RESEARCH DOSSIER
Suspect Interrogation and False Confessions
Abstract. Suspect interrogation is a consequential form of institutional information gathering in which authority, evidence claims, perceived incentives, memory, and individual vulnerability can influence admissions. This research monograph examines how false confessions arise, why innocence does not guarantee resistance, how interrogation methods differ in diagnosticity, and how confessional statements affect subsequent investigation and adjudication. Its critical integrative method compares original experiments on false evidence, minimization, investigator bias, and developmental suggestibility with laboratory meta-analyses, DNA-exoneration case analyses, juror experiments, and international interviewing principles. The literature demonstrates multiple pathways to false admission and reveals the danger of measuring investigative success by confession frequency alone. A 2024 systematic review found greater false-confession odds under accusatorial than information-gathering procedures in experimental simulations while identifying uncertainties about absolute real-world incidence and particular tactics. The synthesis develops an evidence-based account of cognitive and institutional mechanisms, provenance preservation, protections for vulnerable suspects, recording and independent corroboration, and questions for future human–artificial investigative workflows. The principal conclusion is that investigative reliability depends on discriminating true from false information and documenting the conditions under which each statement arose.
Keywords: suspect interrogation; false confessions; coerced-compliant confession; coerced-internalized confession; interrogative suggestibility; investigator bias; minimization; false evidence; information-gathering interviewing; confession diagnosticity; wrongful convictions; evidence contamination; forensic psychology; Artificial Era
Suspect Interrogation and False Confessions as a Psychological Research Object
Suspect interrogation is an institutionally organized interaction in which an investigator seeks information from a person suspected of wrongdoing. Its psychological significance arises from asymmetries of authority, knowledge, control over the setting, and consequences of what is said. A confession is a statement that acknowledges responsibility or involvement; a false confession is a materially incriminating statement by someone who did not commit the alleged act. The research problem concerns the conditions under which a person's decision to speak, assent, narrate, or sign departs from the historical truth of the event. A confession can accurately report genuine involvement, reproduce an investigator's erroneous theory, combine genuine knowledge with suggestion, or express a decision to obtain immediate relief. These pathways demand different explanatory models and different evidentiary tests. The scientific object encompasses the interrogation process, the suspect's cognition and decision-making, the content and provenance of admissions, and the subsequent circulation of the confession as apparently independent evidence. Kassin et al. (2010) Leo (2009) Kassin et al. (2025)
The central outcome is the discriminative capacity of the interviewing procedure: whether it yields information that helps distinguish persons who engaged in the alleged conduct from those who did not. Raw confession counts are an incomplete measure of success because a procedure can increase statements from guilty and innocent people together. A reliable investigation also asks whether a narrative contains independently verifiable details, whether those details were supplied before questioning or introduced by investigators, and whether alternative accounts were objectively tested. The truth status of a confession cannot be inferred merely from its length, emotional vividness, or apparent certainty. An inquiry into false confessions therefore connects experimental psychology, legal psychology, memory research, developmental science, psychometrics, investigative practice, and the sociology of criminal procedure. Its practical importance extends to real victims and communities, whose interest in identifying the actual perpetrator is undermined when investigators build an erroneous case around an unreliable confession. Russano et al. (2005) Catlin et al. (2024)
This monograph applies a critical integrative review of original experiments, controlled mock-crime paradigms, observational police studies, archival analyses of documented exonerations, systematic reviews, and professional interviewing principles. Evidence is evaluated according to the design that produced it. Laboratory studies with established ground truth can estimate causal effects under controlled conditions; archival exonerations establish that severe errors occurred and reveal patterns of propagation but cannot estimate the population prevalence of false confession; self-reported practice surveys describe respondents' beliefs and behavior, not independently audited rates of police conduct. Meta-analyses increase precision by combining eligible studies while inheriting limitations in their scope and measurement. The synthesis examines mechanisms and policy implications without presenting itself as a new systematic review or original participant experiment. The public claim that a measure is protective is assessed separately from the empirical question of how much protection it affords. Kassin et al. (2007) Catlin et al. (2024) Stewart et al. (2018)
Historical Foundations of Confession Research
Confession has long occupied a privileged place in legal fact-finding because a statement against one's own interest appears to carry a presumption of authenticity. Historical legal systems also reveal the danger of treating this presumption as conclusive. The evolution from overt physical coercion to psychologically structured interrogation did not eliminate the scientific problem of elicited compliance: control of time, uncertainty, threats, promised relief, isolation, authority, and the gradual narrowing of acceptable answers can influence testimony even in the absence of physical violence. Modern psychological research developed explanatory tools for studying suggestibility, conformity, social influence, memory reconstruction, and decision-making under situational pressure. Those tools made the confession a researchable product of an interaction rather than a transparent indicator of internal guilt. The international literature includes North American experimentation alongside British and Icelandic traditions of interrogative suggestibility and investigative interviewing, and later transnational guidance on effective interviewing. Kassin et al. (2010) Kassin (2008) Steering Committee of Experts (2021)
A decisive experimental transition occurred when investigators learned to arrange accusations against participants whose innocence was established by experimental design. In the computer-crash experiment, Kassin & Kiechel (1996) falsely accused participants of triggering a computer failure and manipulated whether another person appeared to confirm the accusation. The researchers distinguished signing an admission from subsequently believing in personal culpability and offering details consistent with the supposed event. Its importance lies in demonstrating that false incrimination can affect behavior, belief, and reported memory at distinct levels. Because the experimental accusation concerned a minor laboratory incident, the result establishes a mechanism under controlled conditions rather than reproducing the duration, penalties, and power relations of a criminal interrogation. The study's methodological legacy includes the necessity of distinguishing compliance from internalization and of avoiding automatic generalization from laboratory event scenarios to serious criminal cases.
The next experimental step was to introduce both genuinely responsible and innocent participants. In the cheating paradigm of Russano et al. (2005), some participants violated an experimental rule while others did not. The researchers manipulated minimization and an explicit offer of leniency and observed increases in both true and false admissions. This design improved upon innocence-only paradigms because it permitted a comparison of the procedure's ability to separate the two groups. Subsequent studies examined false-evidence claims, ostensibly pending forensic tests, investigator expectations, and the perceived consequences of confession. Perillo & Kassin (2011) demonstrated false-confession effects of bluffing across three experiments; Narchet et al. (2011) modeled how investigator beliefs can influence the tactics deployed. The evolution of experimental models thus moved the field from asking whether an innocent person can confess to identifying the causal paths by which a method changes the distribution of truthful and untruthful statements.
Scientific reviews and evidence synthesis mark another institutional phase. The 2010 scientific review by Kassin et al. (2010) consolidated suspect-level risks, interrogation practices, safeguards, and consequences for courts. Meissner et al. (2014) compared accusatorial and information-gathering approaches, noting advantages of the latter with limits imposed by the number of independent studies. The 2024 Campbell update Catlin et al. (2024) expanded the evidence base and used network meta-analysis to compare interview approaches in controlled studies, while the 2025 review Kassin et al. (2025) integrated newer findings about cumulative disadvantage, evidence contamination, guilty pleas, and protections for vulnerable people. The 2021 Principles on Effective Interviewing Steering Committee of Experts (2021) show how scientific findings have influenced international professional and human-rights frameworks. This historical sequence reveals growing emphasis on the reliability of obtained information, the independence of corroboration, and testable safeguards.
Conceptual Taxonomy and the Phenomenology of Admission
An accurate taxonomy starts by separating factual truth from voluntariness, legal admissibility, psychological belief, and self-reported motive. A person can knowingly make a false admission to secure an immediate benefit; a person can come to believe an inaccurate accusation; and a person may repeat a jointly constructed narrative without settled personal belief. The classic classification distinguishes voluntary false confessions, occurring without direct external interrogation pressure, from coerced-compliant and coerced-internalized confessions. A voluntary false confession can be associated with motives such as protecting another person, seeking attention, or responding to a personal crisis. A coerced-compliant confession involves deliberate submission despite the person's private conviction of innocence, often to stop distress or avoid an anticipated cost. A coerced-internalized or persuaded confession involves a change in belief about one's own involvement. These categories describe pathways; an individual case may contain transitions and overlapping motives. Kassin & Kiechel (1996) Leo (2009) Kassin (2008)
Compliance and internalization should be operationalized rather than inferred from the final transcript. A signed statement records an overt act; it does not uniquely disclose an enduring belief. Questionnaires or contemporaneous interviews may reveal whether participants attributed the alleged act to themselves, believed evidence against them, or simply anticipated that agreeing would end the encounter. Memory confabulation demands further distinction: generated details may reflect suggestion, imagination, self-explanation, or an attempt to make an accusation coherent. Researchers should trace each detail's entry into the interview record and examine whether it appears in an initial free account before exposure to the relevant fact. Documentary records with gaps cannot always resolve the confessor's internal state. The typology remains useful because it directs different lines of psychological and evidentiary analysis rather than offering a diagnostic checklist. Kassin & Kiechel (1996) Perillo & Kassin (2011)
The phenomenology of innocence adds a paradoxical decision mechanism. A person who knows they did nothing wrong may treat cooperation and waiver of rights as low-risk, expecting the truth to become obvious. Kassin & Norwick (2004) experimentally manipulated the guilt status of mock-theft participants and the demeanor of a purported detective; innocent participants were more likely to waive the mock Miranda rights than guilty participants. Their choices were made in a simulated setting with a modest stake, yet they challenge the assumption that knowledge of innocence automatically promotes caution. The same expectation can make a suggestion of future testing sound reassuring. In Perillo & Kassin (2011), innocent people sometimes understood a bluff about forthcoming evidence as a promise that the evidence would eventually clear them and therefore confessed in the meantime. The mechanism is anticipatory reasoning about future vindication under distorted immediate incentives.
A confession is also a communicative act under institutional surveillance. The suspect is often asked to give an account acceptable to a person who controls the encounter, judges denials, and may repeatedly present a hypothesis of guilt. Participants must reason about what the investigator knows, what the investigator believes, which responses will prolong interaction, and what refusing a requested formulation may cost. An admission can consequently serve several immediate goals at once: resolving disagreement, ending fatigue, protecting relationships, communicating cooperation, or obtaining a shorter path through an apparently inescapable procedure. The scientific challenge is to reconstruct these motives from observable sequences without attributing a single universal psychological cause. Kassin (2005) Scherr et al. (2020)
Diagnosticity, Base Rates, and the Logic of Confession Evidence
Confession diagnosticity concerns how differently a procedure treats genuinely guilty and innocent persons. In experimental settings with known ground truth, a useful descriptive ratio compares the probability of confessing among guilty participants with the probability of confessing among innocent participants exposed to the same method. This is distinct from the percentage of suspects who confess. A tactic may raise the numerator and denominator together and thereby leave the investigator with more admissions but less information about who is responsible. In Russano et al. (2005), strategies implying a favorable consequence increased both kinds of confession, a warning that apparent productivity does not guarantee discrimination. The same logic applies to questions that guide suspects to accept salient pieces of a proposed narrative. A method earns scientific confidence by producing verifiable information and reducing errors under relevant conditions, not by maximizing an isolated operational count.
The evidentiary meaning of a confession also depends on the prior probability of guilt and on other available information. Even a procedure with a favorable ratio can produce a substantial proportion of false admissions when many people subjected to the procedure are innocent. In statistical terms, the positive predictive value of an admission depends on both sensitivity to genuine involvement and the false-positive rate among innocent suspects. Neither laboratory confession percentages nor aggregate exoneration cases directly reveal the base rate of innocence in a particular policing population. Consequently, precise real-world predictive values cannot be responsibly calculated from published laboratory experiments alone. The research challenge is to study known-ground-truth simulations, corroborated field cases, and recruitment or screening decisions together, while tracking different selection mechanisms. Catlin et al. (2024) Kassin et al. (2007)
Diagnosticity must also be considered across stages. Screening creates the pool of suspects; an interrogator may amplify an initial mistaken belief; confirmation-seeking questions shape what is said; and a resulting statement may affect how later evidence is collected. These sequential decisions can produce cumulative error even if each stage appears locally reasonable. Meissner & Kassin (2002) reported that investigator training and experience influenced the tendency to label targets deceptive without improving discrimination accuracy in their study. Narchet et al. (2011) then showed how experimentally manipulated guilt expectations affected use of minimization tactics and increased false admissions among innocent participants. Together these findings supply a plausible chain linking misclassification, treatment, response, and renewed confidence in guilt. The chain does not demonstrate that every biased investigation leads to confession, and different stages may be interrupted by independent checks.
The evidentiary properties of narrative details differ from the epistemic force of a generic admission. A suspect's spontaneous mention of a verifiable, undisclosed feature of an event may carry more diagnostic information than a statement drafted after investigators have repeatedly revealed or suggested that feature. Even a seemingly accurate detail must be examined for alternative sources: news coverage, leaks, co-suspect information, prior questioning, and interviewer prompts. A confession that emerges after a long interaction can contain true crime facts and nonetheless be false as to personal responsibility. The key empirical principle is to preserve the provenance of information and test its independence. This principle connects interrogation psychology with source monitoring, evidentiary dependence, and forensic contamination research. Kassin et al. (2012) Kassin et al. (2025)
Legal and Institutional Context of Psychological Interrogation
Law determines what police may do and which statements can be used, while psychological research determines how interaction conditions influence human behavior. These domains inform one another but answer distinct questions. A statement's technical admissibility under one jurisdiction's rules does not establish its factual reliability; an experimental demonstration of a risk mechanism does not automatically decide a legal dispute about a particular interview. The United States Miranda system provides one influential example of a rights-warning and waiver framework, and other jurisdictions adopt different safeguards, exclusion rules, recording duties, and approaches to counsel. The monograph uses such legal examples to illustrate psychological mechanisms rather than assuming a universal national law. International guidance such as Steering Committee of Experts (2021) joins accuracy, dignity, and legal safeguards within a broader investigative standard.
An institutional focus clarifies the role of professional incentives. Investigators may be assessed by clearance rates or the progress of cases, supervisors may favor decisive narratives, and organizational culture may encourage early certainty. These pressures can interact with the authority hierarchy of the interview room. Affective commitment to a hypothesis is especially important when a suspect's denial is treated as additional proof of deception. Researchers should distinguish the investigator's sincerely held belief in guilt from a knowingly deceptive evidentiary claim, because each can shape questioning differently. Survey evidence Kassin et al. (2007) documents investigators' beliefs and self-reported practices, but self-report cannot independently establish the prevalence of tactics or misconduct. Observational research and recorded interviews are required to study real-time practice.
The boundary of the research object is intentionally maintained. General eyewitness identification belongs primarily to memory and identification research, forensic profiling to behavioral inference, and adjudicative decision-making to the psychology of legal judgments. This dossier investigates those adjacent topics only where they enter the causal sequence of suspect interrogation or determine the evaluation of confession evidence. It therefore complements the broader Science monograph Kassin et al. (2025) informs without duplicating independent forensic subjects. The research program extends beyond one policing system because persons under suspicion experience familiar psychological processes under differing institutional procedures; comparative studies must nonetheless preserve local legal and cultural differences when interpreting results.
Accusatorial Interrogation and the Management of Perceived Evidence
An accusatorial interrogation organizes the exchange around a presumption of culpability. The investigator may present incriminating material, challenge explanations as deceptive, repeatedly assert that an admission is inevitable, and frame the suspect's remaining choice as the manner of accepting responsibility. Such practices create a social environment in which denials have limited communicative effect. A suspect who perceives the investigator as convinced may reason that additional truthful protest will only prolong the encounter. Accusation, evidence presentation, and coercion are analytically distinct: an investigator can fairly disclose accurate evidence, or can misrepresent its existence, provenance, or strength. The psychological effects depend on what information is presented, how it is framed, the alternatives offered, and the suspect's capacity to evaluate the claim. A general label such as 'confrontation' is therefore insufficient for causal inference. Kassin et al. (2010) Meissner et al. (2014)
The presentation of fabricated or misleading evidence can convert external authority into apparent evidence about one's own memory. In Kassin & Kiechel (1996), the claim that a witness observed the participant's purported error increased admissions and reported internalization. Stewart et al. (2018) reviewed experimental false-confession simulations and identified false-evidence ploys as a risk factor across studies, while emphasizing differences among experimental paradigms. The mechanism can include social pressure, discounting of one's memory, doubts about what occurred, or calculation that resistance is futile. Investigators sometimes argue that a false-evidence statement can be justified as a test of reactions; its evidentiary danger is that the assertion becomes an input to the suspect's choices and subsequent story. The experimental record supports caution about fabricated evidence precisely because the procedure can produce the phenomenon it is ostensibly trying to detect.
A bluff is distinguishable from a direct false claim about a conclusive result. The interviewer may suggest that a recording, laboratory test, or other information will soon be available without stating that the result already identifies the suspect. In the series studied by Perillo & Kassin (2011), such bluffs increased false admissions. Participants' reports suggested that some innocent people anticipated their ultimate exoneration and complied in the short term. The effect illustrates the importance of the suspect's interpretation: relatively restrained phrasing from an official actor may communicate an implicit promise of later correction. The evidence does not prove that every bluff has this effect in the field; it establishes a replicable experimental risk and a candidate mechanism for further naturalistic testing.
Interview recording can establish what was actually disclosed and when, but the recording alone does not remove distortions introduced by earlier contact, selective questioning, or a pre-existing narrative. The evidentiary sequence should therefore include the earliest contacts, initial accusation, disclosure of evidence, pauses, and all subsequent restatements. An extracted video clip that begins with the completed admission can make a psychologically consequential process disappear. Distinguishing investigator-supplied information from suspect-originated facts is especially important when the alleged crime has been widely reported or multiple investigators have discussed details with the suspect. A complete time-stamped record supports later scrutiny of whether admissions preceded or followed suggestive material. Kassin et al. (2025) Kassin et al. (2012)
Minimization, Maximization, and Pragmatic Implication
Minimization is a family of techniques that can reduce the perceived moral gravity of conduct or suggest sympathetic explanations for the suspected act. An interviewer may describe the event as understandable, commonplace, an accident, or the consequence of circumstances rather than character. When such language is paired with an expectation that admission will lead to gentler treatment, the psychological consequence may resemble an explicit offer of leniency. Maximization, conversely, can magnify expected punishment, suggest overwhelming proof, or portray continued denial as costly. Both techniques are best analyzed in terms of how the suspect construes the likely consequences of speaking. Their linguistic forms vary, and the same phrase may operate differently across institutional context, language, and personal vulnerability. Horgan et al. (2012) Kassin et al. (2010)
The experiment by Russano et al. (2005) is consequential because it separates the interests of guilty and innocent participants under controlled circumstances. Minimization and an explicit deal increased confessions in both groups, demonstrating that a tactic's ability to increase disclosure is not enough to establish the diagnostic value of the elicited admissions. Horgan et al. (2012) then investigated how minimization and maximization altered participants' perceptions of likely consequences. Its results emphasize the perceived incentive structure: statements that change anticipated outcomes can change confession choices even without a formal promise or threat. Researchers must measure what participants actually understood rather than treating an interrogator's intended meaning as the only psychological input.
Conversational pragmatics matters because speakers infer meaning beyond literal wording. Under conditions of authority, telling a suspect that 'everyone makes mistakes' while urging admission may communicate a possible reduction in culpability or sanction. The critical scientific variable is the anticipated difference between confessing and denying. This difference changes as questioning continues and as investigators reveal, suppress, or invent information. The results of one manipulation are not a general prohibition on all expressions of empathy. Respectful listening, emotional validation, and open questions can be consistent with accurate information gathering. What requires scrutiny is the coupling of apparent sympathy to an implicit requirement of self-incrimination. Horgan et al. (2012) Brimbal et al. (2021)
Maximization raises additional concerns when assertions of inevitable punishment or overwhelming evidence make the suspect believe that a truthful denial has no chance of being accepted. Repeated confrontation can convert the interview into a choice between two anticipated harms rather than a search for what happened. Yet psychologically intense questioning cannot be reduced to any single tactic in isolation. Duration, sleep and food, age, individual reasoning, interpersonal style, uncertainty, and access to assistance can combine to alter perceived costs. Studies that experimentally manipulate one technique have strong internal control over that feature while offering limited evidence about the compounded impact of many techniques over prolonged periods. An adequate model must therefore analyze sequence, accumulation, and interaction. Kassin et al. (2025) Scherr et al. (2020)
Stress, Fatigue, Temporal Pressure, and Decision Horizons
Interrogation outcomes emerge from decisions made under unequal control of time. The investigator may know when questioning is expected to end; the suspect often does not. If continued denial appears to prolong confinement, shame, anxiety, or exhaustion, admission may acquire immediate instrumental value independent of its truth. This creates a short-horizon choice in which present relief is salient and the eventual legal costs of a statement remain uncertain or psychologically remote. The same person might reject an incriminating proposition when rested and supported but accept it after repeated confrontation and uncertainty. Research on false confessions treats excessive interrogation time, isolation, and related demands as interacting situational risks, while acknowledging the difficulty of experimentally reproducing severe detention conditions. Kassin et al. (2010) Kassin et al. (2025)
Stress may impair access to stable autobiographical information, the ability to evaluate competing consequences, or the capacity to maintain resistance to a dominant account. These are plausible process pathways, not interchangeable empirical claims. A rigorous study must specify whether the dependent measure is memory accuracy, compliance, decisional competence, or later persistence in a false narrative. Laboratory manipulations that create mild frustration or time pressure cannot establish effect magnitudes for sleep deprivation or prolonged confinement. Conversely, archival cases cannot usually isolate which component of a complex coercive episode caused admission. The research needs ethically designed converging evidence from validated stress tasks, field interviews, process recordings, and naturally varying procedural safeguards. Kassin (2005) Catlin et al. (2024)
Time pressure can also operate through institutional communication. When an official repeatedly asserts that the process will end when the suspect provides a satisfactory explanation, the expected payoff of resistance may collapse. If a person has limited knowledge of procedural rights, the difference between temporary questioning and a consequential recorded admission may be poorly appreciated. At this point the psychological object is an evolving decision horizon. Researchers should measure perceived exit options, understanding of rights, expectations about evidence, and forecasts of the costs of confession. Realistic assessment requires clear separation of what interviewers said from what suspects inferred. This is one reason a recording must capture questions, pauses, instructions, and the atmosphere preceding the ultimate admission, rather than only the admission itself. Kassin & Norwick (2004) Steering Committee of Experts (2021)
Investigator Bias, Confirmation Processes, and the Escalation of Suspicion
Investigative judgment begins before the first formal accusation. Officers may receive a selective case summary, encounter a suspect's nervousness, or treat ambiguous behavior as deception. The critical issue is whether the evidence justifies the initial suspicion and whether subsequent information can genuinely disconfirm the working hypothesis. Meissner & Kassin (2002) experimentally separated the accuracy of truth–deception discrimination from a tendency to categorize people as lying. In that study, relevant experience and training affected response bias rather than the ability to identify deception. That finding cautions against equating confidence or occupational expertise with reliable diagnosis of guilt. A correct investigative hypothesis can still be pursued unfairly, and an incorrect hypothesis can intensify because confident denials are reinterpreted as resistance.
Confirmation bias is not merely an investigator's private belief. It can alter the questions asked, the evidence withheld, the intensity of confrontation, which explanations are recorded, and the weight assigned to apparent inconsistencies. Narchet et al. (2011) experimentally manipulated perceptions of guilt and investigated the subsequent interaction using a mock-crime paradigm. Bias increased minimization tactics and false confessions among innocents in that setting. The study helps connect initial expectations to behavioral treatment and then to outcomes, rather than treating bias as an abstract mental error. Its ecological limitations remain important: real investigators encounter more complex case files, legal constraints, and institutional supervision than participants in a laboratory model.
Once a confession appears, the process can become self-validating. Investigators may revisit physical traces, obtain new witness statements, or ask forensic personnel to evaluate ambiguous material while knowing the confession narrative. This information can influence ostensibly independent judgments and create convergence between evidence sources. Kassin et al. (2012) analyzed DNA-exoneration case files and found patterns consistent with the hypothesis that confession evidence can contaminate other evidence. Because this was an archival analysis of selected cases, temporal ordering and correlation must be evaluated carefully; the study does not show that every observed corroborating error was caused by the confession. The practical epistemic standard is to protect blind and independent examination where possible and to document information flow.
An open investigative process continually generates and examines alternative explanations. A contradiction is an investigative lead that should be tested, rather than automatic proof of deceit; a refusal to guess should be preserved as such, not rewritten as an evasive admission. Counterhypothesis testing asks what evidence would be expected if the suspect were innocent and which observations would favor another perpetrator or an innocent explanation. These practices are relevant to human–artificial decision support as well: a system that summarizes police files or proposes interview questions can reproduce prior assumptions unless its evidentiary provenance and uncertainty are explicitly managed. This is a prospective interdisciplinary application of established research on confirmation and information contamination, not an empirical assertion that artificial interviewers have already solved the problem. Kassin et al. (2025) Meissner et al. (2014)
Age, Disability, Suggestibility, and Contextual Vulnerability
Suspects vary in developmental maturity, comprehension, cognitive capacities, communication style, and experience with institutions. Youth is a documented risk context because adolescents may evaluate short-term consequences differently, be more responsive to adult authority, and misunderstand the long-term meaning of an admission. In a laboratory study including younger adolescents, older adolescents, and young adults, Redlich & Goodman (2003) found that younger and more suggestible participants were more likely to accept responsibility for an event they had not caused. The result identifies age-related and individual-difference vulnerability under experimental conditions; it is not a diagnostic rule applicable to every juvenile. Policies should attend to the capacity of individual suspects while recognizing that developmental features can systematically affect whole classes of interviews.
Intellectual disabilities, communication difficulties, and some mental health conditions can affect the ability to understand complex warnings, track contradictory questions, resist leading suggestions, or make informed choices under pressure. Such categories are heterogeneous. A diagnostic label alone cannot establish that a statement was false or that a particular suspect lacked decision-making capacity. Assessment requires individualized examination of comprehension, reasoning, suggestibility, symptoms, language proficiency, and the actual interview environment. Kassin et al. (2010) and Kassin et al. (2025) review vulnerability as a multi-factor problem and recommend protective procedures for younger persons and vulnerable adults. Clinical expertise, when relevant, should be applied to specific functional questions instead of substituting a general psychiatric description for evidence.
Interrogative suggestibility concerns the extent to which an individual accepts misleading propositions or changes answers under social pressure and feedback. It can be investigated with structured tests, repeated questioning, and analysis of response changes, while recognizing the limits of psychometric generalization. High suggestibility may be one contributor to a false account, especially where investigators repeatedly introduce contested propositions as settled fact. But vulnerability also depends on situational arrangements: fatigue, fear, confusing procedures, a language barrier, a lack of counsel, and an inability to understand the consequences of agreement can increase effective pressure. Effective safeguards must be evaluated as part of the process rather than treated as a mere attribute of the participant. Redlich & Goodman (2003) Kassin et al. (2025)
Research should avoid replacing individual assessment with stereotypes. Neither youth nor disability licenses an assumption that every confession is unreliable, nor does adult intelligence guarantee resistance to false evidence. Investigators and evaluators must examine the interaction between person's capacities, specific questions, claims about evidence, and actual circumstances. Cleary & Warner (2016) documented how police training and practices were reported for juvenile and adult suspects, reinforcing the importance of investigating the institutional side of developmental vulnerability. The Steering Committee of Experts (2021) principles address heightened vulnerability through appropriate adjustments, access to safeguards, and accountable professional interviewing. Their normative recommendations are informed by science but should be distinguished from an experimental estimate of any particular reform's effect.
Memory Suggestion, Internalization, and the Provenance of Crime Details
Memory is reconstructive, and a narrated account can change as people encounter questions, feedback, and alternative explanations. In an interrogation, an investigator may introduce a proposed sequence and invite the suspect to correct or complete it. A correction may itself become a new source of information for later questioning. The resulting narrative can be internally coherent even where its component details were selected, supplied, or validated by the interviewer. This is a source-monitoring problem: who originally contributed each detail, when did the suspect learn it, and what did the person believe about the source at the moment of endorsement? The distinction matters because the independent discovery of information known only to the perpetrator may have evidentiary value, whereas repetition of an investigator-disclosed detail cannot be counted as independent confirmation. The DNA-exoneration analysis of Kassin et al. (2012) and the contemporary review Kassin et al. (2025) underscore the need to reconstruct information pathways before interpreting a confession's apparently accurate content.
The experimental demonstration of internalized false admissions in Kassin & Kiechel (1996) established a useful distinction among action, belief, and confabulated detail. When a purported witness provided false evidence about a laboratory mishap, some participants signed a confession, some subsequently appeared to believe they had caused the event, and some constructed descriptions that fitted the accusation. Those outcomes should be interpreted through the particular operational measures in the original experiment, not as proof that all confabulatory statements constitute durable autobiographical memories. Belief change can be transient, conditional, or uneven. An account may contain language borrowed from the interviewer while the speaker privately doubts the allegations. Researchers should therefore distinguish an admission, a belief rating, a memory claim, and independently corroborated event knowledge as separate outcomes.
Repeated retelling can make a negotiated story feel familiar, and familiarity may be mistaken for the strength of original recollection. Yet it is difficult to isolate repetition from investigator feedback, anticipated consequences, or the effort to make an accusation intelligible. A stronger process study would record the first unprompted account, the exact sequence of introduced facts, subsequent changes in wording, and immediate and delayed belief assessments. It would use independent coders to distinguish spontaneous disclosures, prompted agreement, guesses, and investigator-provided content. These design choices generate evidence about when the account acquired its features instead of assuming that final fluency demonstrates original knowledge. Kassin et al. (2010) Kassin et al. (2025)
Information contamination may extend outside the interview room. Media reports, social discussions, police press releases, prior interviews, and co-suspect statements can provide ordinary routes by which an innocent person acquires accurate crime facts. Evaluators must examine these possibilities case by case. Investigators can improve independence by withholding selected nonpublic details during initial free recall, testing volunteered statements against objectively documented evidence, and retaining contemporaneous records of what was disclosed. Those recommendations are informed by the established problem of source contamination; they should not be mistaken for a claim that any single detail proves guilt. Properly evaluated corroboration requires a chain of custody for information as well as for physical exhibits. Leo (2009) Kassin et al. (2012)
Experimental Paradigms and the Identification of Causal Mechanisms
The strongest way to estimate whether a tactic causes false admissions is to establish ground truth independently of the interview and randomly vary the relevant tactic. Experimental paradigms differ in how they achieve that goal. In innocence-only settings, participants have not caused the alleged event, allowing the investigators to measure susceptibility to an accusation, false evidence, or suggestion. In guilt–innocence paradigms, researchers arrange for one group to engage in a rule violation while another follows the rules, then expose both to manipulated interviewing conditions. The first type provides a clear test of the possibility and mechanism of false admission; the second additionally permits assessment of diagnosticity. Distinguishing these designs prevents a common interpretive error in which a procedure that produces many statements in one group is called accurate without a corresponding comparison group. Kassin & Kiechel (1996) Russano et al. (2005)
The paradigm introduced by Russano et al. (2005) represents a major experimental advance. Participants worked with a partner under instructions limiting collaboration; some were placed in a condition where a rule violation occurred and others were not. Interrogation manipulations included minimization and an explicit deal. The design permitted comparison of true and false confessions under specified conditions, and the results demonstrated that manipulations which raise confessions can also raise false admissions. However, experimental 'guilt' in a low-stakes task cannot reproduce the moral, social, and legal significance of a serious criminal allegation. Nor can the manipulated confession necessarily be assumed to involve the same cognitive process as a lengthy self-incriminating narrative after prolonged custody. Internal validity and ecological validity must therefore be evaluated together.
Later variants probe specific causal pathways. Perillo & Kassin (2011) experimentally tested false-evidence statements and bluffs about forthcoming evidence, showing how the prospect of future vindication may encourage an innocent participant to accept immediate blame. Narchet et al. (2011) manipulated investigator expectations and measured differences in questioning and outcomes, addressing the path from initial belief to interviewer behavior. Redlich & Goodman (2003) examined age and suggestibility among participants accused of an act they did not commit. Together, these studies establish several distinct levers: evidence credibility, expected future information, interviewer prior belief, and susceptibility to social pressure. Their independent estimates should not be added as if they represented separate nonoverlapping risks in the same real-world population.
Experimental ethics impose necessary limits. Researchers cannot realistically threaten study volunteers with a criminal conviction, deprive them of basic care, or reproduce the complete power of detention. As a result, the most coercive field circumstances are difficult to randomize. Ethical field evaluation can instead compare changes in training, recording, and interviewing policy while preserving safeguards and measuring independent investigative outcomes. Such studies require attention to case selection, investigator assignment, attrition, and changing legal practices. Mixed-method approaches should combine transcripts, behavioral coding, participant understanding, objective evidence, and outcomes. A robust research program asks not simply whether a tactic works in a laboratory but which components of the mechanism remain stable when institutional stakes and safeguards change. Catlin et al. (2024) Steering Committee of Experts (2021)
Meta-Analytic Evidence on True and False Confessions
The accumulated experimental literature offers stronger comparative evidence than any single mock interrogation. Meissner et al. (2014) synthesized research contrasting accusatorial and information-gathering interviewing approaches and reported meaningful differences in confession outcomes. The later Campbell systematic review by Catlin et al. (2024) updated this inquiry through studies available up to 2023. It identified 27 eligible articles contributing 29 independent experiments and 81 effect sizes. The authors performed network meta-analyses to compare broad approaches and selected component tactics, assessed risk of bias, and discussed the constraints of combining experiments with heterogeneous scripts and outcomes. These designs answer a specific question: within controlled experimental simulations, how do the odds of true and false admissions change across interview conditions?
The 2024 synthesis provides quantitatively informative contrasts. Relative to information-gathering approaches, accusatorial approaches had higher odds of false confession, with a combined odds ratio of 4.41 and a 95% confidence interval from 1.77 to 10.97; the authors judged the certainty of this comparison moderate. Relative to direct questioning, accusatorial approaches also had higher false-confession odds, with an odds ratio of 3.03 and interval from 1.83 to 5.02. Information-gathering approaches increased the odds of true confessions relative to direct questioning, with an odds ratio of 2.43 and interval from 1.29 to 4.59. In the comparison of information-gathering with accusatorial approaches for true confessions, the estimate favored information gathering but its confidence interval included no effect; the certainty was low. These are odds ratios from the review's study population, not absolute probabilities of a false confession for real suspects. Catlin et al. (2024)
The qualification about uncertainty is substantive. An odds ratio cannot tell a police agency how many innocent people will confess without the underlying risk distribution and applicable base rates. Some meta-analytic comparisons draw on relatively few direct studies; categories such as 'accusatorial' combine techniques that differ in intensity and wording. The network model permits indirect comparisons under assumptions about comparability, but the six-node analysis of false confession tactics showed important heterogeneity. The Campbell authors also identified potential publication-selection bias and inconsistent reporting of what investigators actually said. Their evidence supports moving toward information-gathering methods while supporting further direct comparisons under transparent and replicable protocols. Catlin et al. (2024)
A separate synthesis by Stewart et al. (2018) assessed prevalence of false admissions in laboratory simulations. That question is valuable for studying the frequency of experimental susceptibility under specific conditions, but an aggregate laboratory rate must not be reported as the population prevalence of wrongful admissions during real police interrogation. Recruitment, stakes, instructions, outcome definitions, and experimental manipulation differ from criminal practice. Even among field records, there is no complete register of all innocent suspects who were questioned and did or did not confess. The sound cumulative conclusion is therefore a causal risk pattern supported by several controlled paradigms, an experimentally supported relative advantage for information gathering in important comparisons, and unresolved uncertainty about general-population absolute risk. Stewart et al. (2018) Catlin et al. (2024) Kassin et al. (2025)
Field Observation, Documented Wrongful Convictions, and Selection Bias
Field evidence provides ecological detail unavailable to controlled laboratory studies. Recorded police interviews allow analysts to observe the progression of questioning, rights explanations, disclosure of evidence, rapport, conflict, pauses, and the provenance of a signed account. Archival case files can reconstruct how investigators interpreted denials, obtained admissions, and moved from a confession to apparently corroborating evidence. The limitation is that ground truth in live cases is often uncertain and the observed interview population is selected by prior investigative decisions. A record of statements from people later convicted cannot serve as an unproblematic sample of truth, especially when the confession influenced the conviction. Strong field work must classify the basis for guilt or innocence independently of the confession wherever possible. Leo (2009) Kassin et al. (2025)
DNA-based exonerations provide unusually powerful evidence that at least some confessions were false. They also illustrate why the known number of cases is a lower-bound record of detected miscarriages of justice, not an estimate of the rate at which innocent people confess. Eligibility for biological testing, preservation of biological material, access to counsel, publicity, postconviction review, and the decision to pursue exoneration all shape which cases are known. Some crimes cannot produce exculpatory DNA evidence, and many recorded admissions may never be tested independently. Scientific inferences must therefore identify the case-selection mechanism. Leo (2009) Kassin et al. (2012)
The study by Kassin et al. (2012) examined DNA-exoneration case files and documented the potential for confession evidence to influence subsequent interpretation of eyewitness, forensic, or other evidence. This is an evidentiary-dependence problem. Suppose a crime scene analyst, witness interviewer, or prosecutor has already read a confession; later expressions of confidence may partly reflect that exposure rather than a separate source of proof. Treating these observations as independent corroboration can produce inflated certainty. Archival sequence evidence is especially helpful for determining whether a detail existed before the confession, whether its interpretation changed afterward, and who had access to the statement. The same process can occur without intentional misconduct, through ordinary cognitive and institutional dynamics.
Field research should avoid reducing every wrongful conviction to a single cause. A false confession may appear alongside mistaken identification, tunnel vision, flawed forensic inference, undisclosed exculpatory evidence, or ineffective legal assistance. These mechanisms interact. Scherr et al. (2020) proposes a cumulative-disadvantage framework in which the initial predicament of an innocent suspect can become the basis for later decisions that increase the chance of conviction. Its explanatory strength lies in directing attention to transitions between stages and mechanisms of feedback. Its limitation is that complex causal chains are not easily quantified from retrospective case collections. The framework should motivate prospective sequence-sensitive measurement and explicit tests of alternative explanations, rather than simply naming a set of correlated errors.
Courtroom Evaluation, Juror Judgment, and the Persistence of Confession Effects
An interrogation produces evidence for later decision-makers, and those decision-makers must judge reliability under conditions of incomplete knowledge. Jurors may assume that self-incrimination is intrinsically more probative than other statements, that no innocent person would confess, or that an elaborate narrative reflects direct memory. A confession can therefore anchor subsequent interpretation of inconsistent testimony and forensic evidence. Kassin et al. (2005) compared police investigators and students in judgments of confession veracity, revealing limits in observers' ability to discriminate true from false statements. Experience and confidence should not be equated with validated detection ability. The research problem in adjudication is how fact-finders combine confession content, interrogation process, external corroboration, voluntariness, and competing explanations.
A systematic review and meta-analysis by Mindthoff et al. (2024) evaluated the influence of confession evidence on mock juror verdict decisions across experimental studies and considered safeguards. It documented the pronounced influence of confession evidence on conviction judgments, including in conditions involving coercion concerns. Such findings establish that a confession's impact can survive information intended to qualify it, but the precise magnitude depends on experimental conditions, instructions, and what alternative evidence jurors saw. Mock-juror verdicts are not identical to actual trial verdicts: participants may face simplified records, no real sentencing consequences, and different deliberative procedures. Nevertheless, controlled comparisons isolate the psychological weight assigned to the statement when other case information is held constant.
Videotape presentation introduces a distinct measurement issue. Lassiter et al. (2002) found that camera perspective in simulations changed how observers interpreted confession voluntariness and trial outcomes; work on the camera-perspective bias shows that an apparently objective recording is still a selectively framed perceptual stimulus. A suspect-focused camera can draw attention to the speaker while reducing access to the interrogator's behavior, even though both participants generated the exchange. Technical safeguards must therefore address completeness of the temporal record and viewpoint, not just the existence of a camera. A recording that preserves both parties and the entire interview can make procedural context more available to evaluators.
A meaningful evaluation should distinguish the factual proposition 'the suspect committed the act' from the procedural proposition 'the suspect made the statement voluntarily under applicable law.' It should test whether the suspect introduced nonpublic accurate details before disclosure, whether investigators falsely described evidence, and whether later 'corroboration' was contaminated by the confession. Expert psychological testimony can explain empirically established risks without purporting to decide the historical truth of a particular disputed statement from a few behavioral cues. Judges and jurors may require instruction about the structure of questioning, the role of incentives, and the limits of intuitive lie detection. Whether any particular instruction or expert presentation improves verdict accuracy must itself be tested, rather than assumed from good intentions. Mindthoff et al. (2024) Kassin et al. (2025)
The Interrogation as a Sequential Decision Process
Interrogation research frequently reports whether a participant confessed by the end of an encounter, but a final binary outcome conceals the sequence by which the decision was reached. A suspect can begin by denying involvement, revise a peripheral detail, accept an investigator's characterization of one event, and eventually sign a statement that organizes those changes into a full admission. Each step alters the evidence available to the interviewer and the suspect's beliefs about which responses will be accepted. A scientifically adequate process model must therefore represent the interview as a sequence of information disclosures, forecasts, social responses, and decisions. The relevant unit is not merely a confession outcome but a trajectory with turning points and observable changes in the suspect's and investigator's behavior. Laboratory manipulations of investigator expectations Narchet et al. (2011) and perceived confession consequences Horgan et al. (2012) supply individual components for this wider model.
A practical causal diagram would begin with pre-interview evidence and prior beliefs, then track the assignment of an interviewing strategy, disclosure of actual and asserted evidence, elicited statements, immediate feedback, and the decision to continue, retract, or amend. Background characteristics such as age and comprehension affect how information is interpreted, and institutional constraints affect the range of available actions. The diagram can distinguish mediator variables, including perceived inevitability of conviction or expected relief, from outcome variables, including signed confession, detailed narrative, and independent evidentiary corroboration. Only some paths are identified experimentally in the existing literature. Field recordings can help describe the sequence, but causal estimation requires independent variation or defensible quasi-experimental assumptions. Russano et al. (2005) Redlich & Goodman (2003) Kassin et al. (2025)
The sequential approach also explains why investigators and suspects can diverge in their interpretations of the same conversation. An investigator may think that an admission follows increasing weight of evidence, while the suspect understands it as the fastest route to terminating an aversive interaction. Without measures of subjective expectation, the transcript by itself may not reveal this divergence. Researchers could collect contemporaneous judgments at defined intervals in simulations and compare them with final statements, permitting analysis of how beliefs and choices coevolve. In field settings, such measurement must respect legal rights and avoid introducing additional pressure. This approach could clarify when false information becomes psychologically credible and which procedures interrupt the transition from uncertainty to an unreliable admission. Perillo & Kassin (2011) Kassin (2005)
Language, Communication, and Cross-Cultural Interpretation
Interrogation is linguistically mediated. Questions contain presuppositions, invitations to agree, temporal claims, descriptions of evidence, and descriptions of consequences. A suspect must interpret literal meaning and pragmatic implication while attending to the authority of the questioner. Limited proficiency in the interview language can increase the risk of misunderstanding modality, negation, legal terminology, or temporal ordering. Interpretation can also change conversational pacing and the perceived firmness of a question. These are theoretically grounded concerns within the broader psychology of confession; quantitative conclusions about their relative prevalence across particular linguistic groups require dedicated studies rather than inference from general susceptibility experiments. The conceptual relevance of implication is supported by work on minimization and consequence perception Horgan et al. (2012), while Steering Committee of Experts (2021) identifies comprehensible communication and safeguards as essential interview principles.
Cross-cultural differences may influence when respect is expressed through agreement, how people respond to authority, and whether a direct denial is socially permissible or perceived as rude. Such possibilities should be investigated with careful attention to within-group variation and the actual setting. Cultural affiliation is not a proxy for an individual's truthfulness or suggestibility. The empirical design should test translated protocols for semantic equivalence, ensure that rights explanations are comprehended, and examine whether interview outcomes depend on the interviewer–suspect language pairing. Comparative results need to distinguish a change in confession rates from a change in the accuracy or informativeness of the resulting statements. Catlin et al. (2024) Steering Committee of Experts (2021)
A reliable record should preserve the original-language questions and answers, the interpreter's interventions, and any subsequent translation used in investigative summaries or court. Translation should be considered a stage in evidentiary provenance: if a suspect uses a phrase that could mean acknowledgment of presence rather than acknowledgment of the act, later paraphrase can inadvertently transform ambiguity into apparent confession. This is a testable research problem, not an allegation that translation usually produces false admissions. Evaluators can compare the original conversation with independent translations and identify which changes alter the probative meaning of a statement. As human–artificial transcription and translation become more common, the same evidentiary requirements apply to algorithmically generated transcripts: preserved original speech, clear uncertainty marking, accountable corrections, and a documented revision history. Kassin et al. (2025)
Assessing a Disputed Confession as a Scientific Evidence Problem
A disputed confession presents a retrospective inference problem. The evaluator first needs an event chronology: evidence that existed before police suspicion, the path by which the suspect was selected, each documented contact, information the suspect possessed, the actual questioning, and all later statements. Particular attention should be given to the earliest unprompted account and to nonpublic details that the suspect correctly reported before they could have been learned through other means. A chronological reconstruction is necessary because the same phrase may have different evidentiary significance depending on what the person already knew. The evaluator should identify gaps, competing explanations, and information that cannot be independently corroborated. Leo (2009) Kassin et al. (2012)
The second step is to assess the person–situation interaction. Age, comprehension, suggestibility, disability, mental state, language, and prior experience can influence how the suspect understands accusations and warnings. Interview duration, isolation, representations of evidence, incentives, and treatment of denials may influence expected costs. These factors should be supported by records and specialized assessments when appropriate rather than inferred from a narrative stereotype. The expert must avoid the inverse fallacy: because a method is associated with false confessions, an individual exposed to it is not thereby demonstrated innocent. Risk factors change the credibility of an inference and guide further testing; they do not determine historical truth independently of case-specific evidence. Redlich & Goodman (2003) Kassin et al. (2025)
The third step examines corroboration as a problem of independence. An apparently accurate statement about a crime should be partitioned into details already public, details disclosed during questioning, details plausibly acquired through ordinary channels, and details that appear demonstrably novel and verifiable. Subsequent forensic or witness judgments should be classified according to whether the decision-maker was already exposed to the confession. The objective is to avoid counting dependent evidence twice. In the archival analysis of exoneration cases, Kassin et al. (2012) identified the potential for a confession to influence multiple later evidence streams. This mechanism provides a scientific reason to require informational separation and to report departures from that separation explicitly.
Finally, an evaluator should state uncertainty proportionately. An incomplete recording, a missing initial interview, uncertainty over who introduced a detail, or an unverified claim about a witness may justify a conditional conclusion rather than a definitive opinion. The legal question of admissibility and the scientific question of reliability should be articulated separately. Conclusions should identify each factual observation, the established empirical mechanism to which it relates, and the remaining alternative explanations. Scientific testimony is strongest when it equips the court to evaluate the process without claiming a validated ability to determine the truth of a unique confession from demeanor alone. Kassin et al. (2005) Mindthoff et al. (2024)
Organizational Learning and Evaluation of Interrogation Reform
An investigative organization can adopt an evidence-informed training policy while retaining performance incentives that reward confessions over case accuracy. Reform must therefore encompass supervision, documentation, feedback, and the evaluation of results. Clear procedural expectations include preserving complete recordings, documenting the introduction of specific evidence, distinguishing suspect-originated details, and recording the investigation of disconfirming leads. Training can develop open questioning, rapport, and alternative-hypothesis testing, but observed practice is the appropriate measure of implementation. The 2021 training study Brimbal et al. (2021) provides evidence that interviewing behavior can change following instruction; whether such improvements persist, transfer to high-stakes cases, and reduce false admissions needs further independent testing.
Organizational outcome measures should reflect both the benefit and the cost of an investigative method. Relevant indicators include the amount of independently verified information, the diversity of plausible hypotheses examined, procedural comprehension, and documented correction of erroneous assumptions. True and false confessions are ideal endpoints only when ground truth can be credibly established; in many real investigations this remains difficult. A policy that appears effective by increasing confession counts may perform poorly on diagnosticity. The comparative logic in Russano et al. (2005) and the experimental synthesis of Catlin et al. (2024) gives agencies a scientific basis for avoiding that metric as the sole measure of professional success.
Independent auditing can also reduce the selective visibility of successful or notorious cases. Agencies should assess a representative range of interviews, including interviews that produce denials, exculpatory information, partial admissions, and no statement. Coding protocols should specify the interviewer behaviors being measured and report interrater agreement and missing records. Evaluation should account for changes in suspect selection, offense mix, available forensic technology, and legal procedure over time. The objective is to identify whether reforms altered causal mechanisms, not simply whether a new manual was disseminated. Cross-agency comparisons require comparable definitions and sufficient transparency about recruitment and context. Snook et al. (2021) Catlin et al. (2024)
The organizational challenge extends to automated workflow systems. Software can help preserve sequences, flag missing records, and link evidence to its original source; it can also make an erroneous narrative more persuasive if summaries suppress uncertainties or repeatedly reproduce the same unverified confession detail. Rigorous evaluation would audit generated outputs against complete source recordings, measure investigators' ability to detect inaccuracies, and ensure that a change in technology does not silently change standards for evaluating admissions. The scientific criterion remains unchanged across manual and computational workflows: information should be judged according to demonstrable origin, independence, and relevance to the event. Artificial participation changes the sociotechnical setting and raises research questions that must be addressed with new evidence. Kassin et al. (2025) Steering Committee of Experts (2021)
Information-Gathering Interviews, Rapport, and Investigative Accuracy
Information-gathering interviewing organizes the encounter around obtaining a detailed and testable account rather than securing an acknowledgment of a predetermined theory. An investigator may begin with an explanation of procedure, invite a free narrative, ask open questions, clarify inconsistencies without treating them as proof of deception, and evaluate an account against independently obtained evidence. Rapport matters as a relational condition for productive communication: respect, predictability, active listening, and appropriate autonomy can increase willingness to disclose without implying a favorable consequence in return for incrimination. This orientation has been developed in investigative interviewing traditions including the British PEACE framework and is reflected in the international Principles on Effective Interviewing. Its psychological value lies in creating opportunities for accurate reporting and for the investigation to change its hypothesis when new evidence warrants doing so. Steering Committee of Experts (2021) Meissner et al. (2014)
The controlled literature supports this direction, subject to study-specific limits. The updated Campbell review Catlin et al. (2024) found information-gathering approaches favorable for eliciting true confessions relative to direct questioning and favorable for avoiding false confessions relative to accusatorial approaches. The evidence is stronger for some contrasts than for others, and the review notes heterogeneity and incomplete reporting of interview scripts. In a field experiment with police investigators and interviewees, Brimbal et al. (2021) examined rapport-based training and interview behavior. It contributes practical evidence about whether trainable communication skills can alter interview process and information provision, while the causal interpretation of its outcomes depends on field assignment, measurement, and the specific participants studied. Such studies complement laboratory comparisons without establishing that all forms of rapport are equally protective.
The term rapport should not become a substitute for specifying procedures. An interviewer may be courteous while still presenting false evidence or implying a reward for confession; the combined procedure has to be studied as a whole. Likewise, openness to alternative explanations must be observable in the handling of a suspect's denials and the timing of evidence disclosure. A well-designed evaluation measures the quality, accuracy, and independently corroborable novelty of information, as well as the frequency of true and false admissions. Training should be audited for fidelity: what practitioners actually did after training, whether they reverted to accusatorial methods under pressure, and whether supervisory feedback rewarded accurate inquiry rather than a predetermined closure. Brimbal et al. (2021) Catlin et al. (2024)
An ethical information-gathering method preserves the suspect's capacity to choose whether and how to answer within the relevant procedural system. The interview should be intelligible and adapted to language and disability needs, with opportunities to clarify misunderstandings and a record of how contested information entered the exchange. Researchers can study whether evidence-based interviewing reduces contamination and improves the calibration of investigators' conclusions. The preferred scientific end point is investigative discrimination coupled with procedural fairness. Successful practice may yield a truthful admission, a credible denial, exculpatory information, or a documented area of unresolved uncertainty. Each is potentially valuable because the process is designed to discover what happened, including evidence inconsistent with the initial suspicion. Steering Committee of Experts (2021) Kassin et al. (2025)
Procedural Safeguards, Recording, and Independent Verification
A protective procedure must intervene at a mechanism that produces error. Early access to legal advice can improve a suspect's understanding of procedural choices; developmentally appropriate support can reduce comprehension failures; complete recording can preserve the sequence of prompts and statements; and limits on deceptive evidence claims can reduce some experimentally demonstrated sources of false admission. These safeguards act at different stages, so their effectiveness should be measured separately and in combination. A rule requiring recording may make later evaluation possible without necessarily preventing an interviewer from using coercive language. A rule limiting evidence fabrication targets the generation of misleading inferences but does not by itself control excessive duration or confirmation bias. The evidence warrants a systems approach that links a particular risk mechanism to the procedural response designed to address it. Kassin et al. (2010) Kassin et al. (2025) Snook et al. (2021)
Recording from the earliest meaningful contact to the final account is especially important for provenance. A complete record helps establish whether a suspect volunteered details, received them through prompting, or adopted them after suggestive confrontation. It can document breaks, changes of interviewer, rights discussions, and retraction or correction. Audio–video geometry matters because observers are sensitive to presentation perspective; the experimental findings of Lassiter et al. (2002) caution that suspect-only framing may change judgments about voluntariness. Recording can make review more transparent, but secure preservation, appropriate access controls, faithful transcription, and the availability of the full file are also necessary. Edited excerpts or summaries can recreate the appearance of an unprompted confession even when the underlying exchange was heavily guided.
Independent verification addresses the downstream influence of confession evidence on other tests. Investigators can maintain clear documentation of nonpublic crime details, evaluate physical evidence without irrelevant case narratives where feasible, and ask whether corroborating information genuinely originated outside the confession process. Kassin et al. (2012) demonstrates why this separation matters in DNA-exoneration files. Evidence management should specify when a forensic examiner, witness, or secondary investigator first learned about the confession. Where independence cannot be maintained, subsequent judgments should disclose their informational dependence instead of presenting apparent agreement as multiple autonomous confirmations. The result is a better calibrated body of evidence and a more accurate reconstruction of institutional reasoning.
Reform outcomes demand evidence beyond compliance checkboxes. A jurisdiction may document that cameras were installed or investigators trained without demonstrating fewer false confessions. Evaluation needs pre-registered outcome definitions where feasible, comparable case populations, auditing of actual interview behavior, reporting of refusals and retractions, and time-series or controlled designs appropriate to the institutional setting. False convictions are too rare and incompletely observed to serve as the only short-term outcome. Intermediate measures such as introduction of nonpublic details, unexplained changes in statements, length and comprehensibility of questioning, and independent corroboration can supply mechanistic indicators, provided their validity is independently assessed. The overarching purpose is continuous calibration of investigative truth-finding rather than mere proof of procedural adoption. Catlin et al. (2024) Steering Committee of Experts (2021)
International Frameworks, Developmental Protection, and Institutional Accountability
Interrogation occurs within legal systems that distribute rights and responsibilities differently, yet many underlying psychological mechanisms recur across settings. The 2021 Principles on Effective Interviewing for Investigations and Information Gathering offer an international framework integrating scientific knowledge, professional practice, and fundamental procedural safeguards. They emphasize effective communication, noncoercive information gathering, early protections for persons deprived of liberty, professional training, and institutional oversight. These are normative principles grounded in convergent evidence, not a randomized evaluation of an entire national justice system. Their relevance is that they align investigative success with the quality and reliability of the information produced and require authorities to consider the circumstances in which an account was obtained. Steering Committee of Experts (2021)
National implementation may differ in rights warnings, legal representation, recording obligations, evidentiary exclusion, interviewing environments, and protections for minors. The United States debate about Miranda waiver has particular psychological significance because Kassin & Norwick (2004) showed how experimentally innocent persons could underestimate the risks of talking without rights protection. Other systems, including British investigative interviewing practice, place more explicit institutional emphasis on planned information-gathering interviews. Comparative research should not equate formal adoption of one model with its actual implementation or infer the superiority of an entire national regime from a selected set of cases. Researchers must identify which procedures were implemented, in which population, by whom, and under what legal constraints.
Developmentally appropriate safeguards respond to identifiable psychological features of adolescence and childhood. Adolescents may be more focused on immediate consequences and more responsive to adult demands, while younger persons may have less experience interpreting legal procedure. Redlich & Goodman (2003) experimentally demonstrated age- and suggestibility-related variation in false responsibility taking. The appropriate response includes comprehensible explanations, safeguards for the presence and independence of suitable support, monitoring for fatigue or misunderstanding, and active assessment of whether an individual can follow the meaning of each question. Legal protection, however, cannot be determined from a single average age effect. Research must distinguish chronological age from functional comprehension, psychosocial development, prior experience, disability, and the specific setting.
Institutional accountability requires supervisors, courts, professional trainers, and researchers to assess the actual reliability of interview processes. Snook et al. (2021) outlines reform issues in North American interrogation practice, while Kassin et al. (2025) integrates more recent psychological evidence and risk-reduction proposals. A credible quality-assurance system makes it possible to audit recordings, detect excessive or misleading questioning, preserve suspect corrections, and track whether hypotheses were genuinely revised. Cross-cultural implementation should include language access and procedures suited to local communication norms; it should also permit independent external evaluation. Reform is an ongoing scientific and institutional process rather than a single replacement of one interview manual with another.
Research Controversies, Measurement Challenges, and the Artificial Era
The literature's strongest findings concern causal susceptibility to defined questioning manipulations and differences among experimental interview styles. Important uncertainties remain about effect sizes in real detention, the influence of multiple interacting tactics, and the absolute incidence of false confessions in differing suspect populations. The most informative studies have established ground truth, a clear transcript of manipulations, and measures that distinguish true admission, false admission, belief, and narrative detail. Many field cases lack those conditions. Reviewers must consequently keep separate the experimental risk of a tactic, its observed presence in documented miscarriages of justice, and its contribution to a particular confession. Treating these as one claim obscures what the evidence can support. Catlin et al. (2024) Kassin et al. (2025)
Variation in terminology creates a further challenge. Researchers use 'accusatorial,' 'minimization,' 'maximization,' 'rapport,' and 'information gathering' to label interventions that sometimes have different scripts and overlapping components. A useful research program would publish complete interview protocols, code tactical sequences with blinded raters, and report deviations from assigned methods. It should measure how suspects construed consequences, whether they believed investigator claims, and how their narratives changed over time. The Campbell authors Catlin et al. (2024) explicitly identified inconsistent manipulation reporting and heterogeneity as limitations. Improved preregistration, independent replication, data sharing when legally and ethically possible, and standardized yet flexible coding would sharpen effect estimates and identify conditions in which one interviewing approach outperforms another.
Future evidence synthesis should include representative participant pools and distinguish age, educational experience, language proficiency, disability, prior justice-system contact, and cultural expectations of authority. It should test combinations of tactics rather than assuming linear accumulation, compare trained practitioners with scripted experimental interviewers, and use process measures linked to objectively verifiable information. Longitudinal research may examine whether a false belief persists after interrogation, whether a retraction is correctly interpreted, and how confessional content influences later eyewitness or forensic assessments. Field studies should preserve confidential information and protect the rights of participants. The emerging goal is a causally explicit model of how investigative decisions transform uncertain information into institutional findings. Scherr et al. (2020) Mindthoff et al. (2024)
The Artificial Era adds a consequential new setting: artificial systems can assist in transcript search, evidence organization, interview planning, translation, risk alerts, and report drafting. Each function can affect the flow of information, the language of questions, and the perceived authority of an analysis. A summarization model may omit qualifying denials; a classification system may treat the presence of an admission as a proxy for guilt; a generated question may reveal an undisclosed fact and destroy its later value as independent corroboration. These are testable hypotheses and foreseeable design problems based on established research on confirmation bias, source contamination, and evidentiary dependence. They are not empirical findings that artificial agents have a particular rate of interrogation error. The appropriate research program compares human-only and human–artificial workflows on blinded ground-truth tasks, provenance preservation, suspect comprehension, auditability, and the detection of alternative hypotheses. Narchet et al. (2011) Kassin et al. (2012)
Artificial support should preserve human accountability and make evidentiary provenance legible. Researchers can investigate whether systems that explicitly trace who introduced each factual proposition reduce contamination; whether transcript interfaces that display interruptions and context change evaluator judgments; and whether automated prompts to seek disconfirming information improve investigative accuracy. These outcomes should be tested before claims of reliability. The wider psychological significance is that a confession is produced and assessed within a socio-technical system whose participants include investigators, suspects, lawyers, witnesses, forensic experts, adjudicators, and increasingly artificial tools. The individual statement must be analyzed as one event in that distributed process. The historical horizon From Homo to Artificial thus creates new objects for forensic psychology without altering the evidentiary requirement that claims be measured and independently tested.
Conclusion
Suspect interrogation is a psychological process in which authority, information, incentives, memory, developmental capacity, and institutional expectations shape the statements people produce. A false confession can arise through deliberate compliance, altered belief, voluntary motives, or combinations of these pathways. The research demonstrates that innocence is compatible with an admission under identifiable conditions. Foundational experiments distinguished behavior from internalization, while guilt–innocence paradigms established that some interrogation strategies increase both true and false admissions. The resulting concept of diagnosticity provides a necessary scientific criterion: a productive investigative method must improve discrimination between culpable and innocent persons rather than simply increase the number of recorded confessions. Kassin & Kiechel (1996) Russano et al. (2005) Kassin et al. (2010)
The accumulated evidence favors investigative information gathering over psychologically accusatorial methods on central outcomes measured in experimental studies. The 2024 Campbell review found higher false-confession odds in accusatorial conditions and an advantage for information gathering in eliciting true confessions relative to direct questioning, while documenting limits of certainty and generalizability. Other research links investigator bias to questioning choices, shows that bluffs can mislead innocent people about future exoneration, and identifies vulnerability associated with youth and individual circumstances. The confession's downstream evidentiary influence is a second major problem: a statement can shape later investigations and courtroom judgments, producing apparently convergent evidence that lacks genuine independence. Catlin et al. (2024) Narchet et al. (2011) Perillo & Kassin (2011) Kassin et al. (2012) Mindthoff et al. (2024)
Scientific and institutional progress depends on conserving the evidentiary history of each statement. Complete recording, transparent information provenance, independent testing of corroboration, appropriate legal and developmental safeguards, and disciplined information-gathering practice address different stages of the causal pathway. The research also requires stronger field comparisons, rigorous assessment of individual vulnerabilities, better reporting of tactics, and clearer estimates of actual real-world risks. A confession deserves evidentiary weight according to its demonstrated reliability, the conditions of its production, and its independently established relationship to the underlying event. That principle extends to emerging human–artificial investigative systems: their contribution will be measured by whether they improve the discovery, preservation, and critical evaluation of truthful information. Kassin et al. (2025) Steering Committee of Experts (2021)
References
- Brimbal, L., Meissner, C. A., Kleinman, S. M., Phillips, E. L., Atkinson, D. J., Dianiska, R. E., Rothweiler, J. N., Oleszkiewicz, S., & Jones, M. S. (2021). Evaluating the benefits of a rapport-based approach to investigative interviews: A training study with law enforcement investigators. Law and Human Behavior, 45(1), 55–67. https://doi.org/10.1037/lhb0000437
- Catlin, M., Wilson, D. B., Redlich, A. D., Bettens, T., Meissner, C. A., Bhatt, S., & Brandon, S. E. (2024). Interview and interrogation methods and their effects on true and false confessions: A systematic review update and extension. Campbell Systematic Reviews, 20(4), e1441. https://doi.org/10.1002/cl2.1441
- Cleary, H. M. D., & Warner, T. C. (2016). Police training in interviewing and interrogation methods: A comparison of techniques used with adult and juvenile suspects. Law and Human Behavior, 40(3), 270–284. https://doi.org/10.1037/lhb0000175
- Horgan, A. J., Russano, M. B., Meissner, C. A., & Evans, J. R. (2012). Minimization and maximization techniques: Assessing the perceived consequences of confessing and confession diagnosticity. Psychology, Crime & Law, 18(1), 65–78. https://doi.org/10.1080/1068316X.2011.561801
- Kassin, S. M., & Kiechel, K. L. (1996). The social psychology of false confessions: Compliance, internalization, and confabulation. Psychological Science, 7(3), 125–128. https://doi.org/10.1111/j.1467-9280.1996.tb00344.x
- Kassin, S. M. (2005). On the psychology of confessions: Does innocence put innocents at risk? American Psychologist, 60(3), 215–228. https://doi.org/10.1037/0003-066X.60.3.215
- Kassin, S. M., Leo, R. A., Meissner, C. A., Richman, K. D., Colwell, L. H., Leach, A.-M., & La Fon, D. (2007). Police interviewing and interrogation: A self-report survey of police practices and beliefs. Law and Human Behavior, 31(4), 381–400. https://doi.org/10.1007/s10979-006-9073-5
- Kassin, S. M. (2008). The psychology of confessions. Annual Review of Law and Social Science, 4, 193–217. https://doi.org/10.1146/annurev.lawsocsci.4.110707.172410
- Kassin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., & Redlich, A. D. (2010). Police-induced confessions: Risk factors and recommendations. Law and Human Behavior, 34(1), 3–38. https://doi.org/10.1007/s10979-009-9188-6
- Kassin, S. M., Cleary, H. M. D., Gudjonsson, G. H., Leo, R. A., Meissner, C. A., Redlich, A. D., & Scherr, K. C. (2025). Police-induced confessions, 2.0: Risk factors and recommendations. Law and Human Behavior, 49(1), 7–53. https://doi.org/10.1037/lhb0000593
- Kassin, S. M., Bogart, D., & Kerner, J. (2012). Confessions that corrupt: Evidence from the DNA exoneration case files. Psychological Science, 23(1), 41–45. https://doi.org/10.1177/0956797611422918
- Kassin, S. M., Meissner, C. A., & Norwick, R. J. (2005). “I’d know a false confession if I saw one”: A comparative study of college students and police investigators. Law and Human Behavior, 29(2), 211–227. https://doi.org/10.1007/s10979-005-2416-9
- Kassin, S. M., & Norwick, R. J. (2004). Why people waive their Miranda rights: The power of innocence. Law and Human Behavior, 28(2), 211–221. https://doi.org/10.1023/B:LAHU.0000022323.74584.F5
- Lassiter, G. D., Geers, A. L., Handley, I. M., Weiland, P. E., & Munhall, P. J. (2002). Videotaped interrogations and confessions: A simple change in camera perspective alters verdicts in simulated trials. Journal of Applied Psychology, 87(5), 867–874. https://doi.org/10.1037/0021-9010.87.5.867
- Leo, R. A. (2009). False confessions: Causes, consequences, and implications. Journal of the American Academy of Psychiatry and the Law, 37(3), 332–343. https://pubmed.ncbi.nlm.nih.gov/19767498/
- Meissner, C. A., & Kassin, S. M. (2002). “He’s guilty!”: Investigator bias in judgments of truth and deception. Law and Human Behavior, 26(5), 469–480. https://doi.org/10.1023/A:1020278620751
- Meissner, C. A., Redlich, A. D., Michael, S. W., Evans, J. R., Camilletti, C. R., Bhatt, S., & Brandon, S. E. (2014). Accusatorial and information-gathering interrogation methods and their effects on true and false confessions: A meta-analytic review. Journal of Experimental Criminology, 10(4), 459–486. https://doi.org/10.1007/s11292-014-9207-6
- Mindthoff, A., Ferreira, P. A., & Meissner, C. A. (2024). The effect of confession evidence on jurors’ verdict decisions: A systematic review and meta-analysis. Law and Human Behavior, 48(3), 163–181. https://doi.org/10.1037/lhb0000563
- Narchet, F. M., Meissner, C. A., & Russano, M. B. (2011). Modeling the influence of investigator bias on the elicitation of true and false confessions. Law and Human Behavior, 35(6), 452–465. https://doi.org/10.1007/s10979-010-9257-x
- Perillo, J. T., & Kassin, S. M. (2011). Inside interrogation: The lie, the bluff, and false confessions. Law and Human Behavior, 35(4), 327–337. https://doi.org/10.1007/s10979-010-9244-2
- Redlich, A. D., & Goodman, G. S. (2003). Taking responsibility for an act not committed: The influence of age and suggestibility. Law and Human Behavior, 27(2), 141–156. https://doi.org/10.1023/A:1022543012851
- Russano, M. B., Meissner, C. A., Narchet, F. M., & Kassin, S. M. (2005). Investigating true and false confessions within a novel experimental paradigm. Psychological Science, 16(6), 481–486. https://doi.org/10.1111/j.0956-7976.2005.01560.x
- Scherr, K. C., Redlich, A. D., & Kassin, S. M. (2020). Cumulative disadvantage: A psychological framework for understanding how innocence can lead to confession, wrongful conviction, and beyond. Perspectives on Psychological Science, 15(2), 353–383. https://doi.org/10.1177/1745691619896608
- Snook, B., Barron, T., Fallon, L., Kassin, S. M., Kleinman, S., Leo, R. A., Meissner, C. A., Morello, L., Nirider, L. H., Redlich, A. D., & Trainum, J. L. (2021). Urgent issues and prospects in reforming interrogation practices in the United States and Canada. Legal and Criminological Psychology, 26(1), 1–24. https://doi.org/10.1111/lcrp.12178
- Steering Committee of Experts. (2021). Principles on effective interviewing for investigations and information gathering. Association for the Prevention of Torture, Anti-Torture Initiative, & Norwegian Centre for Human Rights. https://www.apt.ch/sites/default/files/publications/apt_PoEI_EN_11.pdf
- Stewart, J. M., Woody, W. D., & Pulos, S. (2018). The prevalence of false confessions in experimental laboratory simulations: A meta-analysis. Behavioral Sciences & the Law, 36(1), 12–31. https://doi.org/10.1002/bsl.2327