Co-Witness Contamination: How Other Witnesses Can Change Memory

4 days ago

19 min read

Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy

Co-witness contamination occurs when information supplied by one witness influences another witness’s later recollection or account of the same event. In forensic psychology, the central question is whether a detail reported as personal observation originated in the event itself, in subsequent conversation, or in some combination of the two. Two people can give consistent statements after talking to each other even when their original experiences were different.

A witness can be sincere, attentive, and motivated to help an investigation while reporting something learned from someone else. The effect is supported by controlled research on memory conformity, but it does not mean that every conversation alters memory or that agreement between witnesses proves contamination. The evidence requires attention to timing, source, interview procedure, what was actually said, and what each person knew before the discussion. See the broader eyewitness testimony evidence for the full legal-memory framework.

What is co-witness contamination?

A co-witness is another person who observed, or believes they observed, the same incident. Co-witness contamination is the forensic evidence problem created when one person's post-event information is absorbed into another person's memory report, identification, confidence, or interpretation. Researchers often study the related construct of memory conformity: witnesses’ later reports becoming more alike following their interaction. The label contamination refers to the loss of independence between evidence sources; memory conformity refers to an observable social-memory outcome.

Consider two bystanders who watch a brief confrontation from different positions. One sees a person leave through a rear door. The other never sees the exit but hears the first person discuss it repeatedly. When later interviewed separately, both say they saw the person leave through the rear door. Their accounts now agree, but the agreement may trace back to one original observation. An investigator who counts the two statements as independent confirmation risks exaggerating the strength of the evidence.

Contamination, misinformation, and honest disagreement

The change may involve an incorrect detail, a correct detail the witness did not personally observe, or a judgment presented with increased confidence. Contamination therefore concerns evidentiary provenance as well as factual accuracy. A witness can accurately repeat information supplied by a companion while inaccurately claiming to have seen it. Conversely, witnesses may disagree because of real differences in vantage point, attention, lighting, or recollection rather than because either is deceptive.

Co-witness interaction is one route by which the misinformation effect can appear. Other routes include suggestive questioning, media reports, photographs, and investigative feedback. This article focuses specifically on information exchanged among people who witnessed or discussed the same incident; the general misinformation paradigm belongs to that separate topic.

Why eyewitnesses talk, and why it matters

Conversation after a striking event is ordinary human behavior. Witnesses seek reassurance, check whether others are safe, reconstruct a confusing sequence, and try to resolve discrepancies. Social discussion can begin at the scene, continue while waiting for police, and recur later in messages, family conversations, or online posts. None of these interactions requires a deliberate attempt to influence testimony.

In a questionnaire study of respondents who had witnessed a serious event with someone else, 86% of the follow-up participants reported discussing it with a co-witness. That figure describes the selected survey sample, not a universal rate for every crime, age group, or jurisdiction. Nevertheless, it demonstrates why investigators cannot assume that accounts collected separately were formed independently. The study was published by Paterson and Kemp (2006).

The forensic stakes are distinctive. Independent corroboration means that two accounts have meaningfully separate informational origins. When a witness learns a detail from another witness, a later matching detail cannot automatically be treated as two independent observations. Investigators must preserve the chronology of what was seen, what was heard afterward, and what was first stated before the witnesses interacted.

The landmark experimental demonstration

Gabbert, Memon, and Allan (2003) created different versions of the same filmed crime-related event. Partners viewed subtly different details, then some pairs discussed the event before completing individual memory tests; control participants did not discuss it first. Seventy-one percent of participants in the discussion condition subsequently reported at least one erroneous detail that had been available through their partner’s account.

That 71% result is a finding from the particular experimental manipulation and outcome definition. It is not the probability that a real witness will be contaminated, nor a reliability score for a witness who talked to someone else. The design was important because the experimenters knew exactly which details each participant had or had not encountered in the original video. Real investigations rarely have access to such clean ground truth.

Later experimental work has reproduced the basic possibility of information transfer during co-witness exchange, while showing that the discussion’s conversational structure matters. Gabbert, Memon, and Wright (2006) analyzed how influence arises during discussion rather than treating conversation as an undifferentiated exposure. Such studies support the existence of an effect, but they also show that individual behavior and social context matter.

How one witness changes another witness’s account

1. A new detail enters the shared narrative

A participant says that a car was dark blue, identifies a direction of travel, or describes an object that the other person missed. The second witness has now encountered a post-event description that may be more vivid or coherent than their incomplete original perception. If the new detail is repeated, it can become increasingly accessible during later recall. Repetition and familiarity are relevant, although their effects depend on how the information is presented and tested.

2. The listener decides whether to rely on the other person

Co-witnesses have unequal information, confidence, and social standing. A person who seems to have watched longer, stood closer, or paid more attention may be judged as having superior memory. An experiment manipulating beliefs about viewing duration found that perceived access to the event helped shape conformity even when actual exposure duration was identical; see Gabbert, Memon, and Wright (2007).

A seemingly authoritative assertion can affect what a witness chooses to report before it alters the person's own sense of remembering. Agreement in an interview therefore may reflect strategic deference, uncertainty, confusion about source, or memory reconstruction. A single later statement cannot reliably distinguish these processes.

3. Information from conversation becomes hard to source

Human remembering includes judgments about where knowledge originated. Did the person see the detail, infer it, hear it from a friend, or encounter it in a report? The source-monitoring framework explains how remembered content and its origin can become dissociated. An image of an object discussed after the event can later feel like a visual memory from the event itself, particularly when its source is not explicitly considered.

This process should be distinguished from lying. A person may sincerely experience the borrowed detail as familiar, even vividly so. The responsible forensic question is whether the detail can be traced to an independently documented original observation. Psychological theory offers plausible mechanisms; it does not allow an expert to inspect a statement and conclusively identify the origin of every remembered element.

4. Confidence and reporting precision shift

A confident partner may prompt another witness to adopt both the partner's content and apparent certainty. In a controlled study, Goodwin and colleagues (2017) found that highly confident co-witnesses could increase incorporation of accurate or misleading information into collaborative and subsequent individual reports. This matters because jurors may interpret confidence as a cue to evidential strength, even when that confidence developed after feedback.

Social exchange can also change how much detail a witness volunteers. Rechdan and colleagues (2018) found that disconfirming feedback reduced the precision of some later reports without producing a corresponding decrease in confidence. Memory content, level of detail, and expressed confidence must therefore be assessed as separate outcomes.

The main mechanisms: social influence and memory processes

Normative and informational influence

Normative influence refers to alignment with another person's position because agreement feels socially appropriate or conflict feels undesirable. Informational influence refers to relying on the other person because they appear more accurate or better informed. In co-witness conversations these processes can coexist. One witness may yield because a partner speaks with certainty, because the partner is a close friend, or because the witness doubts their own vantage point.

Neither mechanism requires coercion. Research on source competence indicates that perceived ability and authority can moderate the uptake of a co-witness's claims; a study by Mojtahedi, Ioannou, and Hammond (2020) provides a controlled example. The conclusions apply to patterns within studied conditions, not to universal traits of people who are more susceptible.

Retrieval dynamics and collaborative remembering

Speaking with someone else changes the retrieval environment. One person's prompts can cue useful details, redirect attention to particular aspects, or interrupt another person's line of recall. This is why collaboration need not produce the same outcome for correct details, errors, total quantity, and later independent performance. The forensic review by Hope (2019) surveys the costs and conditions of conversational remembering in applied contexts.

Experimental findings also complicate the slogan that all discussion makes eyewitness memory worse. In Vredeveldt and colleagues’ study, collaborative pairs produced fewer errors than nominal pairs in one test, with evidence of both retrieval disruption and later cross-cuing. The prudent inference is that the effects are outcome-specific. The ability of discussion to elicit correct information does not remove the legal need to distinguish independently observed details from information learned from others.

What recent studies add

Remote discussions can contaminate memory

A 2024 experiment by Kękuś, Szpitalak, Polczyk, and Barzykowski used two differing film versions followed by video-call discussions and individual recognition tests. Seventy-two adults participated in 36 pairs. Accuracy was lower for discussed misinformation-related details than for comparable non-discussed details. When one participant challenged a misleading detail, subsequent answers were more often correct than when the misinformation passed unchallenged.

The study demonstrates that conversation-based influence can occur without witnesses standing together at a physical scene. Its sample size and controlled film materials also matter: the findings support the mechanism under tested online conditions, not precise prevalence estimates for real investigations involving social media, encrypted group chats, or weeks of repeated contact.

Repetition is not the same as having multiple independent sources

O’Donnell, Chan, Foster, and Garry (2023) reported experiments and meta-analytic evidence that repeating misinformation impaired accuracy, while varying the claimed source did not independently increase suggestibility beyond repetition under their tested conditions. That result cautions against assuming that several supposed sources necessarily exert a distinct additive effect. The relevant exposure history includes how often a claim was encountered, not merely how many people appeared to endorse it.

Social media introduces another route

Witnesses may view clips, read other users' accounts, and then discuss those claims with fellow witnesses. Sharman, Danby, and Gray (2024) examined misleading post-event information in a social-media-style video and different warning approaches. This literature is related to co-witness contamination because online reports can seed later conversations, but a mass-media post and an in-person partner are different information sources and should be documented separately.

Who is more likely to be influenced?

Susceptibility is a property of interactions and conditions, not a forensic diagnosis or permanent personality label. Studies examine relative memory confidence, belief about who saw more, relationships between witnesses, information quality, whether misleading details are challenged, and the way recall is tested. These variables can help investigators identify potential routes of influence; they do not yield a validated instrument for deciding that a particular person's testimony is false.

Friends, partners, strangers, and perceived authority

A close companion may be trusted more readily than a stranger. In a study of co-witness relationships, participants who knew their partners before the experiment were more likely to adopt information they had not observed themselves; the full study report also examined mistaken attributions of guilt and innocence. Relationship effects depend on the sample, the social situation, and the kind of information exchanged. Investigators should document the relationship rather than assume friends always conform.

Confidence must likewise be interpreted in context. A confident witness may be accurate, mistaken, or repeating someone else. The key distinction is between confidence documented at an early, uncontaminated stage and confidence recorded only after discussions, identifications, or external feedback. The current eyewitness confidence and accuracy research addresses that broader evidence separately.

Age, culture, language, and vulnerability

Research cannot support a universal hierarchy ranking demographic groups by contamination risk. Effects observed among university students may not generalize straightforwardly to children, older witnesses, people using interpreters, or individuals responding to high-stakes interviews. Culture can shape conventions of agreement, disagreement, deference, and conversational turn-taking, but an individual's susceptibility should not be inferred from group identity. Language barriers can also make it harder to distinguish direct observation from a translated or paraphrased account.

Where developmental, communication, disability, or acute stress issues matter, interviewers should arrange appropriate accommodations and rely on population-specific evidence. Such considerations are reasons to improve interviewing conditions, not reasons to discount testimony categorically. The status of memory content must be established from procedures and records, not stereotypes.

Can witnesses help one another remember accurately?

Yes, under some conditions co-witness exchange can bring out correct details or reduce particular error types. A participant might remind another of the order of two events or raise a question that leads the second person to retrieve previously inaccessible information. Research on collaborative remembering shows both potential benefits and costs, depending on how accuracy is measured, when collaboration occurs, and how the witnesses speak to each other. Vredeveldt and colleagues provide an especially relevant counterexample to absolute claims about harm.

The investigative principle remains to record an unprompted individual account first, whenever feasible, and preserve any later additions as later additions. Information emerging only after collaboration may still be investigatively useful, especially if independently corroborated by recordings, physical traces, or another source. Its origin should remain visible in the evidentiary record.

How contamination changes the meaning of corroboration

Imagine four people observe an altercation. Two separately see a green backpack; a third hears one of them mention it; a fourth reads the group conversation. If all four later describe the backpack, the four statements do not necessarily provide four independent observations. Their accounts share informational ancestry. An appropriate analysis distinguishes the two possible original observations from the two subsequent transmissions.

This is an evidence-dependence problem. Multiple reports can look mutually reinforcing when they depend on a common source. The same issue affects identification decisions: after one witness says a particular person resembles the perpetrator, a second witness’s later selection may reflect both their own memory and the first witness’s suggestion. That possibility does not determine whether the suspect is innocent or guilty, and it does not itself satisfy or refute a legal identification standard.

Courts and investigators must also separate psychological observations from legal findings. A psychologist can discuss known memory mechanisms and conditions that increase contamination risk. The court determines admissibility, credibility, relevance, and the legal significance of those facts under the governing jurisdiction's rules. No research effect size settles whether any specific testimony is admissible or whether a defendant committed an offense.

What investigators should do at the scene

Separate witnesses when operationally possible

When safety and practical circumstances permit, obtain individual accounts without opportunities for witnesses to coordinate details. The National Academies' 2014 eyewitness report recommends that officers manage multiple-witness scenes so interactions are minimized and personnel avoid leaking information. Separation is a procedural safeguard aimed at preserving informational independence; an inability to separate every witness is not, by itself, proof that contamination occurred.

Record each person's first account in their own words

A prompt, open-ended statement establishes a reference point for later changes. An interviewer can ask what the witness remembers, what the witness personally perceived, and what has happened since the incident without implying an expected answer. Audio or video recording, time stamps, contemporaneous notes, and verbatim language help later reviewers distinguish the first report from subsequent elaboration. Exact practice depends on applicable law, local policy, accessibility needs, and the demands of immediate safety.

Ask about contact and sources of information

A useful factual inquiry asks whether the person has spoken with other witnesses, viewed a video, read a report, received photographs, listened to media coverage, or exchanged messages. Where appropriate, record who supplied a specific detail and when. Interviewers should avoid accusatory language: a witness who admits discussing an event can still have many independently remembered facts. Documentation is intended to make the provenance of information assessable.

Preserve original and revised versions

If a witness supplies an altered description, do not silently overwrite the earlier account. Preserve both, with their dates, wording, and known intervening influences. That chronology can later distinguish an ordinary correction from a detail first introduced during a conversation. Investigative corroboration should seek independently obtained evidence rather than only additional people repeating the same conversation.

How interviews can reduce additional contamination

Individual, non-leading interviewing protects the quality of subsequent evidence. Interviewers should let witnesses provide uninterrupted narrative recall before using targeted prompts, explicitly allow uncertainty, and avoid telling a witness what others said as though it were an established fact. Explaining that different people notice different things can reduce pressure to force consistency. Interview records should identify which information came from the witness and which was introduced by the interviewer.

The National Academies review also emphasizes training officers in memory limitations, preventing inadvertent contamination, standardized identification instructions, and prompt documentation of confidence in identification procedures. These recommendations address a larger investigative system. Co-witness contamination is one reason those safeguards matter, not a substitute for the complete set of lineup and interview protections.

What if witnesses have already talked?

Record the fact and approximate timing of contact, identify participants, ask each witness independently for their own observations, and document any details that they attribute to other sources. Existing messages or earlier recorded accounts can be relevant if lawfully available. Asking witnesses to reconstruct precisely which memories were present before discussion may itself be difficult; a retrospective claim of original knowledge should be evaluated alongside contemporary records.

An investigation need not discard all evidence from witnesses who talked. The more defensible approach is to assess which details have independent support, which details appeared only after interaction, and which remain uncertain. The question is the weight and origin of information, not a global label applied to the entire witness.

What forensic psychologists can and cannot conclude

A forensic expert may explain memory conformity, identify conditions consistent with social influence, review the timeline of statements, and describe limits of published evidence. If the record shows that a witness first used a term after reading another person's message, that can be a documented factual sequence. A psychological account of possible source confusion is an interpretation that must be distinguished from the documentary fact.

The expert ordinarily cannot calculate an individualized probability that a particular witness's memory was contaminated from a generic laboratory percentage. There is no accepted co-witness contamination score that determines whether someone is truthful. An association between discussion and an error in a group study is not a case-specific finding of fabrication, false identification, or criminal guilt.

Expert opinions also should avoid treating confidence as a fixed personal trait. The chronology of confidence statements, feedback, discussion, and identification procedure may matter more than how emphatically someone speaks at trial. For the broader identification question, see eyewitness identification practices.

Laboratory evidence, field observations, and external validity

The experimental strength of co-witness studies comes from controlling what each person originally saw and precisely identifying the misleading detail. The same control limits direct transfer to real crimes. Actual incidents may be more emotionally salient, more prolonged, more ambiguous, and associated with personal relationships or serious consequences. Witnesses may discuss events repeatedly, be exposed to body-camera or surveillance footage, or hear conflicting accounts over days and months.

Field surveys help show that witness discussion is common, but their retrospective questionnaires can introduce selection and recall biases. Laboratory experiments often measure recognition or free recall of particular film details rather than a complete legal statement. Review evidence accordingly: laboratory replication establishes the mechanism, whereas inference about a given case requires its own records and contextual facts.

Recent synthesis also emphasizes that eyewitness evidence has different probative value at different stages. Mickes, Wilson, and Wixted (2025) argue that an early test of relatively uncontaminated memory can provide valuable information, including evidence favorable to an innocent suspect. The proper conclusion is to protect initial memory evidence and examine later influences, rather than characterize all eyewitness testimony as inherently worthless.

Reliability, validity, and the problem of numerical certainty

The co-witness literature supports group-level conclusions about the risk of information transfer. It does not supply a universal sensitivity or specificity for identifying contaminated witnesses in practice. Some laboratory outcomes are detail-specific and depend on whether the test asks for free recall, forced-choice recognition, confidence, or attribution of source. Studies also differ in delay, sample size, exposure, and the authority of the misleading speaker.

A false positive would arise if an assessor incorrectly labeled a genuinely independent detail as borrowed; a false negative would arise if an actually borrowed detail were treated as direct observation. Without a validated case-level classification model and an appropriate base rate, percentages from one experiment cannot be translated into calibrated personal risk predictions. The appropriate forensic response is transparency about evidence quality and unresolved uncertainty.

Legal consequences and jurisdictional differences

United States evidence rules, state procedures, common-law witness-sequestration practices, disclosure obligations, and admissibility standards vary by setting. Other countries have different investigative and trial frameworks. A procedural recommendation to interview witnesses separately should therefore be presented as evidence-informed practice, not a universally controlling legal rule. Whether conversation affects admissibility, impeachment, expert testimony, or weight is a question for the relevant jurisdiction.

Defense counsel and prosecutors may both have legitimate reasons to examine co-witness interaction. One party might challenge purported independent corroboration; the other may seek independent physical evidence supporting the witness's original perception. The scientific task is to describe the psychological possibilities fairly and the evidentiary timeline accurately. The legal task is to decide the consequence under applicable law.

Common myths and misconceptions

Myth: If two witnesses agree, both independently remember it

Agreement can arise from common observation, shared inference, shared misinformation, or post-event conversation. The number of accounts and the number of independent sources are different quantities. A careful investigation documents whether testimony converged before or after contact.

Myth: Witnesses who change their stories must be lying

Changes may reflect memory updating, new cues, fatigue, improved retrieval, interviewer pressure, or contact with others. Some changes are errors and some are legitimate corrections. A documented change warrants explanation and evidence review; it does not establish intent to deceive.

Myth: One conversation always destroys eyewitness memory

Witnesses can retain reliable firsthand information after talking with others, and conversation sometimes improves retrieval of particular correct facts. The effect depends on the detail, exposure, conditions, and test. Broad labels such as fully contaminated or completely uncontaminated can conceal important variation within the same statement.

Myth: A witness’s confidence reveals whether information was borrowed

Confidence is influenced by initial memory quality, procedures, post-event information, and social feedback. Strong confidence at trial cannot independently establish the source of a detail. Earlier confidence recorded before feedback can be more informative in specified identification conditions, but it still requires careful interpretation.

Myth: Digital contact is harmless because witnesses were not together

The 2024 online co-witness experiment demonstrates that remote conversation can transmit misleading details. Messages, video calls, public clips, and group chats may all matter. Their effects are not identical, so investigators should document each source rather than collapse every digital interaction into one category.

Practical guidance for witnesses

If you witnessed an incident, record what you personally recall as soon as doing so is safe and appropriate, and distinguish what you observed from information you later received. Tell the interviewer if you spoke with another witness or saw media coverage. If you are unsure about a detail, say so rather than filling a gap from someone else's account. If you later realize that a reported fact came from another source, clarify that chronology through the proper investigative channel.

If you are an investigator or practitioner, treat the witness as a source of observations with a traceable information history, not as a memory test score. Keep the first account, document exposures, preserve revisions, and test claims against independent evidence. If you are involved in a live legal case, qualified local counsel can explain the jurisdiction-specific procedural implications.

Frequently asked questions

Can talking to another eyewitness create a false memory?

Discussion can lead a person to report a detail that was not present in their original view, sometimes with a subjective impression of remembering it. Experiments demonstrate this possibility. Whether the person's internal experience changed or they adopted information at the reporting stage can be difficult to establish in a particular case.

Is co-witness contamination the same as the misinformation effect?

Co-witness contamination is a specific route of post-event influence through another witness. The misinformation effect is the broader research phenomenon in which misleading information presented after an event impairs performance on a later memory test. Co-witness research adds social interaction, relationship, conversational sequence, and evidentiary dependence.

What is memory conformity?

Memory conformity is the tendency for people's later accounts of an event to become more similar after social exchange. It can occur for details that one person never originally encountered. It describes an outcome observed in groups; it is not a clinical condition.

How can investigators tell if witnesses influenced each other?

They can compare the earliest independently recorded accounts with later versions, identify contact between witnesses, review available messages, and trace when a disputed detail first appeared. These methods can reveal opportunities and sequences of influence, but they may leave substantial uncertainty about the origin of individual memories.

Does contamination mean the witness is unreliable about everything?

No. A witness may remember many core elements correctly while borrowing one detail. Reliability and source attribution should be evaluated at the level of individual claims when evidence allows, rather than converted into a blanket personal judgment.

Can social media contaminate eyewitness evidence?

Yes. Online misinformation and co-witness discussion can be connected; experimental work demonstrates that remote interpersonal discussions can change later answers. The precise contribution of social media content, peer messages, and direct memory must be documented in the individual case.

Are children more susceptible to co-witness contamination?

Developmental effects depend on the question, interviewing environment, age, and type of suggestion. Findings from adult dyads cannot simply be imposed on children. Child interviewing requires its own specialized evidence and safeguards.

What if witnesses independently agree before speaking?

Early agreement documented before any contact offers different evidence than later convergence after discussion. Even early agreement does not eliminate ordinary shared misperception or other memory errors. It does reduce one potential explanation for their agreement: post-event communication between those witnesses.

Can courts exclude testimony because witnesses discussed the case?

Consequences depend on local evidence rules, procedural history, the nature of the contact, the relevance of the information, and judicial decisions. Psychology describes how influence may happen; the applicable law determines admissibility and other legal consequences.

Is there a test that proves a co-witness implanted a memory?

There is no general-purpose, infallible clinical or forensic instrument that establishes such an event from a person's present-day account. Documentary chronology, original recordings, information provenance, and independently verified facts are usually more defensible than a stand-alone psychological score.

References

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.

Gabbert, F., Memon, A., & Wright, D. B. (2006). Memory conformity: Disentangling the steps toward influence during a discussion. Psychonomic Bulletin & Review, 13, 480–485. https://doi.org/10.3758/BF03193873.

Gabbert, F., Memon, A., & Wright, D. B. (2007). I saw it for longer than you: The relationship between perceived encoding duration and memory conformity. Acta Psychologica, 124, 319–331. https://doi.org/10.1016/j.actpsy.2006.03.009.

Goodwin, K. A., et al. (2017). The confident co-witness: The effects of misinformation on memory after collaborative discussion. Applied Cognitive Psychology. https://doi.org/10.1002/acp.3320.

Hope, L. (2019). Memory at the sharp end: The costs of remembering with others in forensic contexts. Topics in Cognitive Science. https://doi.org/10.1111/tops.12357.

Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3–28. https://doi.org/10.1037/0033-2909.114.1.3.

Kękuś, M., Szpitalak, M., Polczyk, R., & Barzykowski, K. (2024). Online misinformation can distort witnesses’ memories: Analysis of co-witness discussions using an online version of the MORI-v technique. Frontiers in Psychology, 14, 1239139. https://doi.org/10.3389/fpsyg.2023.1239139.

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.

Mojtahedi, D., Ioannou, M., & Hammond, L. (2020). Intelligence, authority and blame conformity: Co-witness influence is moderated by the perceived competence of the information source. Journal of Police and Criminal Psychology, 35, 422–431. https://doi.org/10.1007/s11896-019-09361-2.

National Research Council. (2014). Identifying the culprit: Assessing eyewitness identification. National Academies Press. https://doi.org/10.17226/18891.

O'Donnell, R., Chan, J. C. K., Foster, J. L., & Garry, M. (2023). Experimental and meta-analytic evidence that source variability of misinformation does not increase eyewitness suggestibility independently of repetition of misinformation. Frontiers in Psychology, 14, 1201674. https://doi.org/10.3389/fpsyg.2023.1201674.

Paterson, H. M., & Kemp, R. I. (2006). Co-witnesses talk: A survey of eyewitness discussion. Psychology, Crime & Law, 12, 181–191. https://doi.org/10.1080/10683160512331316334.

Rechdan, J., et al. (2018). The effects of co-witness discussion on confidence and precision in eyewitness memory reports. Memory, 26, 904–912. https://doi.org/10.1080/09658211.2018.1448872.

Sharman, S. J., Danby, M. C., & Gray, A. D. (2024). Witnesses' susceptibility to misleading post-event information delivered in a social media-style video. Memory, 32, 100–110. https://doi.org/10.1080/09658211.2023.2294692.

Vredeveldt, A., et al. (2017). When discussion between eyewitnesses helps memory. Legal and Criminological Psychology. https://doi.org/10.1111/lcrp.12097.

Psychological Encyclopedia