RESEARCH DOSSIER
Criminal Responsibility and Retrospective Mental-State Assessment
Abstract. Criminal responsibility and retrospective mental-state assessment concern the reconstruction of psychological functioning at the time of an alleged criminal offense and its relationship to jurisdiction-specific legal criteria. This fundamental research monograph traces the history of insanity doctrine, differentiates criminal mens rea from exculpatory incapacity, examines psychotic and substance-related conditions, and evaluates source provenance, timeline reconstruction, structured forensic measures, symptom validity, interrater reliability, memory science, adversarial bias, and the law of expert testimony. Its method is a critical integrative narrative synthesis of original studies, professional guidelines, and primary U.S. legal authorities. Field research on legal sanity evaluations documents substantial expert disagreement; psychometric research provides more limited evidence of structured agreement under particular conditions. The synthesis establishes that diagnostic findings, historical reconstruction, functional inference, and legal responsibility must be assessed as distinguishable stages. It specifies methodological priorities for transparent, reproducible, and culturally responsive evaluations and discusses the prospective research role of artificial systems without treating theoretical possibilities as validated forensic tools.
Keywords: criminal responsibility; insanity defense; retrospective mental-state assessment; mens rea; forensic psychiatry; forensic psychology; criminal culpability; psycholegal assessment; RCRAS; MSE-Offense; legal sanity; expert reliability; source monitoring; malingering; forensic evidence; Rule 702; Rule 704(b); Artificial Era
Scientific Scope: Responsibility as a Retrospective Psycholegal Question
Criminal responsibility concerns the conditions under which a legal system attributes criminal culpability to an individual for conduct occurring at a particular time. Retrospective mental-state assessment investigates the psychological capacities, disturbances, experiences, and functional impairments that may have existed at the time of an alleged offense. The evaluative target is historically indexed: the relevant question concerns a past psychological state under a particular legal rule. A diagnosis made during trial preparation, the seriousness of an offense, or an evaluator's present impression cannot establish that past state on its own. The scientific task is to reconstruct a time-specific account from records, contemporaneous observations, clinical history, interviews, and competing explanatory hypotheses. The legal task is performed by the authorized decision-maker according to the applicable statute and burden of proof. AAPL (2014) 18 U.S.C. § 17
The subject joins three inferential domains. Clinical description addresses symptoms, syndromes, course, and differential explanations. Psychological reconstruction asks which beliefs, perceptions, intentions, control capacities, and understandings were likely active at the relevant moment. Legal attribution asks whether the reconstructed features meet a jurisdiction's threshold for an excuse, a culpable mental element, or a different legally defined consequence. These questions require coordination, because a sound diagnosis may have little bearing on a particular legal criterion and an observed act may admit several psychological interpretations. The evaluation is strongest when it demonstrates exactly which observations support each step from condition to functional impact to legally relevant question. APA (2013) Clark v. Arizona (2006)
This monograph is a critical integrative narrative synthesis of primary legal texts, forensic professional guidance, psychometric investigations, observational field research, and cognitive psychology. It examines U.S. doctrine in detail because the principal primary authorities available here are American; comparative concepts are considered at the level of defined legal traditions rather than presumed uniform international rules. The research is neither a preregistered systematic review nor a new clinical study. The central questions concern construct validity, methods of reconstruction, evidence reliability, diagnostic-to-functional inference, evaluator disagreement, adversarial bias, and transparent communication. No individual forensic opinion or clinical determination is rendered. AAPL (2014) Gowensmith et al. (2013)
The Legal Architecture of Responsibility and Exculpation
Criminal law ordinarily requires proof of legally defined conduct and the applicable mental element, often expressed through categories such as purpose, knowledge, recklessness, or negligence. An insanity defense raises a different question: whether a mental disease or defect affected capacities that the governing law treats as exculpatory even if the elements of the offense were otherwise established. The relation between these inquiries changes with the jurisdiction. Separating the mental element of a charged offense from legal insanity prevents the same phrase, mental state, from carrying two incompatible meanings in the same evaluation. In an opinion, each question should be tied to the specific legal language under examination. Clark v. Arizona (2006) Kahler v. Kansas (2020)
The federal statute, 18 U.S.C. § 17, establishes an affirmative defense when, because of a severe mental disease or defect, the defendant was unable at the time of the acts to appreciate their nature and quality or wrongfulness. It assigns the burden of proving that defense to the defendant by clear and convincing evidence. This is the federal rule, and state jurisdictions may use materially different definitions and allocations. A forensic author should therefore distinguish a scientifically describable impairment from the court's legal conclusion, especially where a statutory threshold uses terms such as severe, appreciate, or wrongfulness that require contextual interpretation. 18 U.S.C. § 17 AAPL (2014)
Related legal constructs include diminished capacity, diminished actuality, intoxication provisions, sentencing mitigation, automatism, and capacity to participate in proceedings. Their labels and admissibility vary. Present competence to stand trial asks whether a defendant can participate in the contemporary legal process, whereas retrospective responsibility addresses a prior alleged act. The temporal distinction affects evidence: improvement in a person's symptoms before trial may change competence without answering the earlier question; similarly, present incompetence does not establish prior insanity. A complete forensic referral identifies the law, offense dates, charged elements, and requested evaluative question before interviewing begins. Federal Rule of Criminal Procedure 12.2 AAPL (2014)
Historical Formation: From M'Naghten to Modern Standards
Legal attempts to accommodate mental disturbance in culpability have developed through different cognitive, moral, and volitional conceptions. The nineteenth-century M'Naghten formulation became a prominent reference for whether a person knew the nature and quality of the act or knew it was wrong. Its influence can be traced in modern statutory language, although its historical role does not entail that every U.S. state now applies one uniform test. An account of the history must therefore track changes in the law as well as changes in psychiatric vocabulary. The legal meaning of knowing wrongfulness and the clinical meaning of a delusional belief have evolved through partially independent traditions. Kahler v. Kansas (2020) Clark v. Arizona (2006)
Twentieth-century approaches also considered mental disease in relation to impaired behavioral control and broader accounts of culpability. Such alternatives foregrounded the difficulty of turning a clinical description into a threshold for legal excuse. The reform after the 1982 Hinckley verdict is especially consequential in federal practice. Congress enacted the Insanity Defense Reform Act in 1984, reshaping the federal standard and imposing the burden described in § 17. The same legislative period produced the special limitation on expert opinions about ultimate mental-state issues in Federal Rule of Evidence 704(b). Historical explanation must keep these policy choices distinct from empirical claims about how accurately clinicians diagnose disorders. 18 U.S.C. § 17 Federal Rule of Evidence 704 Diaz v. United States (2024)
The Supreme Court's decisions in Clark v. Arizona (2006) and Kahler v. Kansas (2020) demonstrate continuing jurisdictional variation. Kahler rejected the proposition that due process requires states to adopt a specific moral-incapacity form of the insanity defense. In other words, the constitutional decision did not establish the scientific truth of one psychological model. It confirmed substantial latitude for legal definitions of excuse. Forensic research consequently needs two parallel histories: the development of clinical methods for studying disorders and retrospective functioning, and the development of normative legal criteria deciding which impairments matter to criminal responsibility. Clark v. Arizona (2006) Kahler v. Kansas (2020)
Construct Definitions and Operational Questions
The phrase mental state may identify a conscious experience, a belief, an intention, a perceived threat, a disorder-related symptom, or a legally specified form of culpability. These are distinct constructs. A hallucination is a perceptual experience; a delusion is a belief-like symptom with specific clinical characteristics; appreciation of wrongfulness is a legally defined functional issue. A forensic conclusion cannot proceed by relabeling one construct as another. The initial analytic move is to translate the referral into observable and historically investigable questions: what the person appeared to perceive, what contemporaneous statements conveyed, how conduct unfolded, and what contextual facts support or challenge a hypothesized impairment. AAPL (2014) APA (2013)
Operationalization also requires attention to temporal scale. A chronic psychotic disorder describes a clinical course over months or years, while legal responsibility may depend on seconds or minutes. An acute episode, substance exposure, sleep loss, or delirious period may produce a sharper fluctuation than a stable diagnostic label conveys. Each claim must identify its time window, supporting evidence, and uncertainty. Time-specific reasoning should avoid treating a single fragment of unusual behavior as if it revealed the entire psychological organization of an offense. AAPL (2014)
Functional appraisal can be divided for analytic purposes into knowledge of physical circumstances, interpretation of social meaning, appreciation of consequences and wrongfulness, and behavioral capacities relevant to the applicable rule. These domains should be measured or reconstructed separately and joined only where a law requires a combined conclusion. Their distinction helps explain why two people with similar diagnoses might have different legally relevant capacities. An evaluation should explicitly address plausible ordinary motives, situational pressures, planning, concealment, impulsivity, and symptom-driven explanations without assuming that any one observable behavior has a unique psychological cause. AAPL (2014) Clark v. Arizona (2006)
Research Methods and the Hierarchy of Evidence
Retrospective criminal-responsibility research cannot reproduce an individual's past experience in a laboratory. It therefore operates through historical inference. Its strongest evidence frequently includes records produced near the relevant time, independently documented treatment history, verified contemporaneous communications, and behavioral observations whose origin can be established. Later statements, retrospective recollection, and present psychological examination can add explanatory information but require separate source evaluation. The inferential question is how each piece changes the relative support for competing explanations. Multiple accounts copied from one original witness or interviewer do not constitute independent corroboration. AAPL (2014) APA (2013)
Different research designs answer different questions. A psychometric cross-validation study can show whether evaluators using a structured instrument agree in a particular sample; it cannot demonstrate that every historical judgment is correct. A routine-practice record study can reveal agreement and disagreement in field conditions; it does not experimentally identify the causal origin of every disagreement. An experiment involving fictional case files can isolate contextual influences, but its outcome may not generalize directly to an actual defendant and legal threshold. These boundaries are central to interpreting the original work of Rogers and colleagues, Gowensmith and colleagues, and Murrie and colleagues. Rogers et al. (1983) Gowensmith et al. (2013) Murrie et al. (2013)
Evidence appraisal therefore includes sampling, referral selection, blinding, evaluator independence, legal jurisdiction, base rates, record completeness, test calibration, diagnostic consistency, and uncertainty in the criterion used to label an evaluation correct. A finding of agreement should be accompanied by the actual denominator and decision categories. Classification accuracy cannot be assessed by treating another clinician's opinion as an unquestioned ground truth. When a clinical outcome, a legal verdict, and a psychometric score diverge, investigators must specify which outcome their method purports to predict. Rogers and Shuman (2000) Federal Rule of Evidence 702
The Retrospective Timeline as a Scientific Instrument
A coherent reconstruction begins with a dated sequence of conditions, observations, and actions before, during, and after the alleged offense. The relevant chronology includes prodromal experiences, contact with clinicians, hospitalization, medication changes, intoxication or withdrawal where documented, interpersonal conflicts, environmental stressors, and contemporaneous communication. For each event the investigator records the primary source, distance from the offense, potential dependence on other reports, and uncertainty in timing. A timeline is an analytic device that makes temporal leaps visible and can reveal when a seemingly convincing narrative rests on evidence acquired only after the outcome became known. AAPL (2014) APA (2013)
Time proximity increases the potential relevance of an observation without guaranteeing its accuracy. An emergency note may capture acute behavior while omitting a detailed explanation; a later interview may describe motives but incorporate subsequent events, questions, legal advice, and reinterpretations. Official records also differ in purpose. A police report focuses on investigative facts, hospital notes on care and risk, and a family statement on observed conduct and prior relationships. No source should be promoted to a complete description of consciousness merely because it was produced in a professional institution. A rigorously reconstructed chronology includes discrepancies rather than silently resolving them. Johnson et al. (1993) AAPL (2014)
The difference between an event, the first record of an event, and the first later interpretation of that record is particularly important. If a defendant is described after arrest as delusional, the investigator should identify when the symptom was first observed, how it was assessed, and whether it preceded the charged behavior. If the same expression appears in several interviews, the investigator should consider whether it was repeated from a common source. Such provenance analysis is a form of source criticism. It converts the timeline from a persuasive story into a set of testable historical propositions. Johnson et al. (1993) APA (2013)
Clinical Disorders, Diagnostic Reasoning, and Legal Relevance
Psychotic disorders can affect perception, belief formation, interpretation of threat, and attribution of agency. Their forensic relevance depends on the nature and timing of symptoms and the relation between those symptoms and the legally relevant capacities at the event. A hallucination documented on one date does not itself prove an impairment at another. A diagnosis of schizophrenia does not settle whether the person understood a particular physical action or appreciated its wrongfulness. A clinically sound reconstruction compares illness course, treatment response, contemporaneous functioning, symptom consistency, and competing explanations for the conduct under review. AAPL (2014)
Mood episodes, dissociative phenomena, cognitive impairment, neurodevelopmental conditions, delirium, and substance-related states may also be relevant to a referral, depending on the governing law. Each calls for a different pattern of corroboration and a different differential diagnosis. For example, an acute confusional presentation raises questions about fluctuating attention and medical precipitants, whereas a reported delusion requires evaluation of its content, conviction, duration, and functional influence. Clinically related conditions need not have the same legal significance. A report should distinguish established observations from a retrospective diagnosis and from the further inference about statutory capacity. AAPL (2014) APA (2013)
Diagnostic certainty and legal certainty are different. A person might satisfy clinical criteria for a disorder while the available records leave the offense-time functional question unresolved. Conversely, observations may establish a marked contemporaneous impairment while diagnostic classification remains disputed. These possibilities become visible when an evaluator reports separate conclusions for evidence of condition, evidence of timing, evidence of causal-functional relationship, and the fit with the applicable criterion. The central professional guidance supports a multi-source evaluation rather than a purely diagnosis-driven conclusion. AAPL (2014) APA (2013)
Psychosis, Delusional Meaning, and Action Interpretation
Unusual beliefs become forensically relevant through their specific behavioral and decision context. A person who reports believing that another individual posed an imminent threat may have held an ordinary mistaken belief, a culturally mediated interpretation, an exaggerated but reality-based concern, or a delusion associated with psychopathology. The investigator must first establish the content and provenance of the asserted belief and then ask what evidence connects it to perception, choice, and behavior at the relevant time. A retrospective statement of subjective certainty cannot, by itself, establish the historical truth of the claimed experience. AAPL (2014) Johnson et al. (1993)
Appreciation is a useful analytic term because it may concern the ability to meaningfully grasp consequences or wrongfulness rather than merely repeat a rule. Yet appreciation is defined within the governing legal standard. The clinician's task is to present evidence on cognitive and experiential capacities with sufficient detail to make the inferential relation intelligible. Expressions of fear, anger, conviction, or even a clinically established delusion should be examined in their functional setting. Where the individual took deliberate steps to conceal an act, the significance of concealment requires analysis of alternative purposes; there is no simple one-to-one mapping from concealment to full capacity. 18 U.S.C. § 17 Clark v. Arizona (2006)
Retrospective evaluation also faces a counterfactual problem: an evaluator is asked whether a mental condition rendered a legally relevant capacity unavailable when the underlying experience cannot be directly observed. Evidence can support a plausible account of what the person perceived, knew, or could appreciate, but it generally cannot reconstruct a complete stream of consciousness. Explicitly separating direct observations, self-report, corroboration, clinical explanation, and legal significance makes the uncertainty auditable. AAPL (2014) Federal Rule of Evidence 702
Substance Exposure, Intoxication, and Mixed States
Substance exposure may change perception, attention, memory, impulse regulation, and psychotic symptoms. Whether voluntary intoxication is legally relevant to a charged element or an insanity defense depends on the jurisdiction and statute. Clinical classifications such as substance-induced psychosis, intoxication, withdrawal, and a co-occurring primary disorder should be evaluated through their actual temporal patterns and records. The question is not resolved by a generic statement that the defendant was under the influence. Quantity, timing, biological data, observed behavior, and an appropriate differential formulation may all matter, while each can remain incomplete. AAPL (2014) Clark v. Arizona (2006)
The empirical importance of these complexities appears in the field reliability study of Gowensmith, Murrie, and Boccaccini. Its abstract reports that evaluators disagreed more often when the defendant had been under the influence of alcohol or drugs at the offense. This is a descriptive association within the studied forensic evaluation system; it does not establish that intoxication causes particular evaluators to be inaccurate or that the same relationship holds in every jurisdiction. The result motivates more explicit measurement of substance-related uncertainty and distinctions among symptoms, exposure history, and legal standards. Gowensmith et al. (2013)
The investigator should consider whether relevant symptoms predated exposure, persisted after expected acute effects, appeared during clinically documented withdrawal, or coincided with an unrelated condition. Self-reported dosage or timing may conflict with laboratory tests or witness descriptions. Each conflict should be handled as an evidence problem rather than resolved by a presumed pattern. Where substance data cannot sustain a reliable reconstruction, the report should state the boundary of inference and avoid filling gaps with generic pharmacological claims. AAPL (2014) Federal Rule of Evidence 702
Memory Reconstruction, Source Monitoring, and Hindsight
Retrospective assessment depends partly on memory, a constructive cognitive process. Johnson, Hashtroudi, and Lindsay's source-monitoring framework explains how people distinguish memories arising from perception, imagination, conversation, and other origins, and why those distinctions may fail under some conditions. This work provides a theoretical reason to ask where details in a later offense narrative came from. It does not establish that every inconsistency is fabrication or that the evaluator can infer a precise historical probability of a specific memory being accurate. Johnson et al. (1993)
Postevent interviews can transform how a person interprets prior motives, experiences, and beliefs. Later medical diagnoses, family discussions, police questions, legal hypotheses, and access to media may create information that is then incorporated into retrospective accounts. A forensic reconstruction should compare the earliest available narrative with later retellings and identify questions or suggestions introduced between them. When the early account lacks a detail later presented as central, this discrepancy invites analysis of both possible memory development and genuine subsequent clarification. There is no universally valid rule that the first statement is invariably the most accurate. Johnson et al. (1993) APA (2013)
Hindsight bias arises when knowledge of an outcome makes preceding events appear more foreseeable or causally unified. In an offense evaluation, the seriousness of the result may influence how earlier symptoms, conflicts, or threats are interpreted. A disciplined evidentiary sequence attempts to reconstruct what was known at each moment before treating later outcomes as explanatory. Evaluators can document what changed when new materials were reviewed, distinguish offense facts from clinical observations, and test whether their formulation would have remained plausible without knowledge of the ultimate charge. These are methodological safeguards rather than a guarantee of unbiased reconstruction. Murrie et al. (2013) Federal Rule of Evidence 702
Collateral Records, Witnesses, and Independent Corroboration
Records from treating clinicians, emergency personnel, acquaintances, and police may add observations unavailable in a later interview. Their value depends on what a source actually observed, when it observed it, and what purpose the record served. A clinician who examined the person shortly after an incident may provide unusually informative signs, but that clinician might never have assessed appreciation of legal wrongfulness. A witness may accurately describe speech or conduct while making a diagnostic inference outside the witness's expertise. Reports should preserve these levels of description. AAPL (2014) APA (2013)
Corroboration requires attention to independence. If three witnesses discussed the incident before interviews, their matching descriptions may reflect shared exposure. If a police summary and a hospital intake repeat a defendant's statement, they are two documents but possibly one informational origin. The scientific question is whether the evidence paths are sufficiently independent to raise confidence in the reconstructed event. A source matrix can document origin, timestamp, directness, dependency, contradictions, and legal relevance without assuming every source deserves equal weight. Johnson et al. (1993) APA (2013)
Conflicting collateral evidence should be preserved in the substantive analysis. For example, a report of bizarre statements during the days before an event may coexist with descriptions of ordinary work performance and a coherent financial plan. Each observation has a time frame and domain; neither automatically eliminates the other. A valid formulation explains what the pattern supports, which historical interpretations remain possible, and where additional information would most affect the conclusion. The ethical expectation of using multiple data sources and disclosing uncorroborated claims is explicit in forensic psychology guidance. APA (2013) AAPL (2014)
Structured Forensic Assessment and the RCRAS Tradition
The Rogers Criminal Responsibility Assessment Scales (RCRAS) emerged from efforts to make insanity evaluations more structured and researchable. Rogers, Dolmetsch, and Cavanaugh described their empirical approach in 1981 using case vignettes and an initial patient sample, thereby bringing issues of decision rules and interrater reliability into a domain often dominated by unstructured judgment. The point of a structured measure is to make observations and evaluative judgments more explicit. It cannot repeal the governing legal standard or replace professional analysis of the case record. Rogers et al. (1981)
Rogers, Seman, and Wasyliw reported a 1983 cross-validation of the RCRAS using 111 cases from five forensic centers. The PubMed abstract reports a mean kappa of .80 for decision variables related to criminal responsibility and differentiating patterns on four of five scales between groups designated sane and insane. These numbers describe agreement and group discrimination under the study's methods. They should not be translated into a universal 80% probability that an RCRAS-based insanity conclusion is factually correct. The criterion and case mix require attention, and later settings may yield different reliability. Rogers et al. (1983)
Structured scoring creates opportunities for standardized training, audit trails, and comparison among evaluators. However, historical information often remains incomplete, the meaning of scores may depend on jurisdictional thresholds, and a measure's apparent accuracy may be inflated by circular comparison with clinical opinions resembling the criterion. A credible evaluation reports precisely which domains an instrument samples, which population supported its validation, which aspects concern current symptoms rather than offense-time functioning, and which conclusions rest on independent evidence outside the test. Rogers et al. (1983) Federal Rule of Evidence 702
The MSE-Offense Debate and Instrument Validity
The Mental State at the Time of the Offense measure (MSE-Offense) became the subject of explicit methodological dispute over whether a screening procedure could be given a determinative role in insanity evaluations. Rogers and Shuman examined its development and validation in a 2000 critique and concluded that serious shortcomings made the measure unsuitable, in their analysis, for screening or determining criminal responsibility under Daubert. This is a published critical assessment of the available validation, not a new independent outcome study and not an automatically binding legal ruling on every possible use. Rogers and Shuman (2000) Daubert (1993)
The debate illustrates a central scientific distinction between face plausibility and validated inference. A tool can appear to ask relevant historical questions but still fail to show that its items, scoring rules, and thresholds discriminate among legally consequential alternatives with acceptable accuracy. Classification may also change as a consequence of who is referred, the proportion of severe disorders in the evaluation sample, and the definitions used for favorable and unfavorable outcomes. A tool validated for describing symptoms cannot acquire validity for identifying statutory incapacity through terminology alone. Rogers and Shuman (2000) APA (2013)
An admissibility-oriented scientific assessment requires transparent evidence about construction, scoring reliability, intended use, comparison criteria, validation groups, and known limitations. Rule 702 asks for sufficient facts, reliable methods, and reliable application; the 2023 revision emphasizes the proponent's burden to demonstrate these requirements by a preponderance and the obligation to keep opinions within evidentiary bounds. A scientifically cautious researcher therefore treats instruments as aids to a differentiated explanation and specifies the inferential steps they can and cannot support. Federal Rule of Evidence 702 Rogers and Shuman (2000)
Symptom Validity, Feigning, and Genuine Complexity
The adversarial context can create incentives for inaccurate symptom reporting, yet such incentives do not establish that a defendant has fabricated symptoms. Forensic interviewing must distinguish deliberate feigning, exaggeration, misunderstanding, memory failure, variable symptom expression, mistrust, and the effects of actual psychiatric disturbance. An inconsistent narrative may have several causes. The scientific problem is to examine response patterns with tools validated for the relevant population and to integrate their results with longitudinal and collateral records. The risk of false positive labeling is especially consequential in a legal context. APA (2013) Wygant et al. (2022)
Research on the Structured Interview of Reported Symptoms (SIRS) and its successor SIRS-2 contributes to the measurement of response style. Rogers and colleagues' 1992 study used an experimental approach to the feigning of particular disorders. Wygant and colleagues' 2022 review discusses the SIRS-2's forensic use and admissibility. These works support the need for psychometric scrutiny, while the existence of a validity instrument does not license an inference from an anomalous score to guilt, conscious deception in a specific account, or a definitive judgment of criminal responsibility. Rogers et al. (1992) Wygant et al. (2022)
Research must also consider genuine symptoms among people who fail validity indicators. Classification errors can occur when distress, rare presentations, neurocognitive limitations, or unexpected response patterns resemble indicators associated with feigning. A transparent report states the exact interpretation permitted by the instrument, the quality of the norms, and how case-specific findings converge or diverge. A finding of questionable symptom validity can appropriately reduce the weight placed on a particular self-report without conclusively erasing independent evidence of past impairment. Wygant et al. (2022) APA (2013)
Interrater Agreement and the Limits of Expert Consensus
Agreement among professionals is important because a forensic opinion should possess an identifiable degree of reproducibility. Gowensmith, Murrie, and Boccaccini examined 483 reports concerning 165 defendants in Hawaii, where multiple independent evaluations allowed an unusually direct study of field reliability. Their 2013 article reports unanimous agreement about legal sanity in 55.1% of cases. The result concerns unanimity across the particular evaluators and cases; it is not an estimate of population-level diagnostic prevalence or an accuracy rate against an independently known historical truth. Gowensmith et al. (2013)
In that study, disagreement was more common when defendants had been intoxicated, while agreement was more common when evaluators concurred on a psychotic-disorder diagnosis or when hospitalization occurred shortly before the alleged offense. Judges followed the majority evaluative opinion in 91% of cases, according to the paper's abstract. This relationship is relevant to the institutional influence of forensic expertise, but judicial agreement with a majority opinion is not independent validation of psychological truth. The published findings motivate examination of how records, diagnostic judgments, and legal interpretations separately contribute to divergent reports. Gowensmith et al. (2013)
Reliability can be expressed at multiple levels: agreement on observable symptoms, diagnostic classification, causal relevance, functional impairment, and ultimate legal conclusion. These levels need not covary. Two clinicians may agree that a person was psychotic and disagree whether the symptoms met a jurisdiction's moral-incapacity criterion. Future research should collect itemized reasoning and common evidence inputs instead of only binary sanity outcomes, permitting investigation of where disagreement originates. Such research would improve the interpretability of field agreement without presupposing that uniformity is always synonymous with correctness. Gowensmith et al. (2013) Rogers et al. (1983)
Adversarial Allegiance and Contextual Bias
Expert judgments can be shaped by information that is legally relevant, scientifically relevant, or merely contextually persuasive. In a controlled experiment, Murrie, Boccaccini, Guarnera, and Rufino paid 108 forensic psychologists and psychiatrists to review offender files while manipulating whether they believed they were retained by prosecution or defense. Experts who believed they worked for the prosecution tended to assign higher scores on risk-assessment tools than experts who believed they worked for the defense; reported effect sizes reached d = 0.85. These findings demonstrate an allegiance effect in the specific experimental risk-assessment task, not direct proof of the same effect magnitude in insanity opinions. Murrie et al. (2013)
The transfer to retrospective responsibility assessment is a credible research question because forensic evaluators also operate in adversarial settings, encounter case theories supplied by attorneys, and may be exposed to strongly framed allegations. However, the scientific argument should distinguish demonstrated experimental bias from an untested extrapolation. An evaluator can reduce the opportunity for avoidable contextual influence by identifying the legal question independently, recording initial hypotheses, separating observation from allegation, and considering disconfirming material as actively as supporting material. These practices are consistent with the APA's impartiality guidance. APA (2013) Murrie et al. (2013)
Bias controls should be evaluated by performance rather than ceremonial statements of neutrality. Prospective studies could randomize case framing, sequence of irrelevant information, and structured alternative-hypothesis prompts, measuring changes in ratings and reasoning. The pertinent outcomes include calibration, interrater agreement, error rates against suitable reference cases, and completeness of disclosed uncertainty. Successful blinding should preserve access to genuinely necessary case facts while preventing extraneous contextual cues from contaminating clinical interpretation. Federal Rule of Evidence 702 APA (2013)
Population Research on the Insanity Defense
Public discussion often treats the insanity defense as a common route to avoiding conviction, but early empirical studies reported a more limited frequency in the sampled jurisdictions. Callahan, Steadman, McGreevy, and Robbins investigated pleas in eight U.S. states and found the defense raised in approximately 1% of felony cases in their sampled data, with 26% of those pleas resulting in not-guilty-by-reason-of-insanity acquittal. The methodological challenge was considerable: identifying actual pleas required differing case-finding methods across courts. The figures belong to the specific historical sample and cannot be asserted as present-day national rates. Callahan et al. (1991)
Janofsky, Vandewalle, and Rappeport studied a Baltimore court cohort over a twelve-month period. Their abstract records 143 insanity pleas among 11,497 indicted defendants, with fourteen later acquitted on that basis. This local cohort illustrates variation in legal processes and emphasizes that a plea, an expert's position, a contested jury trial, and the final legal outcome are different events. A legal system may resolve many questions through agreement or procedural dispositions that cannot be inferred from a courtroom verdict count alone. Janofsky et al. (1989)
Cirincione, Steadman, and McGreevy subsequently examined success rates in seven states and identified differences associated with the frequency of invoking the defense. Together these studies show that legal outcomes are functions of statutory rules, legal representation, case selection, diagnostic evidence, local institutions, and evidentiary judgments. A scientific synthesis should resist converting an historical aggregate into a claim that a particular defendant is likely to succeed or that expert evidence is uniformly determinative. Modern surveillance would require current jurisdiction-specific denominators and comparable case definitions. Cirincione et al. (1995) Callahan et al. (1991)
Legal Evidence, Daubert, and Rule 702
Scientific testimony becomes legally consequential only through rules governing relevance, qualification, methodology, and admissibility. Daubert v. Merrell Dow Pharmaceuticals (1993) articulated a judicial gatekeeping role for scientific expert testimony under federal evidentiary rules. Federal Rule of Evidence 702 now expressly requires a sufficient factual basis, reliable principles and methods, and a reliable application to the case, with the proponent demonstrating admissibility conditions as more likely than not. These requirements matter particularly when a retrospective assessment depends on expert interpretation of documents that cannot themselves be experimentally repeated. Daubert (1993) Federal Rule of Evidence 702
A court's admissibility determination and an evaluator's scientific conclusion answer distinct questions. Expert evidence can satisfy an admissibility threshold while remaining subject to disagreement and uncertainty. Equally, a widely used instrument can be misapplied in a particular population or to a legal construct for which it has not been validated. The 2023 Rule 702 amendment stresses that the opinion itself must remain within the limits of its methodological support. This makes transparent conditional reasoning valuable: which assumptions must be true for the conclusion to follow, what evidence supports them, and how the conclusion would change if contested facts were resolved differently? Federal Rule of Evidence 702 Rogers and Shuman (2000)
Federal and state evidentiary frameworks may differ, and scientific quality does not eliminate the need to check the applicable forum. A useful forensic report identifies the actual legal authority, date of the relevant provision, intended role of the expert, and known boundaries of the opinion. In a retrospective assessment it can be appropriate to say that the record supports evidence of specific symptoms and associated capacities without representing that evidence as a judicial finding of innocence or guilt. The distinction makes a report more accurate and more useful to the factfinder. APA (2013) Federal Rule of Evidence 702
Ultimate-Issue Testimony and the Role of the Factfinder
Federal Rule of Evidence 704(b) provides that an expert in a criminal case must not state an opinion on whether the defendant had a mental state or condition that is an element of the offense or defense. The factfinder retains that ultimate decision. The rule does not prevent all testimony concerning psychological symptoms, behavioral processes, or the general meaning of clinical information. It requires attention to the form and object of expert inference, especially where a clinically phrased conclusion would effectively answer a legally reserved question. Federal Rule of Evidence 704
In Diaz v. United States (2024), the Supreme Court addressed the scope of Rule 704(b) in a drug-transportation prosecution. It held that an expert's general statement concerning what most members of a group knew did not constitute the prohibited opinion about the defendant's own mental state. Diaz concerned knowledge in a criminal offense rather than an insanity assessment, so it should be used as an interpretation of the evidentiary rule, not as experimental evidence about clinical reconstruction. Its broader lesson is that the boundary between population-level testimony and defendant-specific opinion has substantive consequences. Diaz v. United States (2024)
Clear scientific reporting separates observed data, clinical formulation, probabilistic or conditional explanation, and legally reserved findings. An expert might explain documented disorganization, alternative clinical mechanisms, or what a symptom pattern can imply under specified circumstances, subject to local evidence rules. Whether the available information establishes an element or defense remains for the tribunal under the relevant law. This structure can preserve substantive psychological assistance while recognizing the institutional responsibility assigned to the judge or jury. Federal Rule of Evidence 704 Diaz v. United States (2024)
Procedural Rules and Examination Rights
Retrospective evaluations unfold in a legal process with notice, examination, confidentiality, and use restrictions. Federal Rule of Criminal Procedure 12.2 governs advance notice of an insanity defense and specified expert evidence concerning mental condition, and it provides mechanisms for court-ordered examinations. These provisions affect what information may be obtained and how it can be used. Their exact requirements should be consulted in the live rule and case-specific orders rather than reconstructed from general descriptions of forensic practice. Federal Rule of Criminal Procedure 12.2
Forensic interviews differ from therapeutic encounters in referral purpose, obligations, and expected disclosure. The psychologist or psychiatrist must communicate the nature of the examination, who requested it, and the applicable limits on confidentiality and voluntariness. The presence of legal counsel, interpreters, or recording arrangements can affect both procedural rights and interpretation of responses. Ethical guidance emphasizes informed explanation, competence, impartiality, documentation, and careful treatment of collateral materials. APA (2013)
These safeguards also concern research validity. A person who misunderstands the purpose of an interview may respond differently from one who knows that statements can enter a legal report. A translated question may alter a subtle distinction about belief, intent, or recollection. Interview conditions, interruptions, preparatory conversations, and changes in the examinee's current clinical state should therefore be recorded. A legally valid process and scientifically sound elicitation reinforce one another when the evaluation's evidentiary provenance is preserved. APA (2013) AAPL (2014)
Culture, Language, Disability, and Differential Interpretation
Psychiatric symptoms and accounts of action are described through language, culture, social experience, and institutional context. Beliefs concerning spiritual causation, threats, authority, or agency may be interpreted differently across communities. Cultural responsiveness requires investigation of shared meaning and contextual norms, together with clinical assessment of conviction, functional impairment, and deviations from the individual's background. The investigator should not infer pathology merely because a belief is unfamiliar, nor treat cultural description as a substitute for examining whether severe symptoms were present. APA (2013)
Language affects the reconstruction of intent, appreciation, and recall. A forensic interpreter may need to convey distinctions among knowing a rule, believing an act justified, feeling compelled, and being unable to appreciate its wrongfulness. If a psychometric measure lacks appropriate language or population validation, that limitation bears directly on the evidentiary weight assigned to the result. Disability and cognitive difficulties can affect comprehension, autobiographical recall, interview responses, and access to collateral records. The meaning of an apparently inconsistent answer should be assessed in light of these conditions. APA (2013) AAPL (2014)
Empirical fairness is a measurement question. Relevant work would compare error rates, interrater agreement, response-validity classifications, and functional interpretations across appropriately defined populations while accounting for selection and other confounders. Cases referred for insanity assessment may themselves reflect unequal patterns of access to care or expert resources. This possibility requires evidence rather than assumption. The scientific objective is a method that can explain its evidentiary basis and uncertainty in the person's actual linguistic and sociocultural context. APA (2013) Federal Rule of Evidence 702
Reports, Counter-Hypotheses, and Communication of Uncertainty
A rigorous forensic report contains a precise referral question, controlling legal standard, dates and source inventory, relevant history, findings from the current examination, available contemporaneous evidence, alternative explanations, and a reasoned account of how conclusions follow. The report also identifies evidence that could not be obtained. This structure makes the opinion reviewable and permits a decision-maker to separate symptoms from law. An organized narrative should not conceal significant contradictory facts in footnotes or replace explicit reasoning with authority derived from professional credentials. AAPL (2014) APA (2013)
Competing explanations should be genuinely differentiated. For example, an account of an offense may invoke active psychosis, intoxication, intense but nondelusional fear, instrumental planning, an inaccurate later memory, or several interacting conditions. The evaluator asks what each hypothesis predicts about documented speech, behavior, prior course, and subsequent observations. Evidence may be stronger for some propositions than others, and no candidate hypothesis should gain unwarranted certainty merely because it organizes the available facts elegantly. AAPL (2014) Johnson et al. (1993)
Uncertainty can be described in terms of evidence completeness, diagnostic uncertainty, time specificity, agreement among sources, instrument limitations, and sensitivity to disputed assumptions. The terms possible, probable, and supported should be used with stable meanings rather than as rhetorical substitutes for numbers not estimated. Transparent uncertainty serves legal reasoning because decision-makers can see the location of the inferential gap and its relevance to burdens of proof. A report can be scientifically valuable even when available evidence does not justify a categorical historical conclusion. Federal Rule of Evidence 702 APA (2013)
Critical Synthesis: What the Evidence Establishes
The strongest general conclusion is procedural: retrospective responsibility evaluations need explicit legal criteria, temporally anchored clinical data, multiple independent sources, differentiated functional inference, and disclosed uncertainty. Professional guidelines consistently require scientific competence and careful handling of collateral evidence. Primary legal authorities establish that the meaning of exculpatory incapacity is partly jurisdictional. The evidence does not provide a universal test that reads a defendant's past psychological state from a present examination or a diagnostic code. AAPL (2014) APA (2013) Kahler v. Kansas (2020)
Empirical research supplies both encouragement and warnings. The RCRAS cross-validation produced substantial agreement for selected decision variables in its sample, demonstrating the feasibility of structured comparison. The field study of legal sanity opinions documented far less frequent unanimity in routine multiprofessional evaluation, demonstrating the importance of context, construct definition, and task conditions. The MSE-Offense critique illustrates how apparently relevant measures can remain insufficiently validated for their intended forensic use. These studies answer related but different scientific questions, so their summary statistics should be interpreted within their original study designs. Rogers et al. (1983) Gowensmith et al. (2013) Rogers and Shuman (2000)
Further uncertainty concerns response validity, information contamination, adversarial allegiance, and cross-jurisdictional generalization. Controlled bias research demonstrates that misleading retention cues can shift judgments in risk assessment, making similar effects plausible targets for direct sanity-evaluation experiments. Source-monitoring research supplies mechanisms for memory confusion; it does not provide a case-specific truth detector. Together these results support research on transparent reconstruction and independent checking, rather than claims that any diagnostic category or single instrument can establish legal innocence or culpability. Murrie et al. (2013) Johnson et al. (1993) Wygant et al. (2022)
Research Priorities: Reliability, Causality, and Artificial Systems
Future studies should distinguish reliability of symptoms, diagnoses, functional findings, and ultimate legal conclusions. Multi-site designs with independently trained evaluators and shared, sufficiently realistic case files could identify how much variation arises from differences in evidence, legal interpretation, or clinical inference. Predefined outcomes and transparent uncertainty coding would permit meaningful comparison between structured and unstructured methods. Where a reference determination is available, its independence and limitations should be specified. Research should also evaluate whether improvements persist under changes in case severity, intoxication patterns, cultural background, and legal jurisdiction. Gowensmith et al. (2013) Rogers et al. (1983)
A separate experimental program can investigate source provenance, alternative-hypothesis procedures, and adversarial context. Randomly varying task-irrelevant information while holding legally relevant evidence constant may isolate the causal contribution of framing. Longitudinal work could examine whether statements produced immediately after an event differ systematically from later accounts and which procedural features predict greater agreement without treating early records as infallible. Replications should report how training, professional discipline, and case selection affect results. Murrie et al. (2013) Johnson et al. (1993)
Artificial systems now introduce additional research questions in document chronology, information retrieval, consistency checking, and decision support. A software tool can organize records while introducing transcription errors, hallucinated citations, hidden dependence among summaries, or overconfident explanations. Any use in legally consequential assessment requires evaluation of provenance, accuracy, cross-population performance, data protection, and human oversight; an algorithmic output is not a clinical examination or statutory finding. In the Artificial Era, this is a study of how Homo's legal institutions interact with Artificial cognitive systems, with claims of efficacy requiring independent empirical demonstration rather than assertion. Federal Rule of Evidence 702 APA (2013)
Conclusion
Criminal responsibility and retrospective mental-state assessment occupy a distinctive scientific and legal domain: a present evaluator must reason about historical psychological capacities under a rule whose meaning is established by law. The defensible unit of analysis is the complete evidentiary chain from historical source to clinical interpretation, functional assessment, and legally relevant proposition. That chain gains credibility through source independence, temporal precision, alternative-hypothesis testing, measurement validity, and transparent discussion of uncertainty. AAPL (2014) Federal Rule of Evidence 702
Historical doctrine explains why no single psychiatric concept corresponds to all legal tests of insanity. Contemporary federal law defines one standard; states may employ different thresholds. Empirical evidence establishes that evaluators can disagree substantially in ordinary practice and that structured measures require construct-specific validation. Cognitive memory research and experimental work on expert allegiance identify additional mechanisms that can influence retrospective interpretation. These findings support differentiated reasoning and replicable procedures rather than categorical inference from diagnosis alone. Kahler v. Kansas (2020) Gowensmith et al. (2013) Rogers et al. (1983) Murrie et al. (2013)
The enduring scientific task is to make past-state reconstruction reviewable: establish what was observed, when it occurred, how the information traveled, what mechanisms could explain it, and which legal question the evidence can properly inform. The quality of an evaluation is measured by the warrant connecting its claims to the record and by the clarity with which its limitations are communicated. This is the foundation for cumulative research into criminal responsibility across clinical science, psychology, legal institutions, and emerging artificial decision-support systems. APA (2013) AAPL (2014)
References
- 18 U.S.C. § 17. Insanity defense. https://www.law.cornell.edu/uscode/text/18/17
- American Academy of Psychiatry and the Law. (2014). AAPL practice guideline for forensic psychiatric evaluation of defendants raising the insanity defense. Journal of the American Academy of Psychiatry and the Law, 42(4 Suppl.), S3–S76. https://pubmed.ncbi.nlm.nih.gov/25492121/
- American Psychological Association. (2013). Specialty guidelines for forensic psychology. American Psychologist, 68(1), 7–19. https://www.apa.org/practice/guidelines/forensic-psychology.html
- Callahan, L. A., Steadman, H. J., McGreevy, M. A., & Robbins, P. C. (1991). The volume and characteristics of insanity defense pleas: An eight-state study. Bulletin of the American Academy of Psychiatry and the Law, 19(4), 331–338. https://pubmed.ncbi.nlm.nih.gov/1786413/
- Cirincione, C., Steadman, H. J., & McGreevy, M. A. (1995). Rates of insanity acquittals and the factors associated with successful insanity pleas. Bulletin of the American Academy of Psychiatry and the Law, 23(3), 399–409. https://pubmed.ncbi.nlm.nih.gov/8845530/
- Clark v. Arizona, 548 U.S. 735 (2006). https://www.law.cornell.edu/supct/html/05-5966.ZO.html
- Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). https://www.law.cornell.edu/supremecourt/text/509/579
- Diaz v. United States, 602 U.S. 526 (2024). https://www.law.cornell.edu/supremecourt/text/23-14
- Federal Rule of Criminal Procedure 12.2. Notice of an insanity defense; mental examination. https://www.govinfo.gov/content/pkg/USCODE-2024-title18/html/USCODE-2024-title18-app-federalru-rule12.2.htm
- Federal Rule of Evidence 702. Testimony by expert witnesses (amended 2023). https://www.law.cornell.edu/rules/fre/rule_702
- Federal Rule of Evidence 704. Opinion on an ultimate issue. https://www.law.cornell.edu/rules/fre/rule_704
- Gowensmith, W. N., Murrie, D. C., & Boccaccini, M. T. (2013). How reliable are forensic evaluations of legal sanity? Law and Human Behavior, 37(2), 98–106. https://doi.org/10.1037/lhb0000001
- Janofsky, J. S., Vandewalle, M. B., & Rappeport, J. R. (1989). Defendants pleading insanity: An analysis of outcome. Bulletin of the American Academy of Psychiatry and the Law, 17(2), 203–211. https://pubmed.ncbi.nlm.nih.gov/2758121/
- Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114(1), 3–28. https://doi.org/10.1037/0033-2909.114.1.3
- Kahler v. Kansas, 589 U.S. 271 (2020). https://www.law.cornell.edu/supremecourt/text/18-6135
- Murrie, D. C., Boccaccini, M. T., Guarnera, L. A., & Rufino, K. A. (2013). Are forensic experts biased by the side that retained them? Psychological Science, 24(10), 1889–1897. https://doi.org/10.1177/0956797613481812
- Rogers, R., & Shuman, D. W. (2000). The mental state at the time of the offense measure: Its validation and admissibility under Daubert. Journal of the American Academy of Psychiatry and the Law, 28(1), 23–28. https://pubmed.ncbi.nlm.nih.gov/10774838/
- Rogers, R., Dolmetsch, R., & Cavanaugh, J. L., Jr. (1981). An empirical approach to insanity evaluations. Journal of Clinical Psychology, 37(3), 683–687. https://pubmed.ncbi.nlm.nih.gov/7263899/
- Rogers, R., Kropp, P. R., Bagby, R. M., & Dickens, S. E. (1992). Faking specific disorders: A study of the Structured Interview of Reported Symptoms (SIRS). Journal of Clinical Psychology, 48(5), 643–648. https://pubmed.ncbi.nlm.nih.gov/1401150/
- Rogers, R., Seman, W., & Wasyliw, O. E. (1983). The RCRAS and legal insanity: A cross-validation study. Journal of Clinical Psychology, 39(4), 554–559. https://pubmed.ncbi.nlm.nih.gov/6874994/
- Wygant, D. B., Tylicki, J. L., Disney, L. F., & Connelly, A. I. (2022). Structured Interview of Reported Symptoms–2nd Edition (SIRS-2): Use and admissibility in forensic mental health assessment. Journal of Personality Assessment, 104(2), 265–280. https://doi.org/10.1080/00223891.2021.2006673