Eyewitness Testimony: How Memory Works in Legal Cases
Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy
Eyewitness testimony is evidence given by a person who reports seeing, hearing, or otherwise perceiving an event relevant to a legal case. It can help investigators reconstruct events, test competing explanations, identify people, and evaluate allegations. Its reliability depends on what was perceived, how memory was encoded and retrieved, what happened between the event and the first account, and how investigators and courts obtained and interpreted the evidence.
In psychological science, a sincere witness can recall important facts accurately and can also make significant errors. A convincing narrative, a vivid image, an honest belief, and a legally admissible identification represent different questions. The central scientific task is to preserve the earliest available evidence and understand the conditions under which each recollection was produced. The National Academies' landmark eyewitness identification review and a newer 2025 cognitive-science synthesis provide the foundation for this approach.
What Is Eyewitness Testimony?
Eyewitness testimony includes an oral or written account by someone who personally encountered an event. In a criminal case, this might involve describing an offender's actions, a vehicle, spoken words, the sequence of events, or a person's face. In civil litigation, it might involve observations of a traffic collision, workplace event, or disputed interaction. The legal significance of the account depends on the issue the court must decide and the jurisdiction's evidentiary rules.
Three related forms of evidence deserve separate treatment. Event memory concerns what happened; person identification concerns whether a particular individual was present; and source memory concerns where a remembered detail came from. A witness may remember an event accurately but identify the wrong person, or recognize a face while misremembering where it was encountered. These are separate cognitive judgments, each potentially affected by different conditions.
Researchers also distinguish recall, in which someone retrieves information with relatively few cues, from recognition, in which someone evaluates a presented option. A free account of a robbery and a choice from a photo lineup are different tests of memory. The National Research Council emphasized that basic vision and memory science and applied identification research should inform how these tests are interpreted.
How Eyewitness Memory Works
Perception and encoding at the scene
A witness encodes only some of the sensory information available during an event. What enters memory depends on viewing distance, lighting, exposure duration, movement, divided attention, visual obstructions, expectations, and personal familiarity with the observed person or setting. Even when an incident feels unforgettable, its individual visual details may have received little attention. No later interview can fully recover information that was never perceptually registered.
Attention is selective. Someone concentrating on escaping danger may remember a route or a sound better than the face of an unfamiliar person. A bystander watching a vehicle may register its color or direction but not a license plate. Thus accuracy is detail-specific: good memory for one element offers no automatic guarantee for another.
Retention and the time between event and report
After an event, a memory trace can be retained while details become less accessible or more vulnerable to interference. Time alone does not provide a simple mathematical formula for witness accuracy. The retention interval must be considered alongside opportunities for rehearsal, discussion, exposure to photographs or news, intervening events, and the kind of information being recalled. Repetition can strengthen accessibility while also strengthening a mistaken interpretation.
An account obtained promptly, before multiple outside influences, has important evidentiary value. The American Psychology–Law Society recommendations encourage an early recorded witness interview. This safeguards what the witness could report at that point, including uncertainty, rather than treating subsequent confidence as proof of what was remembered initially.
Retrieval and reconstruction
Remembering involves reconstructing information from stored traces and retrieval cues. A memory may preserve its central meaning while omitting peripheral details, or include an inference that the witness later experiences as a perception. Carefully worded questions and open invitations help document the witness's own account. Questions that presuppose an answer can instead introduce information into the retrieval process.
In a classic experimental demonstration, participants who watched footage of vehicle collisions gave different estimates and later reported different details depending on the wording of questions. The Loftus and Palmer experiment80011-3) established that postevent language can shape reported memory. Its controlled film-based design demonstrates a mechanism; it cannot assign an error probability to any particular real-world witness.
Eyewitness Testimony Versus Eyewitness Identification
An eyewitness account may establish that a person saw someone flee a location without establishing the identity of that person. Eyewitness identification is the narrower task of deciding whether a person in a lineup, photo array, showup, or courtroom is the individual seen previously. Recognition decisions are affected by perceptual discriminability, the witness's decision threshold, the fairness of the comparison set, instructions, and suggestive feedback.
A suspect identification, rejection of the lineup, filler choice, failure to identify, and expression of uncertainty all carry information. Investigators should preserve each outcome rather than counting only selections of the police suspect as evidence. The 2025 review of eyewitness memory places special emphasis on what the first reasonably uncontaminated test of recognition can reveal, including information consistent with a suspect's innocence.
Identification procedures deserve their own specialist analysis, including lineup composition, showups, administrator blinding, initial confidence, and simultaneous versus sequential presentation. The purpose of this overview is to explain their relationship to memory science, while the planned EN-58 identification articles will own the detailed procedural comparisons.
Estimator Variables and System Variables
The distinction between estimator and system variables is essential to forensic eyewitness science. Estimator variables are circumstances that the justice system generally cannot control after the event: distance, lighting, exposure time, stress, disguise, witness–perpetrator familiarity, cross-race recognition conditions, and delay. System variables are features that investigators or institutions can control: how witnesses are interviewed, lineup instructions, who administers a lineup, which photographs appear, whether a witness sees the suspect repeatedly, and whether feedback is provided.
Both matter, but only the system variables offer direct opportunities to reduce preventable procedural contamination. The National Academies report and Wells and colleagues' scientific review recommend safeguards aimed at preserving the evidentiary value of the initial account and identification.
Lighting, distance, exposure, and disguise
Clear face recognition requires sufficient perceptual information. A witness who viewed a stranger for a moment in dim light at a substantial distance has a different information base from a witness who interacted with a familiar person in daylight. Case evaluations should reconstruct opportunity to observe with photographs, geometry, timelines, and contemporaneous descriptions when available, while recognizing that reconstructions are approximations.
Disguise, changes in hair or facial presentation, and intervening images can alter recognition. A witness may identify someone through distinctive clothing or context rather than through a clear memory of their face. The source of the claimed recognition should therefore be documented in the witness's own words.
Stress, attention, and weapon focus
Stress changes attention and memory in complex ways. A meta-analysis of high-stress eyewitness studies found lower average accuracy for both person identification and event details under high stress. Its results concern groups and study conditions, not a rule that a frightened witness is invariably mistaken. The real impact depends on timing, duration, task, stimulus, and context.
The weapon-focus literature examines how the presence of a weapon may capture attention and reduce memory for other details. A meta-analytic review reported a measurable average effect, with differences across measures. Because the research mixes controlled settings and varying operationalizations, forensic interpretation must consider whether an actual witness focused on the object and which facts were observed.
Cross-race face recognition
Research has found average differences in recognition performance for faces belonging to one's own racial group versus less familiar racial groups. A meta-analysis by Meissner and Brigham examined the phenomenon across multiple samples. This is a statistical finding about face recognition, influenced by experience and research design. It provides no basis for judging the honesty or competence of a specific witness from racial identity alone, nor does it determine whether an individual identification is correct.
Why Eyewitness Memories Sometimes Change
Suggestion and misleading information
Information encountered after an event may become incorporated into later reports. A leading question, a misleading description from another witness, a photograph, a social media post, or a confident assertion from an investigator can affect what is retrieved or attributed to personal experience. The classic misinformation experiments80011-3) illustrate that the route by which information reaches the witness can matter as much as the apparent vividness of recall.
It is important to document when each new detail entered the record. An account that changes after an interview may reflect ordinary retrieval, clarification, inference, external influence, or some combination. The change itself does not establish deceit and does not identify the exact mechanism.
Conversations with other witnesses
Witnesses sometimes discuss events together before investigators obtain individual reports. Those discussions can improve awareness of shared facts but can also cause one person to report information originally supplied by another. In an experimental study, Gabbert, Memon, and Allan demonstrated memory conformity following co-witness discussion. Its experimental percentage should not be presented as the prevalence of contamination in criminal cases.
Where practical, early separate interviews and records of intervening conversations make it easier to determine which details came from independent perception. Two matching accounts are less corroborative if both derive from one person's narrative.
Source confusion and repeated exposure
A witness may encounter a face in a photograph, news story, or previous identification attempt and later experience that face as familiar. Familiarity alone does not resolve whether the person was seen during the event or afterward. Multiple identification tests with the same suspect can also complicate interpretation by giving the face additional exposure.
This is why the chronology of every identification attempt is part of the evidence: which images were shown, when, by whom, in what order, with which instructions, and with what immediate response. Without that record, a later courtroom identification may conceal how recognition developed.
Is Eyewitness Testimony Reliable?
The scientifically useful answer depends on the procedure, observation conditions, and specific assertion. Eyewitnesses can provide highly valuable evidence. They can also produce false identifications in good faith. Research supports a differentiated evaluation rather than a blanket assumption that eyewitness testimony is either trustworthy or inherently worthless.
A major development concerns initial, uncontaminated identification evidence. Wixted and Wells synthesized evidence that initial confidence can strongly differentiate more accurate from less accurate suspect identifications when the lineup is fair, the test is first and uncontaminated, and the administrator supplies no influence. These findings do not transfer automatically to later confidence, suggestive lineups, repeated tests, or courtroom declarations.
A 2025 cognitive-science review further argues that the first clean memory test can preserve valuable evidence of both guilt and innocence. Its conclusions are part of an active scientific discussion about experimental design, decision criteria, field generalization, and interpretation of identification choices.
Confidence and accuracy are different constructs
Confidence is a person's subjective certainty about a response. Accuracy concerns whether that response corresponds to the relevant external fact. Confidence can carry diagnostic information when measured at an appropriate moment using controlled procedures; it is not a direct measurement of truth. A witness saying “I am certain” at trial may reflect original memory, later feedback, repeated rehearsals, or courtroom preparation.
Bradfield, Wells, and Olson demonstrated how confirming feedback can damage the relationship between certainty and identification accuracy. The National Academies therefore recommend recording the witness's confidence immediately and in the witness's own words.
Why courtroom confidence can mislead
Courtroom testimony often occurs months or years after the event. By then a witness may have read documents, seen media, participated in pretrial proceedings, received feedback, or repeatedly rehearsed an identification. Jurors observe the final performance, not the entire memory history. The appropriate question is how the witness's certainty and details evolved from the first documented report through every later exposure.
A consistent story may reflect stable memory or stable rehearsal. A changed story may reflect correction, genuine retrieval, suggestion, or misunderstanding. Either interpretation requires an evidence-based chronology.
How Accurate Are Eyewitness Identifications?
There is no single universal accuracy percentage for all eyewitness identifications. A study's apparent accuracy depends on whether the perpetrator is present in the lineup, whether the suspect is innocent, how fillers are selected, the witness population, the measure being reported, and how responses are scored. Suspect selections, filler selections, and lineup rejections have different implications.
In signal detection research, discriminability describes how well witnesses separate previously seen from unseen individuals, while response bias describes how readily they choose someone. A procedure can reduce the total number of identifications partly by making witnesses less willing to choose. That numerical change is not necessarily an improvement in discrimination. Recent research examines these competing metrics in detail, including experimental analyses published in 2025.
Likewise, a positive predictive value reported for a particular lineup design cannot be applied mechanically to a different case. It varies with experimental conditions and the proportion of target-present versus target-absent lineups. Interpretation requires base rates, proper comparators, and the distinction between research calibration and individual legal proof.
How Police Interviews Affect Memory Evidence
An evidence-preserving interview begins with the witness's own unprompted account. Investigators can invite the person to describe what happened from the beginning, use neutral follow-up questions, distinguish direct observations from guesses, and record the timing and content of each prompt. The goal is to document accessible information while limiting preventable suggestion.
The National Institute of Justice's eyewitness evidence guide addresses initial reporting, follow-up interviewing, showups, lineups, and recording. The 2020 American Psychology–Law Society recommendations update the evidence base and emphasize early video-recorded interviews and appropriate identification procedures.
Interviews should make room for uncertainty. “I did not see the driver's face” can be an accurate and useful answer; transforming it into a forced choice can create unsupported precision. Investigators should also distinguish what a witness personally noticed from information learned from police, another witness, or public coverage.
The cognitive interview
Structured interviewing approaches may use open retrieval, contextual prompts, and other methods intended to improve the quantity of relevant remembered information. Their value depends on proper training and application, and any gain in details must be evaluated alongside the possibility of additional inaccuracies. A technique shown useful in a study should not be treated as a universal protocol for every witness, population, or legal setting.
Sensitive interviewing of children, traumatized people, or people with communication difficulties requires developmentally and linguistically appropriate methods. Neutral wording, breaks, accommodations, and trained interviewers can improve access and reduce avoidable burdens. A witness's emotional response or communication style is not a psychological lie detector.
Lineups, Showups, and the First Identification
A photo lineup typically presents a suspect with fillers who resemble the witness's initial description. A showup presents one person, often near the incident. Live lineups use people rather than photographs. These practices carry different risks of suggestion; the applicable legal rules and departmental policies vary.
For a fair lineup, an administrator should avoid communicating which person is suspected, witnesses should be told the perpetrator may be absent, and the initial choice or rejection should be recorded fully. The U.S. Department of Justice's 2017 federal photo-array policy emphasizes blind or blinded administration and immediate documentation of confidence. A federal departmental procedure is not the law of every U.S. state or every nation.
The simultaneous-versus-sequential lineup debate illustrates the need to look beyond simple percentages. Different procedures can affect response criteria, discriminability, and the balance between accurate suspect identifications and innocent-suspect errors. The 2025 cognitive review interprets more recent evidence as supporting the diagnostic value of simultaneous presentation, while practice should follow the applicable evidence-informed policy and remain attentive to disagreement about metrics and methods.
What Eyewitness Testimony Can and Cannot Establish
A witness can testify about perceived conduct, descriptions, timelines, conversations, and an identification attempt. Psychology can assess conditions associated with more or less informative memory evidence and can explain how procedures can change reported memory. It cannot turn a witness's narrative into a laboratory-certified determination that a defendant committed an offense.
The legal inference from a witness report to guilt requires attention to other evidence, alternative explanations, opportunity for cross-examination, procedural history, and the governing standard of proof. Psychological research evaluates memory and judgment under defined conditions; a judge or jury applies the relevant law to the case.
Eyewitness evidence can also be exculpatory. A witness who fails to identify a suspect in an appropriately conducted first test, or identifies a filler, may provide information that challenges the investigative theory. The recent scientific synthesis highlights the importance of preserving such initial outcomes rather than replacing them with later confident identifications.
Eyewitness Testimony and Wrongful Convictions
Documented exonerations demonstrate the stakes of mistaken identification. The Innocence Project's historical compilation of the first 375 U.S. DNA exonerations reports eyewitness misidentification in 69% of those cases. This percentage describes a selected exoneration dataset, not the error rate of eyewitness testimony in all prosecutions or the fraction of all convictions that are wrongful.
Wrongful-conviction cases can involve interacting factors, including confirmation bias, suggestive procedures, misleading forensic interpretations, inadequate disclosure, or an inability to revisit early decisions. An eyewitness may sincerely believe a mistaken identification. The scientific lesson is to preserve the original evidence and investigate changes in confidence or content, especially when later statements diverge from the first record.
The latest available Innocence Project impact data use a different, organization-specific denominator. Statistics drawn from these two datasets should not be combined as if they measured the same population.
What Courts Do With Eyewitness Evidence
Admissibility and weight are legal questions. Admissibility asks whether the evidence may be heard under the applicable law. Weight asks how much significance a judge or jury should assign it. Psychology can inform both discussions, but scientific consensus and constitutional doctrine are separate authorities.
In Perry v. New Hampshire (2012), the U.S. Supreme Court held that the federal Due Process Clause does not require a preliminary judicial reliability inquiry when the identification was not obtained through unnecessarily suggestive circumstances arranged by law enforcement. The ruling addresses a U.S. constitutional threshold; state rules and procedural safeguards may provide additional protections. It does not establish that every admitted identification is accurate.
Cross-examination, expert evidence, jury instructions, corroboration, and access to recordings may help assess reliability. Their use and admissibility differ across jurisdictions. Expert witnesses should describe research-based mechanisms and the limits of generalizing laboratory findings to the case, as discussed in a 2022 analysis of memory expert testimony.
Expert testimony about eyewitness memory
A psychologist can explain how postevent information, source monitoring, lineup fairness, administrator influence, and confidence timing operate. The expert can evaluate whether the documented procedure contained known risk factors and identify gaps in the record. The expert normally should avoid presenting a personal judgment that a particular witness is truthful or that a defendant is guilty as a psychological finding.
Research on juror decision-making also matters because witness demeanor and expressed confidence can influence impressions. An evidence-based court evaluation separates credibility, accuracy, procedural fairness, admissibility, and ultimate legal findings.
How to Evaluate an Eyewitness Account in a Specific Case
A case-specific review begins by reconstructing a timestamped evidence history. What did the witness observe before any discussion? How long and from what position? What was said during the first emergency call? Which details were volunteered without prompting? Was the first interview recorded? Did the witness encounter other accounts or images before a lineup? How was the identification administered, and what was the exact immediate response?
These questions require documentation and corroboration rather than intuition about body language. A report of high certainty should be placed next to the earliest confidence statement. A description should be compared with the original transcript, not merely a later summary. Police files should preserve rejections, filler choices, and uncertain answers as well as confident suspect choices.
External consistency can matter when independent evidence supports or challenges particular facts. Yet two witnesses are not fully independent corroboration if they discussed the scene together, and a matching statement may share an information source with the investigator's question. Source tracing is therefore central to forensic evaluation.
Practical Safeguards for Investigators and Legal Professionals
The strongest recurring recommendations are early, recorded, neutral interviewing; careful separation of witnesses before their initial accounts; retention of the full history of statements and identification attempts; fair lineup construction; instructions that the perpetrator might be absent; blind or blinded administration; immediate verbatim confidence documentation; and avoidance of feedback. These practices address procedural risk while leaving perceptual limitations in place.
The 2020 scientific review, National Academies synthesis, and federal justice guidance offer complementary levels of authority: scientific recommendation, national scientific consensus report, and a defined federal policy. They should not be treated as interchangeable legal commands.
A 2025 nationwide U.S. survey report documents continuing variation and implementation challenges across law enforcement policies. Written best practices and what actually happens in a particular investigation must therefore be checked separately.
What Witnesses Can Do to Preserve Their Account
A witness can describe personal observations as soon as reasonably possible, distinguish what was seen from what was inferred, state uncertainty plainly, and keep a record of later exposures to information. Reporting an omitted detail later is legitimate, but the timing and source of that detail are important for interpretation. Witnesses should avoid comparing memories with other witnesses before giving independent accounts when possible.
A witness who has already discussed the event, seen images, or received feedback should disclose that history. This helps an investigator or court evaluate the evidence and does not automatically disqualify the person. Witnesses should follow relevant legal directions and seek appropriate legal or victim-support advice in an actual proceeding.
What Defendants, Attorneys, and Jurors Should Understand
A mistaken identification can be genuinely believed. A low-confidence initial identification should not be retroactively described as high confidence merely because the witness now speaks firmly. Conversely, a well-administered first identification made with high initial confidence can provide meaningful evidence, subject to the quality of the viewing conditions and the rest of the record.
An attorney evaluating a case may compare first reports, lineups, recordings, disclosures, and independent physical evidence. Jurors can ask conceptually whether a detail was observed, remembered, suggested, inferred, or learned afterward. The governing evidentiary instructions remain those given by the court, which vary by jurisdiction.
Common Myths About Eyewitness Testimony
“A confident eyewitness is always accurate.”
Confidence depends on the timing and quality of its measurement. Initial certainty under fair, uncontaminated conditions can be informative; certainty years later after repeated exposure can reflect sources other than the original event. The 2017 synthesis explains the key conditions.
“If a witness remembers one detail, the entire story is accurate.”
Memory is not a single all-or-nothing record. Accurate recollection of an action can coexist with a mistaken estimate of time, distance, or identity. Each claim should be evaluated on its own evidentiary basis.
“Stress makes every witness unreliable.”
High stress can reduce average identification and detail accuracy in research, but it does not diagnose an individual report as false. The stress meta-analysis addresses aggregate effects that require careful case interpretation.
“A lineup selection proves guilt.”
A lineup records a recognition judgment made under a particular procedure. The identity of the perpetrator and the defendant's legal responsibility are determined by the whole evidentiary record and applicable law.
“Repeatedly showing a suspect makes the final identification stronger.”
Repetition can increase familiarity with the displayed face and complicate source attribution. The earliest properly documented test is often the most interpretable, as emphasized in the 2025 research review.
“Changes in a story show that a witness is lying.”
Changes can arise from a variety of processes, including ordinary memory limits, clarification, suggestion, and deliberate fabrication. The pattern and chronology matter, and no single discrepancy operates as a valid lie detector.
Frequently Asked Questions
What is an example of eyewitness testimony?
A person says they saw a vehicle leave an intersection after a crash, describes its direction of travel, and reports what the driver looked like. The account of the vehicle's movement and an identification of the driver are distinct evidentiary claims.
How does memory affect eyewitness testimony?
Perception determines what information is initially encoded. Retention and intervening information affect what remains accessible, while interview prompts and recognition tasks influence what is retrieved and reported. The conditions at each stage shape evidentiary value.
Why do eyewitnesses identify the wrong person?
Possible contributors include limited viewing information, familiarity with an innocent person's face from a later photograph, suggestive identification procedures, source confusion, administrator feedback, and ordinary recognition error. None of these mechanisms must be present in every mistaken identification.
Can eyewitness testimony be used as evidence in court?
Yes. Courts regularly consider eyewitness testimony, subject to evidentiary and constitutional rules that vary by jurisdiction. Admissibility is distinct from scientific reliability and from the ultimate assessment of guilt or liability.
Is eyewitness testimony primary evidence?
It is often direct evidence of what someone claims to have personally perceived. “Primary evidence” has different technical meanings in different legal contexts, so the phrase alone does not indicate that a witness account is conclusive or necessarily accurate.
Can two eyewitnesses corroborate each other?
They can provide independent corroboration when their observations and reports are genuinely independent. Prior discussion or exposure to a shared account can introduce dependence between testimonies, which should be documented.
Does an eyewitness have to identify someone in a lineup?
A fair procedure should allow a witness to reject the lineup or say they are uncertain. An identification test must not be understood as a demand that someone be selected.
Do children have accurate eyewitness memories?
Children can remember events and provide useful evidence, but memory reports depend on development, interview technique, repetition, question wording, and communication needs. Age by itself neither proves nor disproves the accuracy of a particular account.
What matters most about eyewitness confidence?
The most meaningful confidence record is usually an immediate statement collected during a first fair and uncontaminated identification attempt. Later expressions can be influenced by events after identification, as discussed in the National Academies recommendations.
Does eyewitness misidentification explain most wrongful convictions?
It is a major documented contributor in selected DNA exoneration datasets. The Innocence Project's 1989–2020 dataset must not be interpreted as representative of every conviction or legal case.
Can an expert determine whether a particular eyewitness is truthful?
An expert can assess known memory mechanisms and procedural conditions. General psychological research cannot directly observe a past event or certify the truthfulness of one person from demeanor, confidence, or narrative style.
What is the difference between eyewitness testimony and false confession evidence?
Eyewitness testimony concerns a witness's report of a perceived event; confession evidence concerns a suspect's own statement. Both may be affected by interviewing conditions and suggestion, but they involve different investigative and legal mechanisms. For the confession-specific evidence, see Coercive Interrogation: Psychological Mechanisms and False-Confession Risk.
Conclusion
Eyewitness testimony is a valuable form of human evidence whose strength depends on the conditions under which memories were formed, first tested, preserved, and later interpreted. A scientifically responsible assessment preserves contemporaneous information, distinguishes identification from event recall, examines estimator and system variables, and evaluates initial confidence within the proper procedural context. It integrates independent corroboration and respects the separation between psychological findings and legal decisions.
The broader professional setting for this work is described in Forensic Psychology: What It Is and What Forensic Psychologists Do. The accompanying EN-58 eyewitness articles will examine memory errors, misinformation, lineups, confidence, cross-race recognition, weapon focus, co-witness contamination, and court procedures in greater technical depth as those individual canonical pages become available.
Related Articles
References
Bradfield, A. L., Wells, G. L., & Olson, E. A. (2002). The damaging effect of confirming feedback on the relation between eyewitness certainty and identification accuracy. Journal of Applied Psychology, 87(1), 112–120. https://doi.org/10.1037/0021-9010.87.1.112
Deffenbacher, K. A., Bornstein, B. H., Penrod, S. D., & McGorty, E. K. (2004). A meta-analytic review of the effects of high stress on eyewitness memory. Law and Human Behavior, 28, 687–706. https://doi.org/10.1007/s10979-004-0565-x
Gabbert, F., Memon, A., & Allan, K. (2003). Memory conformity: Can eyewitnesses influence each other's memories for an event? Applied Cognitive Psychology, 17, 533–543. https://doi.org/10.1002/acp.885
Innocence Project. (2020). DNA exonerations in the United States (1989–2020). https://innocenceproject.org/dna-exonerations-in-the-united-states/
Innocence Project. (2026). Our impact: By the numbers. https://innocenceproject.org/exonerations-data/
Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13, 585–589. https://doi.org/10.1016/S0022-5371(74)80011-3
Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory for faces: A meta-analytic review. Psychology, Public Policy, and Law, 7, 3–35. https://doi.org/10.1037/1076-8971.7.1.3
Mickes, L., Wilson, B. M., & Wixted, J. T. (2025). The cognitive science of eyewitness memory. Trends in Cognitive Sciences, 29, 655–668. https://doi.org/10.1016/j.tics.2025.01.008
National Institute of Justice. (1999). Eyewitness evidence: A guide for law enforcement. https://nij.ojp.gov/library/publications/eyewitness-evidence-guide-law-enforcement
National Policing Institute / National Institute of Justice. (2025). Nationwide survey of policies and practices in eyewitness identification: Full report. https://www.ojp.gov/library/publications/nationwide-survey-policies-and-practices-eyewitness-identification-full-report
National Research Council. (2014). Identifying the culprit: Assessing eyewitness identification. National Academies Press. https://doi.org/10.17226/18891
Perry v. New Hampshire, 565 U.S. 228 (2012). https://www.law.cornell.edu/supremecourt/text/10-8974
Steblay, N. M. (1992). A meta-analytic review of the weapon focus effect. Law and Human Behavior, 16, 413–424. https://doi.org/10.1007/BF02352267
Tuttle, M. D., Starns, J. J., & Cohen, A. L. (2025). Protecting the innocent in eyewitness identification: An analysis of simultaneous and ranking lineups. Journal of Memory and Language, 104581. https://doi.org/10.1016/j.jml.2024.104581
U.S. Department of Justice. (2017). Justice Department announces department-wide procedures for eyewitness identification. https://www.justice.gov/archives/opa/pr/justice-department-announces-department-wide-procedures-eyewitness-identification
Wells, G. L., Kovera, M. B., Douglass, A. B., Brewer, N., Meissner, C. A., & Wixted, J. T. (2020). Policy and procedure recommendations for the collection and preservation of eyewitness identification evidence. Law and Human Behavior, 44, 3–36. https://doi.org/10.1037/lhb0000359
Wixted, J. T., & Wells, G. L. (2017). The relationship between eyewitness confidence and identification accuracy: A new synthesis. Psychological Science in the Public Interest, 18, 10–65. https://doi.org/10.1177/1529100616686966