatically destroys eyewitness memory, but research shows a more complicated relationship.

A 2021 survey of eyewitness-memory and memory experts found considerable agreement that very high levels of stress can impair eyewitness accuracy. However, experts showed less agreement about exactly how stress affects memory under different circumstances (e.g the intensity and timing of stress and the particular aspect of memory being tested).

This distinction matters.

A witness who experiences a frightening crime should not automatically be assumed to have an inaccurate memory. Rather, extreme stress should be considered as possibly one factor that can interfere with the encoding and later retrieval of information. It would be more appropriate to base one’s judgement on the context of cognitive and environmental variables that possibly affected the eyewitness.

The Weapon Focus Effect

One particularly studied example of attentional narrowing is the weapon focus effect.

Research has found that eyewitness memory for a perpetrator or surrounding circumstances can be poorer when a weapon is present than when a non-weapon object is present.

One of the reasonings is that an unexpected, threatening weapon captures a disproportionate amount of visual attention. Hence, people employ fewer attentional resources available for other details, such as the perpetrator’s face.

Familiarity Matters

Another important consideration is whether the perpetrator is someone the witness already knows.

Recognizing a familiar person and identifying an unfamiliar stranger are not equivalent tasks.

When the perpetrator is a stranger, the witness may have only a short opportunity to encode a face that they have never encountered before. Thus, can the memory that is later used to select one person from a lineup be trusted?

A witness cannot later accurately remember information that was never adequately perceived and encoded in the first place.

The National Academies therefore emphasizes that eyewitness identification is influenced by both characteristics of the event and characteristics of the witness.

But the story does not end when the crime ends. Even a reasonably well-encoded memory can subsequently be altered, weakened, or contaminated.

What Happens to Eyewitness Memory After the Crime?

Once the event is over, the witness’s memory can go through alterations both right before and during investigations. This is because memory is reconstructive; remembering involves reconstructing an earlier experience rather than literally replaying a stored recording. Information encountered after an event can therefore become relevant to what a person subsequently remembers.

Witnesses may encounter information from police officers, prosecutors, other witnesses, family members, friends, news reports, social media, or repeated questioning. Since memory is reconstructive, the new input along with leading questions can also introduce information that was not part of the original experience.

This phenomenon is commonly associated with the misinformation effect.

Thus, a witness may genuinely believe that a detail was part of the original event when that information was actually encountered later.

The problem becomes especially serious when the same witness is repeatedly asked to identify a suspect. Modern eyewitness research suggests that the identification procedure itself can leave a memory trace that affects later identification attempts.

The case of Malcolm Alexander shows how the identification presented in court can be substantially different from the witness’s earliest memory report and level of certainty.

Retrieval of Memory and Eyewitness Identification

Retrieval is the stage at which a witness attempts to bring information about the original event back to mind.

Basically, the witness uses available retrieval cues and reconstructs the earlier experience and is not equivalent to pressing “play” on a mental video. Information encountered between the event and retrieval can therefore become intertwined with the original memory.

There is also an important distinction between recall and recognition.

Eyewitness identification is primarily a recognition task. A witness is presented with faces and asked to determine whether one corresponds to the perpetrator.

This creates another potential problem: a person can recognize a face without necessarily knowing where they recognize it from.

How Does Eyewitness Identification Work?

Remembering a face and identifying a suspect are not necessarily the same psychological task.

A witness may retain some information about a perpetrator’s appearance without being able to reliably determine which person in a lineup corresponds to that memory. The circumstances under which the face was originally viewed (lighting, distance, exposure duration, and attention) can influence the quality of the memory available for identification.

The Problems with Eyewitness Identification

Source Monitoring

The phenomenon known as source monitoring concerns the ability to distinguish the source of a remembered piece of information. In eyewitness situations, a person may remember having seen a particular face but confuse having seen that face during the crime with having encountered it later (for example, in a photograph, television broadcast, or lineup).

The famous case of Donald Thompson illustrates this possibility. Thompson was accused of rape by a woman who had seen him on television shortly before the assault. The case became a classic example of how a recently encountered face can potentially be confused with the face of an offender.

The psychological lesson is that remembering a face and remembering where that face was encountered are separable processes.

Relative Judgment and the Lineup

There is another problem: witnesses do not always evaluate each lineup member independently against their memory of the perpetrator. Under some circumstances, they may instead make a relative judgment, selecting the person who looks most like the perpetrator compared with the other people in the lineup.

This becomes particularly problematic when the actual perpetrator is not present. A witness may still feel compelled to choose the person who most closely resembles their memory, even though the correct response would be to reject the lineup.

Thus, the National Academies recommended several safeguards for eyewitness identification procedures. These include standardized instructions, telling witnesses that the perpetrator may not be present, double-blind administration so that the administrator does not know who the suspect is, recording the witness’s confidence at the initial identification, and videotaping the identification process.

Ronald Cotton: When Identification Becomes Evidence

In 1984, Jennifer Thompson was sexually assaulted in her apartment. During the investigation, she viewed a photographic lineup and initially had difficulty deciding between two photographs before identifying Cotton.

When asked whether she was sure, she said yes.

Cotton was subsequently included in a live lineup and was the only person repeated from the earlier photographic lineup. Thompson identified him again, and the police reinforced the identification by telling her that she had selected the same person.

At trial, Thompson expressed very high confidence that Cotton was her attacker. Cotton was convicted and ultimately spent more than ten years in prison before DNA testing established that he was not the perpetrator and identified Bobby Poole as the actual offender.

Cotton’s case demonstrates why eyewitness identification cannot be evaluated simply by asking whether the witness eventually appeared confident. The circumstances surrounding the first identification and everything that happened afterward matter.

This is where eyewitness memory becomes particularly relevant to forensic psychology: investigative procedures themselves can influence how a witness’s memory is tested and how an identification develops.

Does Eyewitness Confidence Mean Accuracy?

A confident witness is not necessarily an accurate witness.

However, confidence is not completely meaningless either.

The apparent contradiction comes partly from when and how confidence is measured.

Consider Ronald Cotton’s case. Thompson ultimately became highly confident that Cotton was the perpetrator, yet DNA evidence established that she was mistaken. Her identification had also been followed by reinforcing feedback and another identification procedure, providing opportunities for her confidence to change after the initial decision.

This is why simply asking a witness at trial, “How certain are you?” can be misleading. A person’s confidence may have been influenced by everything that happened between the original identification and the courtroom.

Applying Eyewitness Confidence the Right Way

Research summarized by Wixted and Roediger makes an important distinction: confidence expressed during the initial, properly conducted memory test can contain useful information about accuracy, whereas confidence measured after repeated testing and feedback can be substantially less informative. Their analysis argues that the legal system should place particular weight on the witness’s first identification and the confidence expressed at that time.

The National Academies similarly recommends recording the witness’s confidence at the time of the initial identification, rather than relying on a later statement of certainty.

This produces a more nuanced conclusion than either “confidence equals accuracy” or “confidence means nothing.”

The better question is:

How confident was the witness at the first identification, under what conditions was that identification made, and what happened to the witness’s memory afterward?

Conclusion

Eyewitness memory is not a perfect record of an event. What a witness ultimately remembers and reports can be shaped by how the event was perceived, how the memory was later retrieved, and what happened between the event and identification. This does not make eyewitness evidence inherently unreliable; rather, its accuracy depends on the conditions under which the memory was formed and tested. Understanding these vulnerabilities is therefore essential to evaluating eyewitness testimony in forensic settings.