Suicide

Abstract

Suicide is a form of self-injurious behavior: a self-directed, injurious act undertaken with any intent to die. Cognitive psychology studies it not as a single event but as a transition, reframed by the ideation-to-action perspective, which holds that the factors generating suicidal thoughts are distinct from those that convert a thought into an act. The central problem becomes why only a minority of those who think about suicide ever act, and the answer invokes an acquired capability for lethal self-harm, a brief and cognitively constricted acute state, and the lethality of the chosen method. This architecture organises the full spectrum—ideation, the non-fatal attempt, and the fatal act—while explaining why ideation predicts behaviour so weakly and why restricting lethal means most reduces the death rate. The account is mechanistic, not clinical advice; readers in distress should consult the support resources below.

Keywords: suicide, ideation-to-action, acquired capability, means restriction, suicidal behaviour

Suicide is among the most studied and least predictable of human behaviours. For most of the twentieth century, research treated suicidal thinking and suicidal action as points on a single continuum, assuming that stronger thoughts led more or less directly to a greater chance of acting. The decisive shift in the cognitive science of suicide was to reject that assumption. The factors that cause a person to think about suicide, it turns out, are largely different from the factors that govern whether that person will act (Klonsky et al., 2018). Separating the two questions reorganised the entire field and explained a stubborn empirical fact: after fifty years of research, known risk factors forecast suicidal behaviour only slightly better than chance, an accuracy that has not improved across decades (Franklin et al., 2017).

This article is the general entry point to the topic. It surveys the ideation-to-action frameworks that now dominate the field, the capability construct that gates the transition from thought to act, the brief and constricted acute state that precedes most attempts, the reappraisal of impulsivity, the cognitive measures that forecast suicidal behaviour without relying on self-report, and the lethality of means that most sharply separates the attempt that is survived from the one that is not. Each stage of the suicidal process—ideation, attempt, and fatal outcome—is treated in depth in a dedicated companion article; here the aim is to give the architecture that connects them.

What Suicide Is

Suicide is defined as death caused by a self-directed, injurious behaviour undertaken with any intent to die as a result of the behaviour (Van Orden et al., 2010). The definition isolates three cognitive elements: the behaviour is self-directed, it is potentially injurious, and it carries at least some intent to die. Intent is the element that distinguishes suicide from accidental or non-suicidal self-injury, and it is a mental state rather than an outcome. This matters because the outcome of a suicidal act—death or survival—is not a reliable index of the intent behind it: a person with unambiguous intent may survive a low-lethality act while another with more ambivalent intent may die from a highly lethal one.

Because intent and outcome come apart, the field distinguishes the components of the suicidal spectrum by where they sit in the transition from thought to act rather than by outcome alone. Suicidal ideation is the thinking; a suicide attempt is a non-fatal act carried out with some intent to die; and a fatal act is the same behaviour whose outcome was death. Treating these as stages of one process, rather than as a graded series of ever-worse symptoms, is the organising move of modern suicide research. It implies that explaining who thinks about suicide and explaining who acts are different problems with different answers, and that a complete account must reach past the thought to the capability that enables action and the means that determine its result.

Types of Suicide

In the Medical Subject Headings vocabulary, Suicide is a descriptor filed under the broader heading Self-Injurious Behavior, and it has several narrower descriptors that pick out the distinguishable forms and stages of the suicidal process. These subtypes are not mutually exclusive categories so much as points along the ideation-to-action transition and, in one case, a distinct legal and ethical class; a single person may move through several of them. The MeSH tree is an indexing classification built to organise the biomedical literature, not a causal taxonomy, so its branches should be read as retrieval categories rather than as a theory of suicide. The forms below each have a dedicated article.

• Suicidal Ideation — self-reported thinking about engaging in behaviour intended to end one’s own life, ranging from passive wishes to be dead through to active plans with intent; the first stage of every ideation-to-action framework.
• Attempted Suicide — a non-fatal, self-directed, potentially injurious behaviour carried out with at least some intent to die; the behavioural endpoint of the transition when the outcome is survival.
• Assisted Suicide — the provision of the means or knowledge to end one’s life by another person, a class distinguished by its legal, ethical, and medical context rather than by the cognitive mechanism of the act.
• Completed Suicide — a fatal, self-directed act undertaken with intent to die; the same behaviour as the attempt divided from it by an outcome that turns largely on the lethality of the method.

MeSH also files Suicide Prevention as a narrower heading, covering the strategies and interventions that act on this process; its best-evidenced instance, means restriction, is treated at length below.

The Ideation-to-Action Transition

The organising insight of modern suicide theory is that the factors causing suicidal ideation are distinct from those determining whether a person already thinking about suicide will act (Klonsky et al., 2018). The evidence for the separation is blunt: a meta-analysis of fifty years of prediction research found that known risk factors—including the severity of ideation itself—forecast suicidal behaviour and death only slightly better than chance, and that this accuracy had not improved across decades (Franklin et al., 2017). If the thought is a weak predictor of the act, the act must depend on something beyond the thought. Three frameworks converge on the same two-stage architecture while naming the gating variable somewhat differently.

The interpersonal theory of suicide holds that ideation arises from thwarted belongingness and perceived burdensomeness, but that progression to a lethal act additionally requires an acquired capability for suicide—a reduced fear of death and heightened tolerance of physical pain, built up through repeated exposure to painful and provocative experiences (Van Orden et al., 2010). On this account the capability to inflict lethal harm, not the desire alone, is what most sharply separates those who act from the many who think about suicide. The integrated motivational-volitional model makes the division structural: its motivational phase, driven by defeat and entrapment, generates ideation, while its volitional phase enumerates a separate set of moderators—capability, access to and familiarity with means, impulsivity, and planning—that govern whether ideation becomes behaviour (O’Connor & Kirtley, 2018). The three-step theory likewise reserves its final step for capability, resolved into three contributors: dispositional (a largely genetic pain sensitivity), acquired (habituation through experience), and practical (knowledge of and access to lethal methods) (Klonsky & May, 2015).

Note. All three frameworks separate the causes of ideation from a distinct factor that gates the step to an act; they agree that capability and access to means, not the strength of the thought, are what convert ideation into behaviour.

The theories differ in detail, but the shared claim has a direct empirical signature: variables that index capability—prior self-injury, occupational exposure to violence, and fearlessness about death—discriminate those who act from those who only think more reliably than the severity of ideation does (Klonsky et al., 2018). The demonstration below lets the reader vary the strength of ideation and the level of capability independently, and watch how the modelled probability of a suicidal act depends on their conjunction rather than on ideation alone.

The Acute Suicidal State

If capability determines who can carry out a suicidal act, a second body of work addresses when. A persistent assumption held that the path from decision to an attempt is a prolonged deliberation. Direct enquiry with survivors of near-fatal attempts—the closest accessible proxy for the cognition preceding an act—overturns this. A substantial proportion reported that less than ten minutes elapsed between the first current thought of suicide and the act itself, and for many more the interval was under an hour (Deisenhammer et al., 2009). The acute suicidal state is frequently compressed into a window far shorter than the scale on which most assessment and intervention operate.

This brevity has two cognitive implications. First, it reframes prediction: a state that arises and resolves within minutes is poorly captured by assessments administered days or weeks apart, which helps explain why static risk factors forecast behaviour so weakly (Franklin et al., 2017). Second, it describes a distinctive cognitive signature. The cognitive model of suicidal behaviour characterises the acute state as one of attentional fixation, in which attention narrows onto suicide as a solution and disengagement becomes difficult, accompanied by a collapse in the generation of alternative options (Wenzel & Beck, 2008). The same constriction appears in the impaired social problem-solving of those who attempt, who generate fewer and less effective solutions and show reduced ability to imagine specific future events (Pollock & Williams, 2004). When this brief, constricted state coincides with ready access to a highly lethal method, the odds that the act proves fatal rise sharply—not because intent is greater, but because a transient crisis meets an irreversible means.

Impulsivity Reconsidered

The short pre-act interval invites an intuitive inference: that suicidal acts are impulsive and that impulsive people are the ones who act. The cognitive evidence complicates both halves. A systematic reconsideration found that trait impulsivity—the stable disposition to act without forethought—is only weakly and inconsistently associated with suicidal behaviour once related variables are accounted for, and that its apparent role largely reflects its correlation with a lifetime accumulation of painful and provocative experiences (Anestis et al., 2014). Impulsive people encounter more of the experiences that build acquired capability; the capability, not the impulsivity, is the proximal contributor to an act.

The distinction that survives is between impulsivity as a trait and impulsivity as a state. An act may be executed rapidly, with little planning in the final minutes—consistent with the brief pre-act window—without the person being dispositionally impulsive across their life. The three-step theory captures this by placing fast, unplanned action within the acute capability-enabled state rather than attributing it to a personality trait (Klonsky & May, 2015). The upshot is consequential: because a person may move from decision to act faster than help can be summoned, and because an impulsive act with a highly lethal method is as deadly as a premeditated one, interventions that physically interrupt the act—chiefly by removing lethal means during that brief window—matter more than attempts to identify a stable impulsive personality in advance.

Cognitive Predictors Beyond Self-Report

The search for cognitive predictors begins with hopelessness. In a ten-year prospective study of patients hospitalised with suicidal ideation, the severity of hopelessness at intake—a negative expectation about the future, measured by self-report—identified almost all of those who would eventually die by suicide, and did so more sharply than the severity of depression itself (Beck et al., 1985). The finding established that a cognitive variable could forecast eventual death over a long horizon, and it set the agenda for the field: if a self-reported expectation carries prospective signal, what else does, and can the signal be read without relying on a report the person may be unable or unwilling to give? Because intent is difficult to report accurately—through unwillingness, shame, or lack of insight—cognitive psychology has since sought measures of the pre-act state that do not depend on self-report and that forecast suicidal behaviour prospectively. Two have shown that basic information-processing biases carry genuine predictive signal.

The first is an attentional bias toward suicide-related stimuli. Using a modified Stroop task, in which naming the colour of a word is slowed when its meaning captures attention, researchers found that the degree to which suicide-related words held patients’ attention predicted which of them would go on to a suicide attempt—and did so over and above clinicians’ own predictions (Cha et al., 2010). The second is an implicit association between the self and death. Measured by the relative speed of pairing self-related and death-related concepts, patients’ implicit identification with death predicted attempts above known risk factors and clinical judgement (Nock et al., 2010). Both measures tap automatic cognition the person may not report or even recognise, and both forecast behaviour rather than merely thought. A third, slower-moving marker completes the picture: overgeneral autobiographical memory—the tendency to recall the past in summarised categories rather than specific episodes—impairs the capacity to imagine concrete future solutions and is elevated in those with a history of attempts (Williams et al., 2007). For those who die, where no self-report is possible even in principle, the retrospective method of psychological autopsy complements these prospective markers: by systematically reconstructing the mental state of people who have died, through interviews with informants and review of records, such studies establish that the overwhelming majority met criteria for a diagnosable mental disorder, most commonly a mood disorder (Cavanagh et al., 2003). Together the prospective and retrospective evidence describe a cognitive profile of the person at risk, assembled without reliance on a report the person may be unable or unwilling to give.

Lethality of Means and Means Restriction

The most direct behavioural implication of the cognitive account concerns the method. Because whether an attempt proves fatal depends heavily on the lethality of the method rather than on the strength of intent, the case fatality of a method—the proportion of attempts with it that end in death—varies by more than an order of magnitude across means, from a few percent for the most commonly used methods of self-poisoning to the large majority for firearms (Spicer & Miller, 2000). This variation is the engine behind one of epidemiology’s most studied patterns: in many countries women attempt suicide more often than men yet men die more often, a divergence attributable in large part to the greater lethality of the methods more often chosen by men rather than to any difference in intent.

The cognitive logic makes the implication precise. If the acute state is brief and transient, and if case fatality is set largely by method, then reducing access to highly lethal means during that window should lower the death rate—and the evidence bears this out. A systematic review of suicide-prevention strategies identified restricting access to lethal means as one of only two interventions with robust supporting evidence, alongside physician education in depression recognition (Mann et al., 2005); a later evidence-based review reaffirmed means restriction as a core population-level strategy with consistent effects on the suicide death rate (Mann et al., 2021). The mechanism is cognitive as much as physical: because many acts arise from a short-lived crisis and are tied to a particular method, interposing a barrier at the moment of highest risk can prevent death entirely rather than merely displacing it, since a person in an acute, constricted state does not reliably substitute an equally lethal method when a familiar one is unavailable. Means restriction targets the practical component of capability in the three-step theory (Klonsky & May, 2015)—precisely the component most open to external modification. The interactive model below lets the reader vary the baseline rate of attempts and the lethality of available means, and observe how restricting the most lethal means reduces deaths even when the underlying rate of crises is held constant.

Worked Example

The following illustrates how the lethality logic is expressed quantitatively, reproducing the computation in the means-restriction demonstration above. The number of deaths produced by a population of suicidal crises is the product of two quantities: the number of attempts and the case fatality of the methods used. Means restriction works by lowering the second without necessarily changing the first.

Suppose a population experiences 1,000 suicidal crises that result in an attempt. Imagine that, before any intervention, 40 percent of these attempts use a highly lethal method with a case fatality of 0.85, and the remaining 60 percent use a lower-lethality method with a case fatality of 0.05. The expected number of deaths is therefore (1,000 × 0.40 × 0.85) + (1,000 × 0.60 × 0.05) = 340 + 30 = 370. Now suppose a means-restriction measure makes the highly lethal method unavailable to three quarters of those who would have used it, and that—because the crisis is brief and method-bound—those people do not substitute an equally lethal method but shift to the lower-lethality one. The highly lethal group falls to 10 percent of attempts and the lower-lethality group rises to 90 percent, while the total number of attempts is unchanged at 1,000. The expected number of deaths becomes (1,000 × 0.10 × 0.85) + (1,000 × 0.90 × 0.05) = 85 + 45 = 130. Deaths fall from 370 to 130—a reduction of 240, or about 65 percent—with no change in the number of crises or attempts. This is the arithmetic expression of the central point: because case fatality is set largely by method, acting on the method lowers the death rate even when the underlying rate of suicidal crises is held constant. The model quantifies a relationship; it does not describe any individual.

Discussion

The cognitive science of suicide has shifted the central question from who thinks about suicide to who acts, when, and with what result. Its most important contribution is the ideation-to-action separation, which recognised that explaining the thought and explaining the act are different problems and that the second turns on capability, the acute state, and access to means rather than on the intensity of the thought (Klonsky et al., 2018). This layered account explains both the weakness of ideation as a predictor of behaviour (Franklin et al., 2017) and the otherwise puzzling demographic patterns in the death rate, which track the lethality of methods more closely than the prevalence of distress (Spicer & Miller, 2000).

Two implications follow. First, measurement that relies on reported ideation misses the attentional and implicit biases and the acquired-capability markers that actually discriminate those who act—biases measurable directly and adding information beyond self-report and clinical judgement (Cha et al., 2010; Nock et al., 2010). Second, because much of the acute process is brief and state-driven rather than the product of a stable impulsive trait (Anestis et al., 2014; Deisenhammer et al., 2009), the interventions with the strongest evidence act on the environment at the moment of risk—restricting lethal means (Mann et al., 2005; Mann et al., 2021). The limits are real: the cognitive markers are group-level effects from research settings, not individual diagnostic tests, and the weak predictive accuracy of risk factors means that identifying which individual will act, and when, remains largely beyond reach. Understanding the cognition of suicide has improved explanation and sharpened intervention more than it has yet improved individual forecasting.

Current Directions

The most active contemporary research front treats the acute suicidal state as a dynamic process to be sampled in real time rather than inferred after the fact. Having established that the window between decision and act is often minutes rather than days (Deisenhammer et al., 2009), the field has turned to characterising what happens inside that window—the short-term fluctuations in hopelessness, entrapment, and attentional capture that distinguish the hours before an act from ordinary distress. Alongside this near-term, within-person orientation, evidence-based reviews have sharpened the population-level picture, reaffirming lethal-means restriction and the systematic treatment of underlying disorders as the strategies with the most consistent effects on the suicide death rate (Mann et al., 2021). The two strands are complementary: if the crisis is transient and its lethal outcome method-bound, then detecting acute escalation early and removing the method that would make it fatal are two ways to interrupt the same brief, state-driven transition (O’Connor & Kirtley, 2018).

Common Misconceptions

Glossary

Acquired capability

The reduced fear of death and heightened tolerance of physical pain, built up through exposure to painful or provocative experiences, that ideation-to-action theory holds is required to progress from thought to a potentially fatal act.

Acute suicidal state

The transient, often very brief period of intense attentional fixation on suicide and collapse of alternative options that immediately precedes many suicidal acts.

Assisted suicide

The provision by another person of the means or knowledge to end one’s own life; a class of suicidal act distinguished by its legal, ethical, and medical context rather than by cognitive mechanism.

Attentional bias

A systematic tendency for attention to be captured and held by a particular class of stimuli; in those at risk, suicide-related words differentially capture attention and predict future suicidal behaviour.

Case fatality

The proportion of suicide attempts using a given method that end in death; it varies by more than an order of magnitude across methods and is the quantity means restriction acts upon.

Cognitive constriction

A tunnel-like narrowing of attention and perceived options in the suicidal state, in which the range of imaginable solutions contracts until only continued suffering or escape remains.

Entrapment

The appraisal of being trapped in an aversive situation with no possibility of escape; in the motivational-volitional model it converts defeat into suicidal ideation.

Ideation-to-action framework

The organising principle of modern suicide theory that the factors causing suicidal ideation are distinct from those governing the progression from ideation to a potentially fatal act.

Implicit association

An automatic mental link between concepts measured by reaction time rather than self-report; implicit identification of the self with death predicts suicidal behaviour beyond clinical judgement.

Impulsivity

The disposition to act rapidly and without forethought; as a stable trait it is only weakly linked to suicidal behaviour, whereas rapid action within the acute state is common.

Integrated motivational-volitional model

A phased framework in which defeat and entrapment generate ideation in a motivational phase, and capability, means, and impulsivity govern the volitional step to an act.

Interpersonal theory of suicide

A framework in which ideation arises from thwarted belongingness and perceived burdensomeness, and the act additionally requires an acquired capability for suicide.

Lethality of means

The intrinsic probability that use of a particular suicide method results in death; the chief determinant of whether a given act is survived or fatal.

Means restriction

Reducing access to highly lethal methods of suicide; among the best-evidenced prevention strategies because many acts are tied to a particular method during a brief crisis and lowering case fatality lowers the death rate.

Overgeneral memory

A bias toward recalling autobiographical events as summarised categories rather than specific episodes, which impairs the generation of specific solutions to current problems.

Psychological autopsy

A retrospective method for reconstructing the mental state of a person who has died by suicide, through structured interviews with informants and review of records; the principal way of studying cognition that cannot be reported.

Social problem-solving

The cognitive process of generating and evaluating solutions to interpersonal difficulties; impaired in those who attempt suicide and linked to overgeneral memory and hopelessness.

Suicidal ideation

Self-reported thinking about engaging in behaviour intended to end one’s own life, ranging from passive wishes to be dead to active plans with intent; the first stage of the ideation-to-action process.

Suicide attempt

A non-fatal, self-directed, potentially injurious behaviour undertaken with at least some intent to die; distinguished from a fatal act only by its survived outcome.

Three-step theory

An ideation-to-action framework in which pain and hopelessness generate ideation, connectedness governs its escalation, and dispositional, acquired, and practical capability gate the act.

Key Researchers

Aaron T. Beck

(1921–2021). Originated the cognitive model of depression and suicide; with Wenzel developed the account of the acute suicidal state as one of attentional fixation and collapsed problem-solving, and established hopelessness as a prospective predictor of eventual death by suicide. (Wikipedia)

Craig J. Bryan

. Clinical psychologist whose work on fluid vulnerability, lethal-means safety, and crisis response planning addresses the acute suicidal state and the volitional side of the ideation-to-action framework. (ORCID, Google Scholar)

Thomas E. Joiner

. Developed the interpersonal theory of suicide, whose acquired-capability construct is the most influential account of why only some people with strong ideation progress to an act. (Google Scholar, Wikipedia)

E. David Klonsky

. Developer of the three-step theory, which resolves the capability that gates the act into dispositional, acquired, and practical contributors, the last of which is access to lethal means. (ORCID, Google Scholar)

Matthew K. Nock

. Pioneer of implicit and behavioural measures that predict suicidal behaviour beyond self-report, including the death/suicide implicit association test and the suicide attentional-bias paradigm; co-author of the field’s largest meta-analysis of risk factors. (ORCID, Google Scholar)

Rory C. O'Connor

. Developer of the integrated motivational-volitional model, whose volitional phase enumerates the capability, means, and impulsivity moderators that convert ideation into a potentially fatal act. (ORCID, Google Scholar)

J. Mark G. Williams

. Established the impaired problem-solving and overgeneral autobiographical memory that mark the cognitive constriction of the suicidal state, linking basic memory processes to the capacity to generate alternatives to the act. (ORCID, Wikipedia)

Frequently Asked Questions

What does cognitive psychology mean by the ideation-to-action framework?

It is the principle that the factors causing a person to think about suicide are distinct from the factors that determine whether they act. Separating the two explains why the severity of suicidal thoughts predicts behaviour only weakly and directs attention to the capability and means that gate the act (Klonsky et al., 2018).

Why do most people with suicidal thoughts never act on them?

Because the transition to an act requires factors beyond the thought—chiefly an acquired capability to carry out lethal self-harm—that most people who experience ideation do not possess (Van Orden et al., 2010).

How quickly do suicidal acts happen after the decision?

Often very quickly: survivors of near-fatal attempts frequently report that less than ten minutes passed between the first current thought of suicide and the act, so the acute state is often a matter of minutes rather than days (Deisenhammer et al., 2009).

Are suicidal acts usually impulsive?

An act may be executed rapidly, but trait impulsivity is only weakly and inconsistently associated with suicidal behaviour once related factors are considered; the rapid action reflects the acute state rather than a stable impulsive personality (Anestis et al., 2014).

Can suicidal behaviour be predicted from risk factors?

Not reliably at the individual level. A meta-analysis of fifty years of research found that known risk factors forecast suicidal behaviour only slightly better than chance, an accuracy that had not improved across decades (Franklin et al., 2017).

Do cognitive measures add anything beyond asking the person?

Yes. Attentional bias toward suicide-related words and an implicit association between the self and death each predict future attempts over and above self-report and clinical judgement, because they tap automatic cognition the person may not report or recognise (Cha et al., 2010; Nock et al., 2010).

Does restricting access to lethal means actually reduce suicide deaths?

Yes; systematic and evidence-based reviews identify means restriction as one of the few strategies with robust supporting evidence, because case fatality is set largely by method and many acts are not reliably substituted during a brief crisis (Mann et al., 2005; Mann et al., 2021).

What distinguishes a suicide attempt from a completed suicide?

They are the same class of self-directed, intended behaviour divided by outcome: an attempt is survived and a completed suicide ends in death. The division is drawn largely by the lethality of the method rather than by the strength of intent (Spicer & Miller, 2000).

Support Organizations

The organizations below offer free, confidential support to people experiencing suicidal thoughts and to those concerned about someone else.

• 988 Suicide and Crisis Lifeline (United States) — call or text 988 for 24/7 free and confidential support.
• Crisis Text Line (United States) — text HOME to 741741 to reach a trained crisis counsellor.
• American Foundation for Suicide Prevention (United States) — education, research funding, and resources for those affected by suicide.
• Samaritans (United Kingdom and Ireland) — call 116 123 for round-the-clock emotional support.
• International Association for Suicide Prevention (International) — maintains a global directory of crisis centres and resources.

References