Problem Behavior

Abstract

Problem behavior is a type of behavioral symptom: a troublesome or disruptive display that deviates enough from age-expected conduct to draw concern or clinical attention. Cognitive psychology treats it not as a unitary trait but as a small set of empirically derived syndromes, resolving childhood problems through factor analysis into two broad dimensions, externalizing (undercontrolled, acting-out) and internalizing (overcontrolled, distressed), which later work orders inside a hierarchy rising to a general psychopathology factor. This article covers the two-syndrome structure, the coercion model and self-regulation deficits that maintain acting-out, the adolescence-limited versus life-course-persistent developmental taxonomy, and the dimensional turn. What remains contested is whether these dimensions are causal kinds or statistical summaries, and how far a general factor reflects a real shared process. Three demonstrations make the mechanisms manipulable: a coercion-trap simulator, an externalizing/internalizing score profiler, and a general-factor variance splitter.

Keywords: behavioral symptoms, externalizing, internalizing, coercion

What Problem Behavior Is

In MeSH, Problem Behavior (introduced 2016) is defined as troublesome or disruptive behavioral displays, and the descriptor is filed under both behavioral symptoms and child behavior — the twin parents that frame the construct. The label is deliberately broad because the phenomena it names resist a single definition: the same act (a tantrum, a truancy, a fight) is a problem in one context and unremarkable in another, and judgments of excess are made against an age norm rather than an absolute threshold. Cognitive and developmental psychology handle this by abandoning the search for a natural kind and asking instead which behaviors statistically covary — which tend to occur together in the same children — and then naming the resulting clusters.

That empirical move is the organizing idea of the whole field. Rather than theorize categories first, Achenbach (1966) factor-analyzed the symptoms of several hundred child psychiatric cases and found that problems collapse onto two broad, roughly orthogonal dimensions. The first, later called externalizing, loads behaviors directed outward at the environment: aggression, defiance, rule-breaking, overactivity. The second, internalizing, loads behaviors directed inward: anxiety, withdrawal, somatic complaint, dysphoria. A child can score high on one, both, or neither, and the two dimensions have different correlates, courses, and treatments. This two-factor structure has replicated across cultures, informants, and decades and is the backbone of the Achenbach System of Empirically Based Assessment used worldwide.

The phrase problem behavior in everyday and clinical usage most often denotes the externalizing pole, and the rest of this article concentrates there while keeping the two-dimensional frame in view. The crucial point is that externalizing is a dimension, not a diagnosis: a given child sits somewhere on a continuum of severity, and the clinical labels (oppositional defiant disorder, conduct disorder) mark conventional cut-points on that continuum rather than discrete diseases (Fairchild et al., 2019).

The Two Broad Syndromes

The externalizing and internalizing dimensions are separable but correlated, and even within externalizing there is finer structure. Hinshaw (1987) showed that the two most-studied externalizing problems — attention/hyperactivity on one side and conduct/aggression on the other — are statistically distinguishable yet frequently co-occur, a pattern of separability-with-comorbidity that recurs whenever externalizing is decomposed. The practical upshot is that a child who cannot sit still is not thereby a child who breaks rules, even though the two often travel together; the mechanisms, prognoses, and responses to treatment differ. Table 1 summarizes the broadband dimensions and their principal narrow syndromes.

Because the dimensions are continuous, they are measured with norm-referenced scales. A child's raw symptom count is converted to a standardized T-score (mean 50, standard deviation 10) against an age-and-sex reference sample, so that a score expresses how far the child sits from the typical peer rather than how many symptoms were endorsed in absolute terms. A T of 70 — two standard deviations above the mean — is the conventional clinical threshold. The first demonstration below converts raw endorsements into externalizing and internalizing T-scores and marks where the clinical cut falls.

Externalizing / internalizing score profiler

Externalizing T = 70.0 (clinical), internalizing T = 60.0 (normal). Each raw count is standardized against its reference norm (externalizing mean 8, SD 5; internalizing mean 7, SD 4) so the two dimensions are comparable and the T = 70 line marks the conventional clinical threshold.

Reference norms are illustrative of the ASEBA framework, not published standardization values.

How Acting-Out Is Maintained: Coercion

Knowing that externalizing behavior forms a dimension says nothing about why it grows in a particular child. The most influential mechanistic account is Patterson's coercion model (Patterson et al., 1989), which locates the engine of antisocial behavior not in the child's personality but in the moment-to-moment contingencies of family interaction. The core unit is the coercive exchange: a parent issues a demand, the child responds with an aversive protest (whining, arguing, a tantrum), and the parent — to end the unpleasantness — withdraws the demand. The child's protest is thereby negatively reinforced: it was followed by the removal of an aversive stimulus, so it becomes more probable next time. Symmetrically, the parent's capitulation is reinforced by the immediate cessation of the child's protest. Both parties are trained by the exchange to escalate.

Repeated thousands of times across early childhood, this reinforcement trap ratchets the child's aversive behavior upward and the parent's effective demands downward. The model is a direct application of operant reinforcement principles to the family, and it predicts that the behavior will generalize from home to school and peer settings as the child exports a coercive interpersonal style. Harsh, inconsistent discipline accelerates the trap rather than braking it: a meta-analysis of corporal punishment found it associated with more child aggression and externalizing behavior, not less (Gershoff, 2002), because an angry smack both models aggression and, when it fails to stop the behavior, teaches the child that escalation pays.

The second demonstration simulates the coercion trap. Each reinforced bout raises the probability of the child's aversive behavior by a fixed increment; the probability climbs with capitulation, and consistent (non-capitulating) parenting flattens the curve.

The coercion trap

Each time the parent withdraws the demand, the child’s protest is negatively reinforced and its probability rises by 0.12, climbing from 0.20 to 0.92 over 6 bouts.

A deterministic illustration of Patterson’s coercion model; the linear increment is a didactic simplification of the reinforcement process.

The Cognitive Substrate: Self-Regulation

Coercion explains how the environment shapes externalizing behavior; self-regulation explains why some children are more susceptible to that shaping than others. Eisenberg and colleagues (2001) showed that two temperamental qualities jointly predict externalizing problems: low effortful control (the capacity to inhibit a dominant response and perform a subdominant one) and high negative emotionality. A child who is easily angered and poor at braking the resulting impulse is primed to enter and lose coercive exchanges. Effortful control is the temperamental precursor of the executive inhibition that matures through childhood, which is why externalizing problems often attenuate as the prefrontal control system comes online.

The difficulty is that the term self-regulation names a tangle of overlapping constructs — self-control, executive function, cognitive control, impulsivity, delay of gratification — that researchers have used almost interchangeably. Nigg (2017) disentangled them, arguing that externalizing psychopathology is best understood as a failure in a specific, definable control system rather than a vague deficit of willpower. The payoff of precision is mechanistic: a child whose problem is weak response inhibition needs a different intervention from one whose problem is insensitivity to delayed reward, even though both present as impulsive acting-out. This regulatory account also links the externalizing dimension to emotional regulation, since the behaviors that get a child into trouble are frequently failures to modulate anger and frustration under provocation.

Regulation is only half of the cognitive story; the other half is how the child reads the social situation in the first place. Crick and Dodge (1994) reformulated aggression as the output of a sequence of social information-processing steps — encoding the cues in a situation, interpreting them, accessing possible responses, and selecting one — and showed that aggressive children systematically distort early steps, most notably by over-attributing hostile intent to ambiguous acts (the hostile attribution bias). On this account acting-out is not only a failure to inhibit a provoked impulse but a failure to process the provocation accurately: a child who reads a neutral bump as a deliberate attack has already selected the target of an aggressive response before any question of control arises. The social information-processing model is the dominant cognitive complement to the temperamental and operant accounts, locating part of the externalizing dimension in the child's on-line construal of the social world.

Development: Trajectories, Not Snapshots

A single assessment cannot distinguish the child whose problems will remit from the one whose problems will persist, and the distinction matters enormously. Moffitt's developmental taxonomy (1993) proposed two qualitatively different trajectories of antisocial behavior that look similar in mid-adolescence but have opposite origins and outcomes. The adolescence-limited path is common, normative, and transient: it begins around puberty, is driven by a maturity gap between biological and social adulthood, peaks in the teens, and desists in early adulthood as adult roles become available. The life-course-persistent path is rare, begins in early childhood, is rooted in neurocognitive and temperamental vulnerability interacting with adverse rearing, and is stable from preschool into adulthood.

The taxonomy reframes the clinical task as one of prognosis: the mid-adolescent peak in antisocial behavior is mostly the adolescence-limited majority, and treating every acting-out teenager as a budding life-course offender overpathologizes a normal developmental phase. Campbell (1995) made the same point from the other end of development, reviewing which hard-to-manage preschoolers are showing a transient difficulty that will resolve and which minority are showing the early stage of a persistent problem. Early onset, pervasiveness across settings, and co-occurring hyperactivity all predict persistence; situational, late-onset problems usually do not.

Jessor's Problem Behavior Theory (1991) adds the observation that in adolescence the various problem behaviors — delinquency, early sexual activity, substance use, risky driving — tend to covary, forming a loose syndrome embedded in a psychosocial system of risk and protective factors rather than a set of independent habits. A teenager at risk for one tends to be at risk for the others, because they share a common web of social and personal antecedents. This is why prevention programs target the shared risk system (peer context, parental monitoring, school engagement) rather than chasing each behavior one at a time.

The Dimensional Turn: From Syndromes to a Hierarchy

The two broadband dimensions are not the top of the structure. Beauchaine and McNulty (2013) recast the externalizing syndromes as a single developmental spectrum that unfolds from a shared trait vulnerability (early irritability and weak regulation) through a sequence of ontogenic processes, so that oppositional behavior, conduct problems, and substance use are successive expressions of one underlying liability rather than separate disorders. Above even the broadband dimensions sits a general factor. Caspi and colleagues (2014) found that a single general psychopathology (p) factor accounts for the pervasive comorbidity among mental disorders: externalizing, internalizing, and thought-disorder dimensions all load on it, so a child high on p is at elevated risk across the board. The externalizing dimension is thus the shared variance of its narrow syndromes, and p is the shared variance of the broad dimensions — a nested hierarchy of ever-wider abstractions.

This hierarchical, dimensional view is now formalized. The Hierarchical Taxonomy of Psychopathology (Kotov et al., 2017) places the externalizing spectrum inside an empirically derived hierarchy that replaces the discrete categories of traditional diagnosis with continuous dimensions at several levels of breadth. Problem behavior, on this account, is a region of a quantitative psychopathology space, and a child's standing is a profile of scores rather than a list of present-or-absent diagnoses.

The third demonstration makes the general-factor idea concrete: given a child's loading on the general factor, it splits the variance of an observed externalizing score into the part shared with all psychopathology and the part specific to externalizing.

General-factor variance split

A standardized loading of 0.60 on the general p factor means 36% of the observed externalizing variance (loading², = 0.36) is shared broad liability and the remaining 64% (0.64) is specific to externalizing. The higher the loading, the more a child’s acting-out is a marker of general psychopathology risk rather than a distinct problem.

The variance decomposition follows directly from a single-factor model; the loading is illustrative, not an estimate from a specific sample.

Worked Example

Consider Sam, an eight-year-old referred for disruptive behavior, and trace the three mechanisms quantitatively.

Measurement. A caregiver rates Sam on a problem-behavior checklist. On the externalizing items he endorses a raw score of 18 against a reference mean of 8 and standard deviation of 5; on the internalizing items, a raw of 11 against a mean of 7 and standard deviation of 4. Converting to T-scores:

- Externalizing T = 50 + 10 × (18 − 8) / 5 = 50 + 10 × 2.0 = 70.0, exactly at the clinical threshold. - Internalizing T = 50 + 10 × (11 − 7) / 4 = 50 + 10 × 1.0 = 60.0, elevated but subclinical.

Sam's profile is predominantly externalizing — the acting-out pole — with mild internalizing distress, the separable-but-correlated pattern Hinshaw (1987) described.

Maintenance. At home, Sam's protests are intermittently reinforced. Model the probability that he escalates a demand into a tantrum as starting at 0.20 and rising 0.12 with each bout the caregiver ends by capitulating. After one reinforced bout it is 0.32; after two, 0.44; after three, 0.56; after four, 0.68. Four capitulations have more than tripled the odds of a tantrum — the coercion trap Patterson et al. (1989) formalized. Holding the demand (no capitulation) leaves the probability flat at 0.20.

Structure. Suppose Sam's externalizing score loads on the general p factor with a standardized loading of 0.60. The share of his externalizing variance explained by the general factor is 0.60² = 0.36, leaving 1 − 0.36 = 0.64 specific to externalizing. Most of what makes Sam's acting-out distinctive is specific, but a third of it is shared liability that would also raise his internalizing risk (Caspi et al., 2014) — which is why the mild internalizing elevation is unsurprising rather than coincidental.

The three numbers — a T of 70, an escalation probability of 0.68, and a 36/64 variance split — are the measurement, maintenance, and structure of one child's problem behavior, and each corresponds to a control manipulable in the demonstrations above.

Discussion

The cognitive-psychological account of problem behavior is unified by a single methodological commitment: let the data define the categories. From Achenbach's original factor analysis to the HiTOP consortium, the field has preferred empirically derived dimensions to clinically stipulated types, and the payoff is a structure that predicts course and treatment response better than the discrete diagnoses it supplements. The externalizing dimension that most people call problem behavior is maintained proximally by coercive reinforcement, enabled by weak self-regulation, shaped developmentally into transient or persistent trajectories, and situated structurally inside a hierarchy that rises to a general liability.

Two caveats temper the picture. First, the dimensions are descriptive, not etiological: knowing that a child scores high on externalizing does not identify the cause, which may be temperamental, familial, or both, and the same score can arise by different routes. Second, the normative basis of the construct means that what counts as a problem is partly a cultural and contextual judgment — the behavior that is disruptive in a quiet classroom may be adaptive in a dangerous neighborhood — so the measurement framework should inform clinical judgment, not replace it (Fairchild et al., 2019).

Current Directions

Three fronts are active. The first is the push to pin down which control process fails in externalizing problem behavior, replacing the umbrella term self-regulation with specific, separately measurable components; Nigg's (2017) taxonomy of self-regulation, self-control, executive function, effortful control, and impulsivity is the reference statement, and the work now is to map each component to a distinct intervention. The second is the consolidation of dimensional nosology: the Hierarchical Taxonomy of Psychopathology (Kotov et al., 2017) is being validated against clinical outcomes, and the question is whether a continuous externalizing-spectrum score predicts prognosis and treatment response better than a categorical conduct-disorder diagnosis. The third is the clinical endpoint: the modern review of conduct disorder (Fairchild et al., 2019) centers callous-unemotional traits as a specifier that identifies a more severe, more heritable, less treatment-responsive subgroup within severe externalizing behavior, and the open problem is whether that subgroup needs a mechanistically different treatment rather than more of the same. Across all three, the trend is away from problem behavior as a single target and toward a profile of measurable, separately tractable components.

Key Researchers

Thomas M. Achenbach

(b. 1940, University of Vermont). Derived the empirical externalizing/internalizing syndromes from factor analysis and built the ASEBA assessment system that operationalizes them worldwide. ORCID · Wikipedia

Avshalom Caspi

(Duke University; King's College London). Identified the general psychopathology (p) factor onto which externalizing and internalizing problem behavior both load. ORCID · Wikipedia

Nancy Eisenberg

(Arizona State University). Established effortful control and emotion regulation as temperamental predictors of externalizing and internalizing problem behavior. Faculty profile · Wikipedia · Google Scholar

Stephen P. Hinshaw

(University of California, Berkeley; UCSF). Clarified the separability yet comorbidity of the attention/hyperactivity and conduct/aggression dimensions of externalizing problems. ORCID · Wikipedia

Richard Jessor

(University of Colorado Boulder). Formulated Problem Behavior Theory, showing that adolescent problem behaviors covary as a psychosocially embedded syndrome of risk and protection. ORCID · Faculty page

Terrie E. Moffitt

(Duke University; King's College London). Proposed the adolescence-limited versus life-course-persistent developmental taxonomy of antisocial behavior. ORCID · Wikipedia

Joel T. Nigg

(Oregon Health & Science University). Disentangled the overlapping self-regulation constructs whose failure is the proximal cognitive substrate of externalizing problem behavior. ORCID · Google Scholar

Gerald R. Patterson

(1926–2016, Oregon Social Learning Center). Developed the coercion model, explaining antisocial behavior through moment-to-moment coercive family exchanges that negatively reinforce a child's aversive conduct. Institutional biography

Glossary

Adolescence-limited.

A developmental trajectory of antisocial behavior that onsets at puberty, is normative and transient, and remits in early adulthood.

ASEBA.

The Achenbach System of Empirically Based Assessment, a family of norm-referenced questionnaires that score child and adult problems on empirically derived syndrome scales.

Callous-unemotional traits.

A specifier marking lack of empathy, guilt, and emotional responsiveness that identifies a more severe, more heritable subgroup within conduct problems.

Coercion model.

Patterson's account in which coercive family exchanges negatively reinforce a child's aversive behavior, escalating it over repeated bouts.

Conduct disorder.

The clinical endpoint of severe, persistent externalizing behavior that violates the rights of others or major age-appropriate norms.

Effortful control.

The temperamental capacity to inhibit a dominant response and perform a subdominant one; low effortful control predicts externalizing problems.

Externalizing.

The broadband dimension of undercontrolled, outward-directed problem behavior: aggression, defiance, rule-breaking, overactivity.

Hostile attribution bias.

The tendency of aggressive children to interpret ambiguous social acts as deliberately hostile, a distortion in the interpretation step of social information processing.

Internalizing.

The broadband dimension of overcontrolled, inward-directed problems: anxiety, withdrawal, somatic complaint, dysphoria.

Life-course-persistent.

A rare developmental trajectory of antisocial behavior rooted in early neurocognitive vulnerability and stable from childhood into adulthood.

Negative reinforcement.

The strengthening of a behavior by the removal of an aversive stimulus; the operant process at the heart of the coercion trap.

p factor.

A general psychopathology factor onto which externalizing, internalizing, and thought-disorder dimensions all load, indexing broad shared liability.

Problem Behavior Theory.

Jessor's framework in which adolescent problem behaviors covary as a syndrome embedded in a psychosocial system of risk and protective factors.

Self-regulation.

The umbrella set of control processes — executive function, inhibition, impulse and delay control — whose failure is the proximal substrate of acting-out.

Social information processing.

Crick and Dodge's sequential model of how children encode, interpret, and respond to social cues; systematic distortions in its early steps predict aggression.

Syndrome.

In this literature, a cluster of problems that statistically covary, derived by factor analysis rather than stipulated by diagnostic convention.

T-score.

A standardized score (mean 50, SD 10) expressing a child's standing relative to an age-and-sex reference sample; 70 is the conventional clinical threshold.

Frequently Asked Questions

Is problem behavior a diagnosis?

No. It is a descriptive construct, a troublesome or disruptive display judged against an age norm, that cognitive psychology resolves into continuous dimensions. Clinical diagnoses such as oppositional defiant disorder and conduct disorder mark conventional cut-points on the externalizing dimension rather than discrete diseases (Fairchild et al., 2019).

What is the difference between externalizing and internalizing problems?

Externalizing problems are undercontrolled and directed outward (aggression, defiance, rule-breaking); internalizing problems are overcontrolled and directed inward (anxiety, withdrawal, dysphoria). Factor analysis first separated these two broad dimensions (Achenbach, 1966), and a child can score high on one, both, or neither.

Why do some children's problems get worse over time?

The most influential answer is coercion: when a parent withdraws a demand to stop a child's protest, the protest is negatively reinforced and becomes more probable, escalating across repeated exchanges (Patterson et al., 1989). Harsh, inconsistent discipline accelerates rather than brakes this trap (Gershoff, 2002).

Does acting-out in adolescence predict lifelong antisocial behavior?

Usually not. Moffitt's taxonomy distinguishes a common, transient adolescence-limited course that remits in early adulthood from a rare, stable life-course-persistent course that begins in early childhood (Moffitt, 1993). Most teenage acting-out is the adolescence-limited kind.

What cognitive process underlies externalizing behavior?

A failure of self-regulation. Low effortful control and high negative emotionality jointly predict externalizing problems (Eisenberg et al., 2001), and the broad construct of self-regulation resolves into specific control components such as inhibition, executive function, and delay of gratification, each separately measurable (Nigg, 2017).

Can problem behavior be measured objectively?

It is measured norm-referenced rather than absolutely. Raw symptom counts are converted to standardized T-scores against an age-and-sex reference sample, so a score expresses how far a child sits from the typical peer; a T of 70 is the conventional clinical threshold (Achenbach, 1966).

What is the p factor?

A general psychopathology factor that captures the shared variance among all mental-health problems. Externalizing, internalizing, and thought-disorder dimensions all load on it, so a child high on p is at elevated risk across the board (Caspi et al., 2014).

How does the modern dimensional model differ from traditional diagnosis?

It replaces discrete categories with empirically derived continuous dimensions arranged in a hierarchy. The Hierarchical Taxonomy of Psychopathology places the externalizing spectrum inside a quantitative structure of several levels of breadth rather than a list of present-or-absent disorders (Kotov et al., 2017).

References