Abstract

Impulsive behavior is a type of behavior: action taken with little forethought or tolerance of delay, typically trading a long-term cost for a short-term gain. Rather than a single trait, it is a cluster of loosely correlated tendencies that cognitive psychology fractionates into impulsive choice — the preference for a smaller-sooner reward, measured by delay discounting — and impulsive action, the failure of motor restraint measured by the stop-signal task, with self-report adding emotion-driven urgency facets. Because these dimensions load on different neural systems and dissociate across disorders, a single impulsivity score predicts outcomes poorly. This article sets out the taxonomy, the hyperbolic-discounting and race-model accounts, and the measurement debate that has led some researchers to reject the construct entirely. Three interactive demonstrations trace the discounting curve, the stop-signal race, and the trait profile.

Keywords: impulsivity, delay discounting, response inhibition, urgency, executive function

Few words in psychology are used as loosely as impulsive. In ordinary speech it labels the person who interrupts, the gambler who chases a loss, the shopper who buys on a whim, and the patient who cannot sit still — as though a single quality ran through them all. For most of the twentieth century the science followed suit, treating impulsivity as one dial that some people turn higher than others. The last three decades have dismantled that picture. When the behaviours are measured directly rather than merely rated, they come apart: the person who discounts future rewards steeply is not reliably the person who cannot stop a prepared button-press, and neither is reliably the person who acts rashly under strong emotion (Evenden, 1999; Dalley, Everitt & Robbins, 2011).

What survives is a family of related but separable tendencies, each with its own task, its own neural substrate, and its own developmental and clinical profile. Getting the distinctions right matters because impulsivity is a transdiagnostic thread running through attention-deficit/hyperactivity disorder, substance use, pathological gambling, bipolar disorder, borderline and antisocial personality, and the eating disorders (Moeller et al., 2001; Amlung et al., 2019). A treatment that targets the wrong facet — patience training for someone whose problem is motor disinhibition — will miss.

Key Takeaways
  • Impulsivity is not a single trait; it fractionates into at least two experimentally independent axes — impulsive choice and impulsive action — plus emotion-driven urgency facets.
  • Impulsive choice is captured by delay discounting: the subjective value of a reward falls hyperbolically with delay, which mathematically predicts preference reversals.
  • Impulsive action is captured by the stop-signal task, modelled as a race between a go and a stop process; the stop-signal reaction time (SSRT) estimates the latency of inhibition.
  • The facets correlate only weakly, load on different brain systems, and dissociate across disorders, so a single impulsivity score has poor predictive and construct validity.
  • Steep delay discounting is one of the most robust transdiagnostic markers in psychiatry, elevated across addictions and many other conditions.

What Impulsive Behavior Is

Impulsivity is most usefully defined not by a feeling but by a contrast with control. An impulsive act is one that is under-regulated relative to the actor's own longer-range goals: it is executed before its consequences are adequately weighed (a failure of reflection), it favours an immediate payoff over a larger deferred one (a failure of patience), or it escapes a prepared attempt to withhold it (a failure of inhibition). Each clause points to a different measurable process, and that is the first lesson of the modern literature: the everyday word bundles together things the laboratory keeps apart.

Two broad methods have grown up around the construct, and they do not agree. Self-report questionnaires — the Barratt Impulsiveness Scale (BIS-11; Patton, Stanford & Barratt, 1995) and the UPPS-P model (Whiteside & Lynam, 2001; Cyders & Smith, 2008) — ask people to rate their own tendencies and yield tidy factor structures. Behavioural tasks — discounting procedures, the stop-signal task, go/no-go — measure performance directly. A meta-analysis by Sharma, Markon and Clark (2014) found that self-report and behavioural measures of supposedly the same facet correlate near zero, and that the behavioural tasks barely correlate with one another. Impulsivity, in other words, is less a thing than a region of trait space that several imperfectly aligned instruments each illuminate from a different angle.

Figure 1

Impulsivity as Under-Regulation

Impulsivity as under-regulation A two-by-two grid crossing speed of response against regard for consequences, locating four behavioural styles; the impulsive style is the fast response emitted while its consequences go unweighed. fast / prepotent response slow / withheld response consequences weighed consequences ignored quick & adaptive (expertise, reflex) IMPULSIVE (rash, disinhibited) deliberate & controlled (planning) cautious / inhibited (deliberation)
Note. Speed of response is not itself impulsive — fast action can be expert and adaptive. Behaviour becomes impulsive only when a fast, prepotent response is emitted while its consequences go unweighed (upper right). The same two ingredients, re-crossed, give the other three styles.

Types of Impulsive Behavior

In the Medical Subject Headings vocabulary, Impulsive Behavior (D007175) sits directly under behavior itself, and MeSH records one narrower descriptor beneath it. MeSH is an indexing classification built for retrieving the biomedical literature, not a theory of how impulsivity is organised, so its single child reflects historical indexing practice rather than the two- or three-factor structure the experimental literature favours; the two ways of carving the domain are orthogonal and should not be conflated.

Narrower MeSH descriptor What it denotes
Compulsive Behavior Repetitive acts performed to reduce anxiety or distress rather than for reward; classified by MeSH as a kind of impulsive behavior, though cognitive psychology increasingly treats impulsivity and compulsivity as opposite poles of a single control dimension.

Compulsive Behavior does not yet have its own article on this site, so the term is listed here but not linked. The MeSH placement is worth a caveat of its own: contemporary models (Dalley & Robbins, 2017; Chamberlain et al., 2018) argue that compulsivity — action that persists despite being no longer rewarding — is better understood as the mirror image of impulsivity than as a species of it, the two anchoring opposite ends of a dimension that runs from under-controlled to over-controlled action.

The Two Experimental Axes

The behavioural literature converges on two dissociable systems. The first governs when a reward is worth waiting for; the second governs whether a motor response can be cancelled once set in motion. They are measured by different tasks, implemented in different cortico-striatal loops, and can be independently impaired (Bari & Robbins, 2013; Dalley, Everitt & Robbins, 2011).

Impulsive choice: delay discounting

Offered £60 today or £100 in a month, many people take the £60 — even though waiting earns a return no investment could match. The phenomenon is captured by delay discounting: the subjective value of a reward shrinks as the delay to it grows. Crucially the shrinkage is not exponential, as rational economic models assume, but hyperbolic (Ainslie, 1975; Mazur's one-parameter form):

V = A / (1 + kD)

where V is present subjective value, A the reward amount, D the delay, and k a person's discounting rate. A hyperbola falls steeply at first and then flattens, whereas an exponential falls at a constant proportional rate. That difference is not a technicality: only a hyperbola crosses itself, which is why a hyperbolic discounter will reverse a preference as a fixed pair of rewards both move closer in time — preferring the patient option when both are distant, then flipping to the impatient one as the sooner reward comes within reach. Exponential discounters never reverse. The worked example below re-derives exactly such a reversal.

Steep discounting (a high k) is the single most reliable behavioural marker of addiction: a meta-analysis across substances found robustly elevated discounting in dependent groups (MacKillop et al., 2011), and the pattern now appears to be genuinely transdiagnostic, raised across a wide range of psychiatric disorders (Amlung et al., 2019).

Delay discounting and preference reversal

A hyperbolic discounter values £60 now against £100 in 30 days. Drag the extra delay to push both rewards further off together, and watch the curves cross: the preference flips from impatient to patient even though nothing but the common delay has changed. The discount rate k sets the steepness.

Discount rate k (per day)0.050
Extra delay added to both rewards (days)+0
0255075100present value (£)extra delay added to both rewards (days)£60 sooner£100 later
At an extra delay of +0 d (k = 0.050): the £60 is worth £60.0, the £100 worth £40.0 — so this chooser prefers the smaller-sooner (£60). The curves cross at +25.0 d: below it the impatient option wins, above it the patient one does. That reversal is the signature of hyperbolic discounting.

Impulsive action: response inhibition

The second axis concerns the brakes, not the accelerator. In the stop-signal task a participant makes speeded responses to a stream of go stimuli; occasionally, a short delay after the go stimulus, a stop signal instructs them to cancel the response already under way. Logan's horse-race model treats each trial as a race between a go process and a stop process: if the stop process finishes first the response is inhibited, if the go process wins it escapes (Verbruggen & Logan, 2008). Because the stop process is not directly observable, its latency — the stop-signal reaction time, SSRT — is estimated from the probability of responding at different stop-signal delays. A long SSRT means slow brakes and is seen in ADHD, in substance use, and after damage to the right inferior frontal cortex.

Response inhibition is a component of executive function and overlaps with psychological inhibition more broadly, but the stop-signal construct is narrower and sharper: it isolates the cancellation of an already-initiated action, which is conceptually distinct from never starting it (go/no-go) and from choosing patiently (discounting).

The stop-signal race

On a stop trial a go response (whose finishing times form the curve below) races a stop process that completes at SSD + SSRT. If the go process finishes first — in the shaded tail — the response escapes; otherwise it is inhibited. Lengthen the SSRT (slower brakes) or the SSD (a later signal) and the escape probability climbs. Go-RT is fixed at a mean of 490 ms, SD 90.

Stop-signal delay SSD (ms)250
Stop-signal reaction time SSRT (ms)240
200400600800go finishing time (ms)stop finishes 490 ms
Stop process finishes at 490 ms (= 250 + 240). Probability the response escapes, p(respond | signal) = 50%; inhibited on 50% of stop trials. At the defaults (SSD 250, SSRT 240) the finish lands on the mean go-RT, giving a 50/50 race — the point from which the integration method recovers an SSRT of 240 ms.

Why one score will not do

If impulsivity were unitary, the discounting rate and the SSRT would track each other and both would track self-reported impulsiveness. They do not. The two experimental axes are largely independent, and self-report facets form yet another, loosely related, cluster. The UPPS-P model decomposes trait impulsivity into five facets — negative urgency, positive urgency, lack of premeditation, lack of perseverance, and sensation seeking — precisely because factor analysis refused to collapse them into one (Whiteside & Lynam, 2001; Cyders & Smith, 2008). The urgency facets, which capture rash action under strong negative or positive emotion, turn out to be the ones most predictive of the severest clinical outcomes, and they are almost invisible to the cold cognitive tasks. The demonstration below maps a chosen trait profile onto behaviour, showing how differently the individual facets contribute.

One number, many profiles

The UPPS-P model splits trait impulsivity into five facets that do not collapse into one. Set each independently: two very different people can share the same average score while differing entirely in which facet drives it. The correlate notes are illustrative of the literature, not computed from your input.

Negative urgency3/5
Positive urgency2/5
Lack of premeditation2/5
Lack of perseverance2/5
Sensation seeking3/5
NUPUPMPSSS
Naive global impulsivity score (facet mean): 2.4/5. That single number hides the shape below:
  • no single facet is elevated — this profile would look average on a global impulsivity score

Worked Example

Consider a hyperbolic discounter with rate k = 0.05 per day, choosing between a smaller-sooner reward of £60 and a larger-later reward of £100.

At the moment of choice, the £60 is available now (D = 0) and the £100 in 30 days:

- VSS = 60 / (1 + 0.05 × 0) = 60 / 1 = £60 - VLL = 100 / (1 + 0.05 × 30) = 100 / 2.5 = £40

The smaller-sooner reward has the higher present value, so our discounter takes the £60 — the impulsive choice.

Now delay both rewards by a further 30 days, so the £60 arrives in 30 days and the £100 in 60, and re-evaluate from today:

- VSS = 60 / (1 + 0.05 × 30) = 60 / 2.5 = £24 - VLL = 100 / (1 + 0.05 × 60) = 100 / 4 = £25

Now the larger-later reward wins, 25 to 24. Nothing about the rewards has changed except that both are further off — yet the preference has reversed from impatient to patient. This is the signature of hyperbolic, not exponential, discounting: the same person is a spendthrift up close and a saver at a distance. Setting the two present values equal at the original delays gives the indifference rate k\* = (100/60 − 1) / 30 ≈ 0.022 per day; a discounter steeper than this takes the £60, one shallower waits for the £100. The `DelayDiscountingDemo` above traces both curves and marks the crossing point where the reversal occurs.

Discussion

The fractionation of impulsivity is now the mainstream view, but it raises a sharp question that Strickland and Johnson (2021) have pressed to its conclusion: if the facets neither cohere empirically nor share a mechanism, is impulsivity a scientific construct at all, or merely a folk term that survives because it is convenient? Their argument is that the word imports a spurious unity — a single explanatory entity, impulsivity, is invoked to explain behaviours that in fact have distinct causes — and that the field would be better served by naming the specific processes (steep discounting, slow SSRT, high negative urgency) and dropping the umbrella. This is not a fringe position; it is the logical endpoint of three decades of dissociation evidence.

The counter-case is pragmatic. The facets, however loosely bound, cluster together far more often than chance in clinical populations, and the umbrella term earns its keep as a signpost toward a family of related vulnerabilities even if it names no single mechanism. The transdiagnostic reach of delay discounting (Amlung et al., 2019) suggests that something general is captured, even if that something is better described as a dimension of self-regulation than as a trait. Either way, the practical injunction is the same: measure the specific facet of interest, never a global score, and never assume that improving one will improve another.

Current Directions

Three active fronts stand out. The first is the impulsivity–compulsivity dimension: large-sample latent-variable work (Chamberlain et al., 2018) supports a continuum from under-controlled to over-controlled action that cuts across diagnostic categories, reframing conditions as varied as ADHD and obsessive-compulsive disorder as different points on shared control dimensions — part of the wider move toward dimensional, rather than categorical, psychiatry. The second is urgency: because the emotion-driven facets predict self-harm, binge behaviour and relapse better than the cold measures do, emotion-regulation-focused interventions are being tested as facet-specific treatments. The third is computational phenotyping — replacing a single SSRT or k with parameters recovered from formal models of the whole task (race-model and reinforcement-learning fits), on the argument that model parameters are more mechanistically interpretable and more stable markers than summary scores. Whether these parameters finally deliver the clean, mechanism-linked measures the field has sought — or simply confirm Strickland and Johnson's (2021) suspicion that there is no unitary trait to find — is the open question.

Common Misconceptions

Impulsivity is a single personality trait.
Its defining empirical feature is the opposite: the laboratory measures of its supposed facets correlate near zero with one another and with self-report (Sharma, Markon & Clark, 2014). It is a loosely bound family, not a trait.
Acting fast is being impulsive.
Speed alone is not impulsivity. Expert, reflexive, and well-practised actions are fast and fully adaptive. Behaviour is impulsive only when a prepotent response escapes regulation relative to the actor's own goals (Evenden, 1999).
Impulsive people simply discount the future exponentially steeply.
The discounting is hyperbolic, not exponential, and the distinction is the whole point: only hyperbolic curves cross, producing the preference reversals that exponential models cannot (Ainslie, 1975).
Impulsivity and compulsivity are simply unrelated opposites.
They are increasingly modelled as two ends of one control dimension — under-regulated versus rigidly over-regulated action — not as unrelated states, and the same person can show both (Dalley & Robbins, 2017).

Glossary

Barratt Impulsiveness Scale (BIS-11).
The most widely used self-report measure of trait impulsiveness, yielding attentional, motor, and non-planning subscales.
Compulsivity.
The tendency to perform repetitive acts that persist despite being no longer rewarding; increasingly modelled as the over-regulated mirror of impulsivity on a single control dimension.
Delay discounting.
The decline in the subjective value of a reward as the delay to receiving it increases; the standard behavioural measure of impulsive choice.
Discounting rate (k).
The single parameter of the hyperbolic discounting function; a higher k means steeper devaluation of delayed rewards and more impulsive choice.
Exponential discounting.
Discounting at a constant proportional rate, as assumed by rational economic models; unlike hyperbolic discounting it never produces preference reversals.
Go/no-go task.
A paradigm in which participants respond to go stimuli and withhold to no-go stimuli; indexes action restraint, the failure of which is a commission error.
Hyperbolic discounting.
Discounting in which value falls as 1/(1 + kD); unlike exponential discounting it predicts preference reversals.
Impulsive action.
Failure to withhold or cancel a motor response, measured by the stop-signal and go/no-go tasks; distinct from impulsive choice.
Impulsive choice.
Preference for a smaller-sooner over a larger-later reward; measured by delay discounting.
Iowa Gambling Task.
A card-selection task indexing decision making under uncertainty; poor performance is linked to ventromedial prefrontal damage and insensitivity to future consequences (Bechara et al., 1994).
Negative urgency.
The UPPS-P facet capturing the tendency to act rashly under strong negative emotion; among the strongest facet-level predictors of severe clinical outcomes.
Positive urgency.
The UPPS-P facet capturing the tendency to act rashly under strong positive emotion.
Preference reversal.
A switch in which of two rewards is preferred as both move nearer or further in time; the behavioural signature of hyperbolic discounting.
Response inhibition.
The suppression of a prepared or ongoing action; a component of executive function.
Sensation seeking.
The UPPS-P facet capturing a preference for novel and intense experience; only weakly tied to harmful outcomes.
Stop-signal reaction time (SSRT).
The estimated latency of the covert stop process in the stop-signal task; longer SSRTs indicate poorer response inhibition.
Stop-signal task.
A paradigm requiring cancellation of an already-initiated response when a stop signal appears; modelled as a race between a go and a stop process.
Transdiagnostic.
Describing a process, such as steep delay discounting, that is elevated across many distinct psychiatric diagnoses rather than being specific to one.
UPPS-P model.
A five-facet model of trait impulsivity: negative urgency, positive urgency, lack of premeditation, lack of perseverance, and sensation seeking.
Urgency.
The UPPS-P facets capturing the tendency to act rashly under strong negative (negative urgency) or positive (positive urgency) emotion.

Key Researchers

George Ainslie

(Temple University School of Medicine; Coatesville VA Medical Center). Demonstrated preference reversal and hyperbolic discounting of delayed reward, the formal core of intertemporal impulsiveness, and founded the field of picoeconomics. Wikipedia · Homepage

Ernest S. Barratt

(University of Texas Medical Branch, Galveston; 1925–2005). The founder of modern impulsivity research. He built the first impulsiveness scale in 1959, which evolved into the widely used BIS-11, and founded the International Society for Research on Impulsivity. Memorial

Antoine Bechara

(University of Southern California). Developed the Iowa Gambling Task and the somatic-marker account of decision making under uncertainty, linking ventromedial prefrontal damage to insensitivity to future consequences. Wikipedia

Melissa A. Cyders

(Indiana University Indianapolis). With Gregory Smith, introduced the emotion-based urgency traits — positive and negative urgency — that extended the UPPS model and reshaped how trait impulsivity is measured. ORCID · Faculty

Jeffrey W. Dalley

(University of Cambridge). Established the rodent neural substrates of waiting (motor) impulsivity and trait-like premature responding, linking impulsive action to striatal dopamine function. Faculty · Google Scholar

Trevor W. Robbins

(University of Cambridge). With Dalley and Everitt, framed the influential modern dissociation of impulsivity, compulsivity, and top-down cognitive control, and mapped its cortico-striatal substrates. ORCID · Wikipedia

Frequently Asked Questions

Is impulsivity the same as a lack of willpower?

No. Willpower is a folk notion of a single self-control resource; impulsivity, as measured, splits into independent components (impatience, disinhibition, urgency) that do not share a common resource. Someone can have excellent motor restraint and still discount the future steeply.

What is the difference between impulsive choice and impulsive action?

Impulsive choice is about valuation over time: preferring a smaller reward now to a larger one later, measured by delay discounting. Impulsive action is about motor control: failing to withhold or cancel a response, measured by the stop-signal task. They are largely independent.

Why does hyperbolic discounting matter more than exponential?

Because hyperbolic curves cross and exponential curves do not. Only hyperbolic discounting predicts the everyday preference reversals in which a person resolves to wait, then abandons the plan as the tempting option draws near.

What does the stop-signal reaction time (SSRT) measure?

It estimates how long the covert process that cancels an action takes to finish. Because the stop process cannot be observed directly, the SSRT is inferred from how often people fail to stop at different stop-signal delays, using the race model.

Is impulsivity always bad?

No. Rapid, decisive action is adaptive in many settings, and some facets (such as sensation seeking) are only weakly tied to harm. Impulsivity becomes a problem when it is chronic, pervasive, and misaligned with a person's own longer-term goals.

Impulsiveness is one of the core symptom domains of ADHD, and the disorder is associated with both steeper discounting and longer SSRTs, though, consistent with fractionation, not every individual shows both.

Can impulsivity be reduced?

Facet-specific interventions show promise: episodic future thinking and commitment devices can reduce discounting, inhibitory-control training targets response inhibition, and emotion-regulation therapies target the urgency facets. A global cure is unlikely because there is no single mechanism to fix.

Why do some researchers want to abandon the term entirely?

Because the facets do not cohere empirically or mechanistically, Strickland and Johnson (2021) argue that treating impulsivity as one construct imports a false unity, and that science would progress faster by naming the specific processes directly.

References

Ainslie, G. (1975). Specious reward: A behavioral theory of impulsiveness and impulse control. Psychological Bulletin, 82(4), 463–496. https://doi.org/10.1037/h0076860

Amlung, M., Marsden, E., Holshausen, K., Morris, V., Patel, H., Vedelago, L., Naish, K. R., Reed, D. D., & McCabe, R. E. (2019). Delay discounting as a transdiagnostic process in psychiatric disorders: A meta-analysis. JAMA Psychiatry, 76(11), 1176–1186. https://doi.org/10.1001/jamapsychiatry.2019.2102

Bari, A., & Robbins, T. W. (2013). Inhibition and impulsivity: Behavioral and neural basis of response control. Progress in Neurobiology, 108, 44–79. https://doi.org/10.1016/j.pneurobio.2013.06.005

Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50(1–3), 7–15. https://doi.org/10.1016/0010-0277(94)90018-3

Chamberlain, S. R., Stochl, J., Redden, S. A., & Grant, J. E. (2018). Latent traits of impulsivity and compulsivity: Toward dimensional psychiatry. Psychological Medicine, 48(5), 810–821. https://doi.org/10.1017/s0033291717002185

Cyders, M. A., & Smith, G. T. (2008). Emotion-based dispositions to rash action: Positive and negative urgency. Psychological Bulletin, 134(6), 807–828. https://doi.org/10.1037/a0013341

Dalley, J. W., Everitt, B. J., & Robbins, T. W. (2011). Impulsivity, compulsivity, and top-down cognitive control. Neuron, 69(4), 680–694. https://doi.org/10.1016/j.neuron.2011.01.020

Dalley, J. W., & Robbins, T. W. (2017). Fractionating impulsivity: Neuropsychiatric implications. Nature Reviews Neuroscience, 18(3), 158–171. https://doi.org/10.1038/nrn.2017.8

Evenden, J. L. (1999). Varieties of impulsivity. Psychopharmacology, 146(4), 348–361. https://doi.org/10.1007/pl00005481

MacKillop, J., Amlung, M. T., Few, L. R., Ray, L. A., Sweet, L. H., & Munafò, M. R. (2011). Delayed reward discounting and addictive behavior: A meta-analysis. Psychopharmacology, 216(3), 305–321. https://doi.org/10.1007/s00213-011-2229-0

Moeller, F. G., Barratt, E. S., Dougherty, D. M., Schmitz, J. M., & Swann, A. C. (2001). Psychiatric aspects of impulsivity. American Journal of Psychiatry, 158(11), 1783–1793. https://doi.org/10.1176/appi.ajp.158.11.1783

Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt Impulsiveness Scale. Journal of Clinical Psychology, 51(6), 768–774. https://doi.org/10.1002/1097-4679(199511)51:6<768::aid-jclp2270510607>3.0.co;2-1

Sharma, L., Markon, K. E., & Clark, L. A. (2014). Toward a theory of distinct types of impulsive behaviors: A meta-analysis of self-report and behavioral measures. Psychological Bulletin, 140(2), 374–408. https://doi.org/10.1037/a0034418

Strickland, J. C., & Johnson, M. W. (2021). Rejecting impulsivity as a psychological construct: A theoretical, empirical, and sociocultural argument. Psychological Review, 128(2), 336–361. https://doi.org/10.1037/rev0000263

Verbruggen, F., & Logan, G. D. (2008). Response inhibition in the stop-signal paradigm. Trends in Cognitive Sciences, 12(11), 418–424. https://doi.org/10.1016/j.tics.2008.07.005

Whiteside, S. P., & Lynam, D. R. (2001). The Five Factor Model and impulsivity: Using a structural model of personality to understand impulsivity. Personality and Individual Differences, 30(4), 669–689. https://doi.org/10.1016/s0191-8869(00)00064-7