Dangerous Behavior

Abstract

Dangerous behavior is a type of behavior: the willingness to engage in, and the tendency to carry out, actions that could harm oneself or others—what the research literature calls risk-taking. (MeSH also files it among social problems.) Cognitive psychology treats risk-taking not as a single appetite but as the joint product of how a person perceives a risk, how it makes them feel, and how well a developing control system restrains the impulse to pursue it. This article sets out how risk-taking is measured, why the same person is bold in one domain and cautious in another, how analysis and affect compete to value a risky option, and why the propensity peaks in adolescence. Three demonstrations make domain-specific risk profiles, the expected value of a behavioral risk task, and the dual-systems developmental gap concrete.

Keywords: dangerous behavior, risk-taking, sensation seeking, dual-systems model

Dangerous behavior is the umbrella the National Library of Medicine uses for conduct that exposes the self or others to harm; its central entry term, and the term the research literature actually works with, is risk-taking. The cognitive question is not why anyone would court harm—every worthwhile action carries some—but how people weigh an uncertain prospect of loss against its reward, and why that weighing produces such different behavior across people, situations, and ages. The answer that has emerged is that risk-taking is not one thing: it is assembled, each time, from perception, affect, disposition, and control.

What Dangerous Behavior Is

The defining move is to treat risk-taking as a decision under uncertainty rather than as recklessness. A risky choice is one whose outcomes are probabilistic and include a meaningful chance of loss; to take the risk is to accept that chance in exchange for a possible gain. Framed this way, dangerous behavior becomes continuous with ordinary decision making, and the interesting variation is in how the uncertain prospect is valued. Two people facing the identical odds can value them differently because they perceive the probabilities differently, because the prospect feels differently to each, or because one can restrain an impulse the other cannot.

How that valuation departs from a plain expected-value calculation is itself lawful. Prospect theory shows that people evaluate outcomes as gains and losses relative to a reference point rather than as final states, and that the resulting value function leaves them risk-averse for gains but risk-seeking for losses (Kahneman & Tversky, 1979). A direct consequence is the framing effect: the same prospect is accepted or refused according to whether it is described in terms of what is gained or what is lost, so how a risk is posed can tip a person toward or away from taking it (Tversky & Kahneman, 1981).

That is why a single number—how much risk a person “likes”—explains so little. The modern instruments were built precisely to decompose risk-taking into components: the perceived likelihood and severity of harm, the affective reaction to it, and the behavioral tendency to engage anyway (Blais & Weber, 2006). Perception, feeling, and action dissociate, and dangerous behavior is what results when they align toward engagement.

Types of Dangerous Behavior

Within the Medical Subject Headings, Dangerous Behavior is the parent of several more specific headings, each a recognized form in which the willingness to accept harm is enacted. The table lists those with a dedicated article on this site. The groupings are a library indexing classification, not a causal taxonomy: a given episode may belong to more than one, and the categories describe where a behavior is filed, not the psychological mechanism that produces it. The parent heading sits in turn beneath the general category of behavior, and (through its second tree position) beneath social problems.

Table 1

Forms of Dangerous Behavior Indexed Under the MeSH Descriptor

Note. Direct narrower descriptors of Dangerous Behavior (D003617) that have an article on this site; a fourth, Aggressive Driving, is indexed by MeSH but not yet covered here. Descriptors are an indexing scheme, so the forms are neither exhaustive nor mutually exclusive.

Measuring Risk-Taking

Because risk-taking fractionates by domain, the workhorse self-report instrument is explicitly domain-specific. The Domain-Specific Risk-Taking (DOSPERT) scale asks how likely a person is to engage in risky activities across separate areas—financial, health and safety, recreational, ethical, and social—and scores each domain on its own, together with the person's perception of the risks and expected benefits involved (Blais & Weber, 2006). The crucial empirical result is low cross-domain correlation: the same respondent is routinely a financial conservative and a recreational daredevil. A single risk-appetite score averages this structure away and predicts poorly; the domain profile predicts well.

Behind the domains sits a common decomposition. Reported risk-taking in a domain is well modeled as the perceived benefit of the activity minus its perceived riskiness, each weighted by the person's attitude toward that trade-off (Blais & Weber, 2006). Apparent recklessness often turns out, on measurement, to be an ordinary cost-benefit judgment resting on an unusual perception of the odds or an unusual taste for the reward, rather than an indifference to harm.

The domain-specific risk profile (DOSPERT)

Risk-taking fractionates by domain, so the same person can be a financial conservative and a recreational daredevil. Set each domain’s likelihood of risk-taking (0–10) and compare the profile with the single average score. The wider the spread, the more a one-number risk-appetite summary throws away.

A single average of 4.2 hides a spread of 8 points across domains. The domain profile and the average tell very different stories — the average predicts poorly.

Risk in the Moment: Behavioral Tasks

Self-report captures what people say they would do; behavioral tasks capture what they do when a real, escalating risk is in front of them. The Balloon Analogue Risk Task (BART) asks participants to inflate a virtual balloon one pump at a time, with each pump adding to a temporary bank but also raising the chance the balloon bursts and wipes that balloon's earnings out (Lejuez et al., 2002). The measure is the average number of pumps on balloons that did not burst—a behavioral index of the willingness to push an advantage toward its breaking point. BART scores correlate with real-world risk behaviors such as smoking and drug use, giving the laboratory task external validity that a questionnaire alone cannot claim (Lejuez et al., 2002).

The task is informative because it has a correct answer that most people miss. Each pump trades a fixed gain against a rising probability of ruin, so expected earnings climb, peak, and then fall as the burst risk overtakes the reward. Cautious players stop short of the peak and leave money on the table; the incautious push past it and lose balloons. The Worked Example below computes exactly where that peak lies, and the demonstration locates it interactively.

Finding the optimal stopping point in the BART

A balloon bursts at a point equally likely to be any pump from 1 to 128; each successful pump banks $0.05, a burst forfeits the balloon. Choose how many pumps to plan (k) and watch expected earnings EV(k) = 0.05 × k × (128 − k) / 128 climb, peak, and fall. The peak sits at 64 pumps — neither timid nor reckless.

EV(64) = 0.05 × 64 × (128 − 64) / 128 = $1.60 — optimal: this is the expected-value peak.

Why People Take Risks

If perception and valuation drive risk-taking, what drives those? Three strands of evidence converge. The first is perceptual: risks are judged not on statistics but on qualitative features—how dreaded, how controllable, how catastrophic, how well understood—that the psychometric paradigm first mapped, so that a risk's felt size can diverge sharply from its measured size (Slovic, 1987). The second is affective: people evaluate a risky prospect partly by consulting the feeling it evokes, and this risk-as-feelings response can override, rather than merely inform, the cognitive assessment of probabilities and outcomes—which is why vivid, emotionally charged dangers loom larger than frequent but dull ones (Loewenstein et al., 2001).

The third strand is dispositional. Some of the variance is a stable trait: sensation seeking, the need for varied, novel, and intense experience and the willingness to take risks to obtain it, is a heritable dimension of personality that predicts risk behavior across many domains (Zuckerman & Kuhlman, 2000). But trait is not destiny. Whether a given person takes a given risk depends on the situation, the domain, and the person's momentary affective state acting together, so risk-taking is best predicted by the interaction of who, what, and when rather than by any one of them alone (Figner & Weber, 2011).

The Developmental Window

Risk-taking is not flat across the lifespan: it rises sharply in adolescence and declines into adulthood, a pattern so robust it appears across cultures (Steinberg, 2008). The leading explanation is the dual-systems (or maturational-imbalance) model. A socioemotional system, sensitive to reward and strongly activated at puberty, matures early; a cognitive-control system, supporting impulse control and the weighing of long-term consequences, matures slowly into the mid-twenties. For a few years the accelerator outpaces the brakes, and the gap between them is the window of heightened risk-taking (Steinberg, 2008; Shulman et al., 2016).

The model's sharpest contribution is to dislodge the intuitive view that adolescents take risks because they misjudge them. Fuzzy-trace theory shows that adolescents often reason about risks more quantitatively than adults—tallying odds that adults dismiss out of hand on the gist that a danger is simply not worth it—so better deliberation is not obviously protective (Reyna & Farley, 2006). The behavior reflects a reward system in ascendancy and a control system still maturing, not a deficit of knowledge; the implication for policy is that information campaigns alone are weak levers (Steinberg, 2008).

The maturational imbalance of the dual-systems model

A reward-sensitive socioemotional system matures early and peaks in mid-adolescence; a cognitive-control system matures slowly into the mid-twenties. Where the accelerator outruns the brakes, the gap is the window of heightened risk-taking. Move the age marker to read the two systems and their imbalance.

Imbalance (sensation − control) = 7.0 — a wide gap: reward sensitivity far outruns control, the peak risk window.

Worked Example

Consider a simplified Balloon Analogue Risk Task. A balloon's bursting point is equally likely to be any whole number of pumps from 1 to 128, and each successful pump banks $0.05, but a burst forfeits that balloon's bank. A player who plans to stop after k pumps keeps the bank only if the balloon survives all k, which happens when its bursting point exceeds k—a probability of (128 - k) / 128. Expected earnings are therefore EV(k) = 0.05 x k x (128 - k) / 128. This rises, peaks, and falls: the term k x (128 - k) is largest when k = 64, giving EV(64) = 0.05 x 64 x 64 / 128 = 0.05 x 32 = $1.60. Stopping too early at k = 30 yields only 0.05 x 30 x 98 / 128 = $1.15, and pushing on to k = 80 falls back to 0.05 x 80 x 48 / 128 = $1.50, because by then the roughly 37% survival chance eats the extra pumps. The optimal policy—64 pumps—is neither timid nor reckless, and the task scores how far a person's average lands from it. The numbers are illustrative, but the shape is the real claim: more risk raises reward only up to a point, after which the growing chance of ruin dominates.

Discussion

Dangerous behavior is a productive topic for cognitive psychology precisely because it resists the folk theory built into its name. “Dangerous” implies a defect—ignorance, impulsivity, a death wish—yet the evidence keeps returning the behavior to the ordinary machinery of judgment and choice. Risk-taking is domain-specific, so there is no single reckless type; it is valued by analysis and affect at once, so it is neither purely rational nor purely emotional; and its developmental peak reflects the ordinary timetable of brain maturation rather than a failure of understanding. The behavior is a normal output of normal cognition operating on uncertain, affectively charged prospects.

That reframing has a cost worth stating plainly. Treating risk-taking as decision-making under uncertainty captures the deliberative and affective components well, but some dangerous behavior is compulsive or habitual rather than chosen, and the decision frame fits those cases less comfortably. The construct is also normative at its edges: what counts as an unacceptable risk is a social judgment, which is why the same descriptor umbrella spans a laboratory balloon task and the forms of harm indexed in Table 1.

Current Directions

Recent work has pressed the developmental account in two directions. The first is global. A study of over 5,000 individuals across eleven countries confirmed that both heightened sensation seeking in adolescence and the slow maturation of self-regulation appear worldwide, supporting the dual-systems model as a species-typical pattern rather than an artifact of Western samples (Steinberg et al., 2018). A companion analysis traced the specific age curves of risk-taking and its psychological correlates across the same countries, finding broad similarity alongside culturally patterned differences in timing (Duell et al., 2018).

The second direction concerns how risk information reaches the decision. Much laboratory work describes the odds to participants, but real adolescents usually learn risks by experience, through feedback over repeated choices. When the two are compared, described and experienced risks produce systematically different behavior, and the age differences in risk-taking depend on which mode is in play—so how a risk is learned, not only what it is, shapes who takes it (Rosenbaum et al., 2018). Together these lines are moving the field from a single developmental curve toward a contextualized account of when, where, and how the window of risk opens.

Cognitive Implications

Read as cognition, dangerous behavior is a case study in how valuation and control interact. It bears on three larger themes. First, it shows dual-process valuation at work outside the laboratory: a deliberate, analytic estimate of a prospect and a fast, affective reaction to it are computed in parallel and can conflict, and the behavior reveals which won (Slovic, 1987; Loewenstein et al., 2001). Second, it makes the development of executive control visible in behavior: the adolescent risk peak is, in effect, a natural experiment on what happens when reward sensitivity runs ahead of inhibitory control, and it connects risk-taking to the broader study of self-regulation (Shulman et al., 2016). Third, it illustrates the power of representation: whether a risk is encoded as a precise probability or as a categorical gist, and whether it is learned by description or by experience, changes the choice, linking dangerous behavior to core questions about how information is mentally represented before it is acted upon (Reyna & Farley, 2006; Rosenbaum et al., 2018).

Common Misconceptions

Risk-taking is a single personality trait.

It is largely domain-specific. The correlation between a person's financial, health, recreational, ethical, and social risk-taking is low, so a single “risk appetite” score misdescribes most people (Blais & Weber, 2006).

Adolescents take risks because they don't understand the dangers.

Adolescents often estimate risks as accurately as, or even more deliberately than, adults. The peak reflects a reward system maturing ahead of the control system, not a knowledge deficit, which is why information alone changes little (Reyna & Farley, 2006; Steinberg, 2008).

Taking a risk is inherently irrational.

A risky option can carry the higher expected value; accepting a chance of loss for a larger expected gain is rational. Risk-taking is a normal cost-benefit judgment that becomes maladaptive only when perception or valuation is miscalibrated (Lejuez et al., 2002).

Glossary

Affect heuristic.

The use of the immediate feeling a prospect evokes as a cue to judging its risk and benefit, so that liked options are judged safer and disliked ones riskier.

Balloon Analogue Risk Task (BART).

A behavioral measure in which participants inflate a virtual balloon for escalating reward against a rising chance of bursting; the average pumps on intact balloons index risk-taking.

Dangerous behavior.

The MeSH umbrella for conduct that accepts a prospect of harm to oneself or others; its central entry term is risk-taking.

Domain-specificity.

The finding that risk-taking varies by area of life—financial, health, recreational, ethical, social—so that boldness in one domain does not predict boldness in another.

DOSPERT scale.

The Domain-Specific Risk-Taking scale, a self-report instrument that scores risk-taking, risk perception, and expected benefit separately within each of five domains.

Dual-systems model.

The account of adolescent risk-taking as a gap between an early-maturing reward (socioemotional) system and a late-maturing cognitive-control system.

Expected value.

The probability-weighted average of an option's possible outcomes; the benchmark against which a risky choice is judged optimal, cautious, or reckless.

Framing effect.

The finding that logically equivalent descriptions of the same prospect, cast as a gain or as a loss, reverse risk preference, so the wording of a choice can tip a person toward or away from taking a risk.

Fuzzy-trace theory.

A theory holding that decisions rest on two parallel representations, a precise verbatim trace and a categorical gist, with mature decision-makers relying more on gist.

Gist.

The bottom-line, categorical meaning of information (e.g. “this is dangerous”), as opposed to its exact quantitative detail; the representation that often guides adaptive risk avoidance.

Prospect theory.

The account of choice under risk on which outcomes are valued as gains and losses relative to a reference point, yielding risk-aversion for gains and risk-seeking for losses.

Risk as feelings.

The hypothesis that emotional reactions to a risky prospect can drive behavior directly, sometimes diverging from and overriding the cognitive evaluation of its odds and outcomes.

Risk perception.

The subjective assessment of a hazard's likelihood and severity, shaped by qualitative features such as dread, controllability, and familiarity rather than by statistics alone.

Risk-taking.

Engaging in behavior with an uncertain outcome that includes a meaningful chance of loss or harm, in exchange for a possible reward.

Self-regulation.

The cognitive-control capacity to restrain an impulse and weigh long-term consequences; its slow maturation is one half of the dual-systems imbalance.

Sensation seeking.

A heritable personality trait: the need for varied, novel, and intense experience and the willingness to take risks to obtain it.

Socioemotional system.

The reward-sensitive neural system, strongly engaged at puberty, whose early maturation drives the adolescent surge in reward-seeking.

Verbatim representation.

In fuzzy-trace theory, the precise, quantitative encoding of information (e.g. exact probabilities), as distinct from its gist.

Key Researchers

Valerie F. Reyna

. Professor of human development at Cornell University; with Frank Farley she developed fuzzy-trace theory's account of adolescent risk, arguing that reliance on gist, not superior calculation, underlies mature risk avoidance. ORCID

Paul Slovic

. Professor of psychology at the University of Oregon and founder of Decision Research; he established the psychometric paradigm of risk perception and, with colleagues, the affect heuristic that links feeling to risk judgment. ORCID

Laurence Steinberg

. Professor of psychology at Temple University; a leading architect of the dual-systems model of adolescent risk-taking and of the cross-national evidence that it generalizes worldwide. Google Scholar

Elke U. Weber

. Professor of psychology and public affairs at Princeton University; she co-authored the DOSPERT scale and the account of risk-taking as a domain-specific trade-off of perceived risk against perceived benefit. Faculty page

Marvin Zuckerman

(1928-2018). Professor of psychology at the University of Delaware; he originated the construct of sensation seeking and the scale that measures it, tying risk-taking to a heritable dimension of personality. Wikipedia

Frequently Asked Questions

What is dangerous behavior in psychology?

It is the willingness to engage in, and the tendency to carry out, actions that could harm oneself or others, the sense in which the National Library of Medicine uses the term. Its central research label is risk-taking, and it is studied as a form of decision-making under uncertainty rather than as a disorder (Blais & Weber, 2006).

Is risk-taking a single personality trait?

No. It is largely domain-specific: the same person can be cautious with money and bold in sport, so risk-taking is measured separately across financial, health, recreational, ethical, and social domains (Blais & Weber, 2006).

How do psychologists measure risk-taking?

With two complementary approaches: self-report scales such as DOSPERT that ask about likely behavior across domains, and behavioral tasks such as the Balloon Analogue Risk Task that record real choices as a risk escalates (Blais & Weber, 2006; Lejuez et al., 2002).

Why do teenagers take more risks than adults?

Because a reward-sensitive socioemotional system matures earlier than the cognitive-control system that restrains it. For a few years the gap between them widens the propensity to take risks, a pattern seen across cultures (Steinberg, 2008; Shulman et al., 2016).

What is the difference between risk perception and risk-taking?

Risk perception is the subjective judgment of how likely and severe a harm is; risk-taking is the behavior of accepting it. The two can come apart: someone may perceive a danger clearly yet still pursue the reward (Slovic, 1987).

What is sensation seeking?

A heritable personality trait: the need for varied, novel, and intense experience and the willingness to take physical, social, legal, and financial risks to obtain it. It predicts risk behavior across many domains (Zuckerman & Kuhlman, 2000).

Does taking a risk mean a person is being irrational?

Not necessarily. A risky option can have the higher expected value, in which case accepting the chance of loss is rational. Risk-taking becomes maladaptive when the perception or valuation behind it is miscalibrated, not simply because a risk was taken (Lejuez et al., 2002).

Can risk-taking ever be beneficial?

Yes. Measured willingness to accept uncertainty underlies exploration, enterprise, and learning; the developmental surge in adolescence may itself serve adaptive ends such as independence-seeking. The concern is calibration and context, not the existence of risk-taking (Figner & Weber, 2011).

References