Insanity Defense

Abstract

The insanity defense is a doctrine of forensic psychiatry: a legal excuse under which a defendant who committed an act is acquitted because a mental disorder deprived them of responsibility. It rests on an intuition — that punishment presupposes a mind capable of choosing — operationalized through successive tests, from the cognitive M'Naghten rule to broader volitional standards and back. This article distinguishes the defense from competency to stand trial, traces the major tests and what each asks, and confronts the gap between how the public imagines the defense and what the record shows: it is raised in under one percent of felony cases, succeeds in about a quarter of those, and rests on expert sanity judgments of modest reliability. The defense is less a loophole than a test of where the law draws the line between madness and blame.

Keywords: insanity defense, criminal responsibility, M'Naghten rule

What the Insanity Defense Is

The insanity defense is an affirmative defense in criminal law: the defendant does not deny committing the act but argues that, because of a mental disorder at the time, they should not be held criminally responsible for it. It is a doctrine of forensic psychiatry, the field where clinical assessment meets legal judgment, because it turns a psychiatric question — the defendant's mental state at the moment of the offense — into a legal verdict. A successful plea yields a special verdict, usually not guilty by reason of insanity (NGRI), which is an acquittal in name but almost never in consequence: the acquittee is typically committed to a secure psychiatric facility, often for longer than the prison term the conviction would have carried (#ref-steadman-1993).

The defense expresses a principle older than any statute — that blame requires a guilty mind, a mens rea, and that a mind sufficiently disordered cannot form one in the culpable sense the law demands. Isaac Ray's 1838 treatise on the medical jurisprudence of insanity gave the principle its first systematic medical footing, and the tests that followed are successive attempts to say, in words a jury can apply, exactly which derangements of mind excuse and which do not (#ref-perlin-1994). What has never been settled is where to draw that line, because it is not a medical fact to be discovered but a moral boundary to be chosen, and each legal test embeds a different choice.

The Legal Tests

The history of the insanity defense is a history of competing tests, each answering a different question about the disordered mind. The oldest surviving standard is the M'Naghten rule, formulated by the English House of Lords in 1843 after Daniel M'Naghten was acquitted of murder while under a paranoid delusion. It is purely cognitive: a defendant is excused only if, through a defect of reason from disease of the mind, they did not know the nature and quality of the act, or did not know that it was wrong (#ref-mnaghten-1843). M'Naghten asks nothing about self-control; a defendant who knew an act was wrong but could not stop themselves is, under it, still responsible.

That gap prompted the irresistible impulse test, which supplements M'Naghten by excusing a defendant who knew an act was wrong but was driven to it by an uncontrollable impulse — a volitional prong added to the cognitive one. A more radical departure came in 1954, when the D.C. Circuit in Durham v. United States adopted the product test: a defendant is not criminally responsible if the unlawful act was the product of mental disease or defect (#ref-durham-1954). Durham threw the question open to psychiatric testimony so completely that it proved unworkable — it made experts, not juries, the effective arbiters — and was abandoned within two decades. The synthesis that displaced it was the American Law Institute (ALI) standard from the Model Penal Code: a defendant is not responsible if, as a result of mental disease, they lacked substantial capacity either to appreciate the criminality of the conduct (cognitive) or to conform their conduct to the law (volitional). By softening “know” to “appreciate” and “did not” to “lacked substantial capacity,” the ALI test was meant to be both humane and administrable (#ref-borum-1999).

Then the pendulum swung back. After John Hinckley was acquitted by reason of insanity for shooting President Reagan in 1981, public outcry drove the federal Insanity Defense Reform Act of 1984, which narrowed the federal standard to a near-purely cognitive test, dropped the volitional prong, and shifted the burden of proof onto the defense. Many states followed, and Stephen Morse's contemporaneous analysis defended a narrow, cognitively grounded excuse against both abolition and volitional over-reach, arguing that the law should excuse only those who cannot rationally grasp what they are doing (#ref-morse-1985). The reforms did not, however, change the empirical picture the defense's critics most misunderstood.

The legal tests disagree about the same facts. Pick a test and see whom it would excuse. Notice that the volitional cases — knew it was wrong but could not stop — turn on which test the jurisdiction uses.

M'Naghten excuses only a failure of understanding, so it holds the "knew-it-was-wrong-but-could-not-stop" defendant responsible; the ALI test adds a volitional prong and excuses him; Durham excuses any act that is the product of mental disease, the breadth that made it unworkable. The 1984 federal reform moved the federal standard back toward M'Naghten.

Table 1

Principal Legal Tests of Insanity, by What Each Asks of the Mind

Note. The tests differ mainly in whether they excuse a failure of self-control (the volitional prong) as well as a failure of understanding (the cognitive prong). The 1984 federal reform, prompted by the Hinckley acquittal, dropped the volitional prong and returned the federal standard to a near-purely cognitive form (Borum & Fulero, 1999).

Insanity Is Not Incompetency

The single most common confusion about the insanity defense is to mistake it for competency to stand trial, and the two are entirely distinct in both timing and question. Competency concerns the defendant's mental state now, at trial: the standard, set by the Supreme Court in Dusky v. United States, asks whether the defendant has a rational and factual understanding of the proceedings and can consult with counsel with a reasonable degree of rational understanding (#ref-dusky-1960). A defendant found incompetent is not tried at all until competency is restored; the trial is simply paused.

Insanity, by contrast, concerns the defendant's mental state at the time of the offense, often years earlier, and it is a defense raised at trial to defeat responsibility. A defendant can be perfectly competent to stand trial yet mount an insanity defense about their state during the crime; conversely, a floridly disordered defendant may be incompetent to be tried while their sanity at the time of the act is never adjudicated. The evaluations differ accordingly: a competency assessment probes present understanding and reasoning, while a criminal-responsibility (sanity) evaluation is a retrospective reconstruction of a past mental state, drawing on records, accounts, and the defendant's report — a far harder inference, and one standardized instruments were later built to discipline (#ref-rogers-2000). Keeping the two apart is the first thing any forensic evaluation must do.

How Often It Is Used — and How Often It Works

Few areas of criminal law are as badly misunderstood by the public as this one, and the misunderstanding has been measured. In the landmark eight-state study, Lisa Callahan and colleagues assembled the first large multi-jurisdiction dataset on insanity pleas and found the defense was raised in under one percent of felony cases — roughly 0.9% — a figure orders of magnitude below public belief (#ref-callahan-1991). Eric Silver, Carmen Cirincione and Henry Steadman then quantified the belief itself: surveyed citizens estimated the defense was raised in about 37% of felony cases and succeeded most of the time, when in reality it is raised rarely and succeeds in only about a quarter of the cases in which it is even attempted (#ref-silver-1994). The authors called their paper “demythologizing” for good reason: almost every intuition the public holds about the insanity defense is quantitatively wrong.

The success figure is equally instructive. Carmen Cirincione, Henry Steadman and Margaret McGreevy found that insanity acquittals, far from being won by adversarial theatrics, are overwhelmingly the result of agreement — the defense typically succeeds only when the prosecution's and defense's evaluators concur that the defendant was insane, so that the verdict is effectively uncontested (#ref-cirincione-1995). The image of a cunning defendant fooling a jury is almost the inverse of the reality: contested insanity pleas usually fail, and the ones that win are the ones nobody seriously disputes. When Steadman and colleagues examined the aftermath of the Hinckley-era reforms, they found that tightening the standard changed the volume and disposition of pleas far less than reformers expected, precisely because the defense was already so rare and so seldom successful (#ref-steadman-1993). The worked example below makes the arithmetic concrete.

Of 100,000 felony cases, the defense is raised in a small fraction and succeeds in a fraction of those. Adjust the two rates and compare the result with what surveyed citizens believe.

At the empirical defaults (raised in 0.9% of cases, succeeding in 26% of those), successful insanity acquittals run about 0.23% of felony cases — roughly 1 in 427 — while the public imagines a rate near 16%. The defense that dominates public anxiety is, in the aggregate, a rounding error in the criminal docket.

The Reliability of Sanity Evaluations

If insanity verdicts track expert agreement, the reliability of the underlying evaluations becomes the hinge of the whole system — and that reliability is only moderate. Neil Gowensmith, Daniel Murrie and Marcus Boccaccini exploited a natural experiment in Hawaii, where statute requires three independent evaluators to assess each defendant's sanity, and asked how often the three agreed. Unanimous agreement on legal sanity occurred in only about 55% of cases; on the question of criminal responsibility the evaluators disagreed far more than the field's confidence in its judgments would suggest (#ref-gowensmith-2013). Because courts follow evaluator consensus, this modest inter-rater reliability propagates directly into verdicts: the same defendant might be found sane or insane depending substantially on which evaluators happened to be assigned.

Standardized instruments were developed partly to constrain this variability. Richard Rogers built structured approaches to criminal-responsibility assessment that decompose the global “sane or insane” judgment into rated components — the presence of a disorder, its relation to the offense, and the resulting impairment of cognition or control — so that evaluators reason through the same explicit steps rather than reaching a holistic verdict by private routes (#ref-rogers-2000). A recent systematic review by Giovanna Parmigiani and colleagues catalogued the structured and semi-structured tools now available for insanity evaluations and found the evidence base for their validity still thin, even as their use grows (#ref-parmigiani-2023). Structure improves consistency, but it cannot manufacture a ground truth against which a retrospective judgment of a past mind can be scored, and that missing criterion is why the field's reliability ceiling remains stubbornly low.

Some jurisdictions require several independent evaluators to judge a defendant's sanity. If each reaches an "insane" call at some rate and they decide independently, how often do all of them agree?

With three evaluators calling "insane" about a fifth of the time, unanimity appears in only about half of cases — close to the ~55% Gowensmith and colleagues measured in Hawaii's three-evaluator system. Requiring more evaluators lowers agreement further. This idealized independence model understates real correlation between examiners, but it shows why modest per-evaluator reliability translates into frequent panel disagreement.

Worked Example

Suppose a jurisdiction handles 100,000 felony cases in a year, and we want the true rate of successful insanity acquittals against the public's estimate of it. Use the empirical figures: the defense is raised in about 0.9% of felony cases and succeeds in about 26% of the cases where it is raised.

Cases in which the defense is raised: 0.9% × 100,000 = 900. Of those, the number that succeed: 26% × 900 = 234. So successful insanity acquittals number 234 out of 100,000 felony cases — a rate of 0.234%, or about 1 in 427.

Now compare the public's picture. Survey respondents estimated the defense was raised in roughly 37% of cases and, of those, succeeded in a large fraction — take the surveyed figure of about 44% success. That imagined world would produce 0.37 × 100,000 = 37,000 pleas and 0.44 × 37,000 ≈ 16,280 acquittals — a rate of 16.3%, or about 1 in 6.

The public's estimate of successful insanity acquittals thus exceeds the real rate by a factor of 16,280 / 234 ≈ 70. Put differently, people imagine the insanity defense succeeding roughly seventy times more often than it does. The gap is not a rounding error or a quirk of one jurisdiction; it is the central empirical finding of the field, replicated across the eight-state data, and it explains why reforms aimed at a supposed epidemic of insanity acquittals have so little to bite on. The defense that dominates public anxiety is, in the aggregate, a rounding error in the criminal docket.

Discussion

The insanity defense endures because it protects something the criminal law cannot easily abandon: the premise that punishment is reserved for those who could have done otherwise. Strip the defense away entirely and the law would punish the deluded and the rational alike, severing blame from its moral basis; make it too capacious and it dissolves responsibility into a medical explanation for everything. Every legal test is a compromise on that spectrum, and the historical oscillation — cognitive M'Naghten, then the volitional expansions, then the retreat to cognition after Hinckley — is not confusion but a society repeatedly renegotiating how much loss of control it is willing to treat as loss of blame. Some states responded not by narrowing the test but by adding an alternative guilty but mentally ill (GBMI) verdict, which convicts the defendant while formally acknowledging mental illness and allowing treatment during the sentence — a compromise that preserves the conviction rather than replacing it with an acquittal (#ref-borum-1999). Morse's argument that the excuse should track a failure of rationality rather than a failure of will is one influential resolution, but it remains contested precisely because the underlying question is moral, not empirical (#ref-morse-1985).

What the empirical record adds is a sharp correction to the politics that surrounds the defense. The reforms that followed Hinckley were driven by a belief — that the insanity defense is a common and easily exploited escape from justice — that the data flatly contradict: it is rare, usually unsuccessful, typically uncontested when it does succeed, and followed by confinement rather than release (#ref-silver-1994). Michael Perlin named the force behind the misperception sanism, an irrational prejudice against people with mental illness that shapes insanity-defense law as powerfully as evidence does, and that survives repeated empirical refutation because it was never grounded in evidence to begin with (#ref-perlin-1994). The defense's real problems are not the imagined ones. They lie in the modest reliability of the sanity judgments courts depend on, and in the honest difficulty of reconstructing a mind at a past moment — problems of measurement and inference, not of a loophole waiting to be closed.

Current Directions

Two developments are reshaping the field. The first is neuroscientific. Advances in brain imaging and cognitive neuroscience have raised the prospect of grounding claims about impaired responsibility in measurable neural findings rather than clinical impression alone, and Desmond Egbenya and Samuel Adjorlolo survey both the promise and the hazard: neuroscience may sharpen the assessment of a defendant's mental state at the time of an offense, but a brain abnormality is not itself an excuse, and importing neural evidence without a theory of how it bears on legal responsibility risks a new, more technical version of the Durham error (#ref-egbenya-2021). The live question is not whether the brain is involved in every act — it is — but which neural facts, if any, the law should treat as excusing.

The second direction is empirical follow-through: closing the evidentiary gaps the field has long tolerated. Haseeb Haroon and colleagues examined two decades of insanity-acquittee outcomes in one state forensic program — the release and rehospitalization trajectories that begin after the verdict — an understudied phase in which the acquittee's long confinement is actually decided (#ref-haroon-2023). Parallel work continues to test and standardize the evaluation instruments themselves, with reviews pressing for validity evidence to match their spreading use (#ref-parmigiani-2023). The through-line is a field maturing from doctrinal argument toward measurement: less debate over which words a test should use, more data on how reliably any test can be applied and what becomes of those it acquits.

Common Misconceptions

The insanity defense is a common way to escape conviction.

It is raised in under one percent of felony cases and succeeds in only about a quarter of those, so successful insanity acquittals are a tiny fraction of the criminal docket — the public overestimates the rate by roughly seventy-fold (#ref-silver-1994).

An insanity acquittal means the defendant goes free.

Almost never. A person found not guilty by reason of insanity is typically committed to a secure psychiatric facility, often for as long as or longer than the sentence a conviction would have brought (#ref-steadman-1993).

Insanity and incompetency to stand trial are the same thing.

They are different questions at different times: competency concerns the defendant's present ability to understand and assist at trial, while insanity concerns their mental state at the time of the offense (#ref-dusky-1960).

Winning insanity cases are courtroom battles between duelling experts.

The opposite is usually true: successful pleas are overwhelmingly ones in which the opposing evaluators agree the defendant was insane, so the verdict is effectively uncontested — contested pleas mostly fail (#ref-cirincione-1995).

Glossary

ALI standard.

The Model Penal Code test excusing a defendant who, because of mental disease, lacked substantial capacity to appreciate the criminality of the conduct or to conform conduct to the law; combines a cognitive and a volitional prong.

Cognitive prong.

The part of an insanity test concerned with knowledge or understanding — whether the defendant knew the nature of the act or that it was wrong.

Competency to stand trial.

A defendant's present capacity to understand the proceedings and assist counsel; assessed at the time of trial and distinct from sanity at the time of the offense.

Criminal responsibility.

The legal accountability of a person for a criminal act; the insanity defense is a claim that a mental disorder negated it at the time of the offense.

Durham (product) rule.

The 1954 test excusing a defendant when the unlawful act was the product of mental disease or defect; abandoned as unworkable because it ceded the question to expert testimony.

Dusky standard.

The U.S. Supreme Court test for competency to stand trial: a rational and factual understanding of the proceedings and the ability to consult with counsel with a reasonable degree of rational understanding.

Forensic psychiatry.

The branch of psychiatry at the interface of mental health and the law; the parent kind under which MeSH files the insanity defense.

Guilty but mentally ill (GBMI).

An alternative verdict adopted by some states that convicts the defendant while recognizing mental illness, allowing treatment during a criminal sentence; not an acquittal.

Insanity Defense Reform Act.

The 1984 U.S. federal law, passed after the Hinckley acquittal, that narrowed the federal standard to a near-purely cognitive test, dropped the volitional prong, and placed the burden of proof on the defense.

Irresistible impulse test.

A supplement to M'Naghten excusing a defendant who knew an act was wrong but could not control the impulse to commit it; adds a volitional prong to the cognitive one.

M'Naghten rule.

The 1843 English test, still the majority U.S. standard, excusing a defendant who, from disease of the mind, did not know the nature and quality of the act or did not know it was wrong; purely cognitive.

Mens rea.

The “guilty mind” — the mental element of a crime; the insanity defense contends that a mental disorder prevented the defendant from forming it in the culpable sense the law requires.

Not guilty by reason of insanity (NGRI).

The special verdict returned when an insanity defense succeeds; a legal acquittal that ordinarily results in commitment to a secure psychiatric facility rather than release.

Sanism.

Michael Perlin's term for the irrational prejudice against people with mental illness that shapes insanity-defense law and sustains public misperceptions in the face of contrary evidence.

Volitional prong.

The part of an insanity test concerned with self-control — whether the defendant could conform their conduct to the law; present in the ALI and irresistible-impulse tests, dropped by the 1984 federal reform.

Key Researchers

W. Neil Gowensmith (living). Forensic psychologist at the University of Denver whose studies of field reliability — including the Hawaii three-evaluator natural experiment — measured how often independent examiners actually agree on a defendant's legal sanity. ORCID

Thomas Grisso (born 1942). Emeritus professor of psychiatry at the University of Massachusetts Chan Medical School and a foundational scholar of forensic assessment, whose work established the conceptual framework for evaluating competency and criminal responsibility. ORCID - Wikipedia - Wikidata

Stephen J. Morse (living). Professor of law and of psychology and law in psychiatry at the University of Pennsylvania; a leading philosopher of criminal responsibility whose “Excusing the Crazy” defends a narrow, rationality-based insanity excuse. Faculty Page - Wikipedia - Wikidata

Daniel C. Murrie (living). Forensic psychologist at the University of Virginia's Institute of Law, Psychiatry and Public Policy who studies evaluator bias and the field reliability of forensic mental-health judgments. ORCID - Faculty Page

Isaac Ray (1807-1881). Founding superintendent of Butler Hospital and a founder of American psychiatry; his A Treatise on the Medical Jurisprudence of Insanity (1838) shaped nineteenth-century insanity law and influenced the Durham product test. Wikipedia - Wikidata

Richard Rogers (born 1950). Regents Professor of psychology at the University of North Texas and the developer of standardized forensic assessment instruments, including structured approaches to criminal-responsibility evaluation. ORCID - Wikipedia

Henry J. Steadman (living). Sociologist and co-founder of Policy Research Associates who led the landmark multi-state empirical studies of the insanity defense — the eight-state study and the Hinckley-reform evaluation — that demythologized public perceptions of the plea. Faculty Page - Policy Research Associates

Frequently Asked Questions

What is the insanity defense? It is an affirmative defense in criminal law under which a defendant who committed an act is acquitted because a mental disorder at the time of the offense deprived them of criminal responsibility. A successful plea returns a special verdict, usually not guilty by reason of insanity, and it is a doctrine of forensic psychiatry.

How is insanity different from competency to stand trial? Insanity concerns the defendant's mental state at the time of the crime and is a defense to responsibility; competency concerns their present ability, at trial, to understand the proceedings and help their lawyer. A defendant can be competent to be tried yet argue they were insane when the offense occurred.

What is the M'Naghten rule? It is the 1843 English test, still the majority U.S. standard, that excuses a defendant who, because of a disease of the mind, did not know the nature and quality of the act or did not know that it was wrong. It is purely cognitive and says nothing about self-control.

How often is the insanity defense actually used? Rarely. Large multi-state studies find it is raised in under one percent of felony cases (about 0.9%) and succeeds in only roughly a quarter of the cases in which it is even attempted, far below what the public believes.

Does someone found not guilty by reason of insanity go free? Almost never. An insanity acquittee is typically committed to a secure psychiatric facility, frequently for as long as or longer than the prison sentence a conviction would have carried.

What changed after the Hinckley case? John Hinckley's 1982 insanity acquittal for shooting President Reagan prompted the federal Insanity Defense Reform Act of 1984, which narrowed the standard to a near-purely cognitive test, eliminated the volitional prong, and shifted the burden of proof onto the defense. Many states enacted similar changes.

How reliable are the expert evaluations of sanity? Only moderately. When three independent evaluators assess the same defendant, they reach unanimous agreement on legal sanity in only about 55% of cases, so the same defendant might be judged sane or insane depending in part on which examiners are assigned.

What is the difference between the cognitive and volitional prongs? The cognitive prong asks whether the defendant understood what they were doing or that it was wrong; the volitional prong asks whether they could control their conduct even if they did understand. Tests differ chiefly in whether they include the volitional prong, which the 1984 federal reform removed.

References