Social Learning and Criminal Behavior
Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy
Social learning and criminal behavior concerns the processes through which relationships, observed actions, learned beliefs, and behavioral consequences may contribute to unlawful conduct. In Ronald L. Akers' influential criminological formulation, people encounter models of behavior, acquire interpretations of what is acceptable, experience rewards or disapproval, and revise their actions through social interaction. These processes can support offending, but they can also support lawful conduct, rehabilitation, and desistance.
The most important empirical distinction is between a plausible learning mechanism and evidence that it actually contributes to behavior. A major meta-analysis of social learning theory found that differential association and favorable definitions of offending were more strongly associated with crime and deviance than measured imitation and differential reinforcement. A longitudinal network meta-analysis additionally found evidence both that friends influence behavior and that people choose friends who resemble them. Social learning therefore belongs within a multicausal account of offending, supported by specific findings and bounded by their limitations.
What is social learning theory of crime?
Social learning theory of crime explains how social interaction contributes to acquiring, maintaining, or changing behavior that violates criminal law. Edwin H. Sutherland's differential association theory supplied a foundational account of learning through intimate social contacts; Ronald L. Akers extended that approach by specifying reinforcement, modeling, and the learned meanings attached to actions. The criminological formulation is detailed in Akers and Jennings' handbook chapter, while the Oxford Handbook of Criminological Theory situates learning approaches within developmental and correctional psychology.
The theory studies processes, not a criminal personality type. Criminal behavior is a category defined by applicable law; deviance is a broader social category, which can include actions that are unusual or disapproved without being illegal. Delinquency often refers to prohibited or antisocial conduct by young people, but researchers use different measures. An arrest, self-reported offense, charge, conviction, and assessment score are also different outcomes.
Social learning is a framework for explaining possible influences on conduct. It cannot establish whether a named person broke the law, prove a particular motive, supply a forensic diagnosis, or determine criminal responsibility. Those conclusions require their own evidence and legal standards.
The four elements of Akers' social learning theory
1. Differential association: who participates in the learning environment?
Differential association concerns repeated interaction with people whose attitudes, routines, opportunities, or behavior favor or discourage offending. Friends, partners, relatives, colleagues, co-participants in an activity, and online connections may supply different kinds of exposure. The nature of a relationship matters more than a simple count of acquaintances: its duration, intensity, frequency, and significance can alter what the person learns.
For instance, peers may communicate that minor theft is acceptable, provide opportunities to participate, reward risk-taking, or oppose the behavior. A young person can also learn from friends who disapprove of illegal conduct and support alternatives. Relationships are a setting where competing definitions of behavior are encountered.
Association is not identical to influence. A person who already commits certain acts may select others with similar interests, and a person may leave a network as their behavior changes. Being acquainted with someone who offended is neither evidence of guilt nor a valid substitute for assessing the individual's own conduct.
2. Definitions: how an action becomes acceptable or unacceptable
Definitions are learned beliefs, attitudes, rationalizations, and moral evaluations of particular actions. Someone might regard a behavior as wrong in general but construct a justification for a special circumstance. Social groups can also convey expectations about loyalty, retaliation, status, and fairness.
Researchers measure definitions with questions about approval, permissibility, excuses, and perceived norms. These are psychological constructs, not formal legal defenses. A respondent can endorse a hypothetical justification without performing an act; conversely, someone can offend despite reporting disapproval. The Pratt et al. meta-analysis found strong associations involving definitions, but the measures and study designs differed, and association alone does not establish which belief came first.
A person may acquire a favorable belief before an offense, offer an excuse after an offense, or voice an opinion to obtain social acceptance. These possibilities require temporal evidence rather than inference from a single interview.
3. Differential reinforcement: what happens after behavior
Differential reinforcement involves the actual or anticipated rewards and costs associated with a behavior. Money, approval, excitement, perceived protection, belonging, and social status may serve as rewards in some settings. Disapproval, damaged relationships, lost opportunities, or formal punishment may serve as costs. Immediate social feedback can matter even when a distant legal consequence is substantial.
Reinforcement may be unplanned. Laughter, admiration, or increased group status following a story about misconduct can reward the performance or presentation of that conduct. Supportive reinforcement can also strengthen lawful problem-solving, conflict avoidance, and treatment participation.
The empirical question is whether rewards and costs were measured independently and whether they preceded later behavior. Criminological studies frequently infer reinforcement from perceived consequences, which makes causal testing difficult. The 2010 theory meta-analysis found more modest or less stable associations for reinforcement measures than for association and definitions.
4. Imitation: observing behavioral models
Imitation is learning a possible action from watching another person. A model can communicate what a behavior looks like, when it is used, and how others respond. Close friends and respected group members can be influential models, but observing an act is only one step in a longer possible process.
Not everyone exposed to a model adopts the behavior. Existing attitudes, available alternatives, skills, opportunity, personal goals, anticipated consequences, and social disapproval can alter the response. The evidence summarized by Pratt and colleagues found imitation to be a weaker measured correlate than some other learning constructs. Therefore, an allegation that someone copied a television scene, video, or notorious offender needs case-specific evidence; resemblance alone is an uncertain basis for explanation.
How these processes work together: a hypothetical example
Imagine a teenager who begins spending time with peers who regularly shoplift. Group members discuss theft as acceptable under certain circumstances, describe how they act, and celebrate when someone succeeds. The teenager witnesses the behavior and later participates. The relationships represent differential association, the favorable interpretations are definitions, the observed acts provide models, and positive group reactions may provide reinforcement.
This is a hypothetical illustration of four constructs, not an established causal sequence. The teenager might have shoplifted before meeting the group and sought out similar peers. Another teenager in the same setting might refuse because of personal values, relationships with supportive adults, anticipated consequences, or a lack of opportunity. Identical exposure does not entail identical outcomes.
The pathways can be reciprocal. Behavior may change friendships, friendships may change beliefs, and beliefs may affect which opportunities appear attractive. Research needs repeated observations, evidence of actual conduct, and credible comparison conditions to untangle those sequences.
What does the strongest evidence show?
Meta-analysis: some mechanisms are better supported than others
Pratt, Cullen, Sellers, and colleagues published The Empirical Status of Social Learning Theory: A Meta-Analysis in Justice Quarterly in 2010. Their synthesis identified relatively strong associations between crime or deviance and measures of differential association and definitions favorable to offending. Associations with differential reinforcement and imitation were weaker overall. Effect magnitude and stability varied by research design and model specification.
This supports a significant relationship between social learning constructs and behavior without proving that all four mechanisms are equally well tested or equally important. A theoretical explanation and an observed association should remain distinct in both scientific and forensic writing.
The included literature also varies in outcomes and populations. Findings about self-reported minor delinquency cannot simply be reinterpreted as evidence about homicide, adult fraud, violent recidivism, or legal guilt. The units of analysis and operational definitions matter.
Longitudinal network studies: both peer influence and selection
The Gallupe, McLevey, and Brown meta-analysis synthesized social network studies capable of estimating both influence among friends and selection into friendships. Both processes received statistical support. People sometimes change after interacting with peers, and they also choose relationships partly because of existing behavioral similarity.
These results improve on cross-sectional findings in which people with similar behavior are measured at one point in time. However, network models still require assumptions about unmeasured influences, how ties are defined, timing between observation waves, and which people are included in the network.
The McGloin and Thomas review explains additional measurement problems. Asking a respondent whether friends offend is different from surveying those friends independently. People may misperceive, overestimate, underestimate, or project behavior onto others. Studies that directly measure nominated peers often find smaller associations than studies using perceived peer conduct, although perceived norms may themselves influence behavior.
Youth crime: learning operates among multiple factors
In 2024, Ayano and colleagues' umbrella review examined systematic reviews and meta-analyses of risk and protective factors for youth crime. The evidence included links involving peer pressure, school environments, family supervision, adverse experiences, previous offending, and other factors. The review categorized peer pressure as having suggestive evidence of association rather than a universally established causal effect.
The broad implication is that social context matters along with family conditions, development, resources, and opportunities. The review did not establish that treating any single correlate would necessarily prevent crime. Its findings also cannot be used to diagnose a young person or calculate a valid individual risk probability.
Peer influence, peer selection, and shared circumstances
Peer influence means that the behavior or expectations of peers contribute to change in an individual's conduct. Peer selection means people form or retain friendships partly because of prior similarities. Shared circumstances mean peers may engage in similar conduct because they attend the same school, experience comparable opportunities, face similar constraints, or participate in the same setting.
All three can produce a correlation between a person's behavior and their friends' behavior. Research designs that ignore selection or shared circumstances can exaggerate an influence interpretation. Even longitudinal models face challenges if friendship nominations omit important relationships or if a substantial influence occurs between observation waves.
Peer selection is itself relevant to prevention: people may change their connections when they enter new schools, jobs, neighborhoods, institutions, or relationships. Influence also varies according to closeness, behavior type, perceived norms, and social position. The question is about mechanisms in a changing network, not permanent labels such as 'good' and 'bad' peers.
Why social learning is important in adolescence
Adolescence often involves growing autonomy, intense peer contact, changing supervision, and new behavioral opportunities. Social rewards connected to belonging and status can become particularly salient. At the same time, family support, adult relationships, school belonging, and prosocial opportunities may provide competing models and reinforcement.
The 2024 umbrella review of youth crime illustrates the importance of considering protective as well as risk factors. Peer pressure is one measured correlate among several, and its effects may differ across youth, contexts, and countries.
Developmental evidence should not be generalized to all adults or to all forms of crime. A school-based network study of minor delinquency does not supply a calibrated estimate of future serious violence, nor does a group-level pattern identify which young person will offend.
Families and early social learning
Families provide early experiences of rules, approval, disagreement, conflict resolution, and behavioral consequences. A child can observe caregivers handle frustration constructively or harshly, and can learn how people respond to requests, boundaries, and mistakes. Consistent expectations and supportive relationships may reinforce effective coping and respectful behavior.
An adverse family experience does not determine subsequent offending. Child development involves many relationships and contexts, and resilience is common. Research on youth crime identifies associations involving supervision and adversity, yet those findings do not authorize blame or a claim that an individual offense was caused by parenting.
The forensic question is narrower: what specific, documented family learning experiences are relevant to the behavior under examination, and what alternative explanations or protective experiences exist? Clinical and family assessment should remain separate from a legal finding.
Co-offending and group behavior
When an offense is committed with others, group interaction may shape decisions, norms, coordination, and perceived rewards. Members might provide knowledge, access, approval, status, or material opportunities. They may also exert pressure, create dependencies, or discourage withdrawal from planned conduct.
Co-offending, however, can reflect strategic cooperation, previous individual motives, opportunity, coercion, and shared resources. These possibilities cannot be reduced to imitation. Membership in a group cannot substitute for evidence of an individual's actions or intentions.
Social learning theory offers questions to investigate: Were norms stated? Which acts were observed? What consequences followed? Did behavior change after the relevant association formed? The answers depend on credible records and observations, not a generalized offender profile.
How social learning can operate in digital environments
Online spaces can convey approval, disapproval, models of conduct, narratives that justify actions, and rapid social feedback. Digital communication changes visibility, audience size, timing, persistence, and the ability to coordinate. Social learning provides a framework for hypotheses about fraud networks, online harassment, or other cyber-enabled acts.
The research examined in this article, however, is strongest in traditional delinquency and social-network contexts. Applying it to a specific online offense requires evidence from that offense and population. Content exposure, platform membership, technological skill, and actual commission of a criminal offense are separate facts.
An online post that appears approving may reflect performance, irony, imitation in speech, or a sincere view; it is not automatically proof of criminal intent. Digital evidence must be interpreted with its provenance and applicable legal rules.
How does social learning relate to other theories of offending?
Self-control approaches focus on how individuals regulate impulses, delay rewards, or respond to immediate incentives. Opportunity perspectives explain how settings make certain acts feasible. Strain accounts consider pressures and responses to adversity. Developmental and life-course approaches trace patterns of onset, persistence, and desistance across time and changing institutions.
These explanations may be complementary. A person can enter a peer network because of a pre-existing preference, encounter opportunities through that network, and subsequently adopt attitudes that make participation more acceptable. Demonstrating which processes actually occurred requires evidence designed to test competing explanations.
Moral disengagement concerns thought processes that minimize self-sanction for harmful acts. It overlaps conceptually with favorable definitions but has its own literature and measures. Likewise, general social influence, aggression, impulsivity, and psychometric assessment have broader research domains. The forensic ownership of this article is the contribution of specific social learning mechanisms to unlawful behavior.
Does social learning theory explain violent offending?
Learning mechanisms can be relevant when violence is observed, normalized, praised, or rewarded in a social context. But violence includes many different acts and situations; the importance of any mechanism varies with developmental history, relationships, threat, opportunity, and immediate circumstances.
Evidence from general delinquency cannot be treated as a specific explanation for murder, intimate partner violence, or targeted violence. Social learning constructs are not standalone violence risk instruments. A clinician should not infer violent conduct from diagnosis, social group, or exposure to a behavioral model.
An individualized violence risk assessment, where legally and professionally appropriate, requires its own validated approaches, multiple information sources, and transparent uncertainty. Social learning may contribute a hypothesis; it is not a finding that a person will commit a future violent act.
Persistent offending, desistance, and change
Persistent offending sometimes occurs in settings where unlawful behavior remains possible, socially rewarded, or embedded in long-standing routines. A person may receive status or resources from activities that make transition difficult. Yet persistence can arise through pathways that are not principally explained by social learning.
Desistance may involve new roles, supportive relationships, meaningful work, altered opportunities, and a change in the personal meaning of conduct. Someone can learn and practice alternative responses while retaining contact with some prior peers. Social networks are dynamic, and rehabilitation research must observe actual change rather than assume that one relationship explains it.
These pathways suggest intervention targets without implying that every social bond must be severed or that all people require the same level of supervision.
Prevention: learning lawful alternatives
Prevention can cultivate relationships and environments where cooperation, problem-solving, emotional regulation, responsibility, and lawful achievement are modeled and reinforced. Schools and communities may support these alternatives through meaningful activities, stable opportunities, consistent expectations, and responsive adults.
The effectiveness of a program cannot be inferred solely from its resemblance to social learning theory. Interventions need appropriate outcomes, comparisons, follow-up, and monitoring for harms. Associations between a protective setting and lower offending may partly reflect other advantages, so causal claims require stronger designs.
A fair prevention approach addresses behaviors and resources rather than stereotyping neighborhoods, friendships, cultural groups, or young people who have faced adversity.
Peer contagion: can interventions make behavior worse?
Some interventions unintentionally place young people together in ways that reward stories of misconduct and normalize disruptive behavior. Dishion, McCord, and Poulin (1999) described evidence for deviancy training and possible harmful effects of certain peer group programs.
A later review by Dishion and Tipsord examined peer contagion across childhood and adolescence, highlighting how social reinforcement and group structure may influence problem behaviors. It also discussed protective features of adult supervision and social context.
These findings do not justify a blanket conclusion that every group program is harmful. Supervision, facilitator skill, group composition, treatment content, and implementation change outcomes. The practical question is whether the exact intervention reliably improves measurable outcomes in the population receiving it.
Correctional rehabilitation and cognitive-social learning
Correctional programs sometimes combine practice of alternative skills, behavioral feedback, prosocial modeling, cognitive restructuring, and supportive social contacts. The U.S. Courts evidence-based practices overview describes supervision methods that include practical assistance and responses to behavior. The Correctional Service of Canada identifies social learning and cognitive-behavioral methods among its program foundations.
The risk–need–responsivity framework often guides intervention intensity, target selection, and delivery. It is important, however, to avoid overstating the supporting evidence. Fazel and colleagues' 2024 umbrella review found mixed support and methodological weaknesses across meta-analyses of RNR principles, including concerns about causal inference, heterogeneity, and transparency. A meta-analysis of randomized community correction studies also emphasized limitations and variation.
The sound conclusion is that socially informed rehabilitation methods are plausible and may be beneficial within properly evaluated programs. They require independent outcome evidence, careful implementation, adjustment for individual needs, and continuous monitoring. Desistance and rehabilitation are not synonymous with high or low scores on a theoretical variable.
Forensic use: what can be assessed?
In a properly defined forensic referral, a psychologist may evaluate offense-related beliefs, significant relationships, behavioral histories, contextual reinforcement, protective supports, and available alternatives. Evidence may come from interviews, case records, collateral reports, repeated observation, and measures validated for the population and purpose.
A case formulation should specify which facts are directly documented, which claims are reported by others, and which conclusions remain interpretations. It should consider alternative explanations and avoid presuming a causal sequence because it appears consistent with social learning theory.
Keep psychological constructs and legal findings separate
A learned definition is a research construct. A quoted statement or witnessed act is a behavioral observation. A history of offending is a documented or alleged historical variable, depending on the source. A score is a psychometric result. A risk factor expresses an association with an outcome in a defined population. A forensic opinion is an evidence-based professional interpretation. A legal finding is a conclusion reached by a legally authorized decision-maker.
These categories have distinct evidentiary demands. A favorable attitude cannot establish guilt, a friendship cannot establish participation, and a psychological explanation cannot determine legal intent or capacity. Criminal responsibility, competence, and admissibility depend on jurisdiction-specific law.
Prediction, calibration, and error
Social learning theory does not yield a validated personal probability of future crime. Prediction requires an instrument and model tested on the relevant population, outcome, and time horizon. Proper evaluation includes discrimination, calibration, outcome base rates, false positives, false negatives, and the effects of thresholds.
When an outcome is infrequent, even a predictor with some statistical value may label many people who will never display that outcome. Performance can also differ across demographic groups and justice settings. A measured association is not an individualized forecast; a risk score is not evidence of guilt and cannot replace a legal decision.
Cultural fairness and institutional context
Friendship patterns, family arrangements, social norms, and definitions of offending are shaped by culture and setting. Administrative indicators such as arrests also reflect enforcement, reporting, and institutional choices. Treating those indicators as neutral measures of behavior may distort assessment and compound disparities.
Fair use requires checking measurement equivalence, relevant comparison groups, language, context, alternative accounts, and possible disparities in surveillance or opportunity. A protected characteristic, neighborhood, economic circumstance, or association with a stigmatized group is no substitute for individual evidence.
What does social learning tell us about legal responsibility?
Social learning can explain how a person may have encountered a model, learned a belief, or experienced reinforcement. It does not determine whether the person committed an offense under the law, possessed the legally required mental state, acted under coercion, met a competency standard, or qualified for a criminal defense.
Legal systems establish their own standards and evidentiary rules, which vary across jurisdictions. Neither social learning theory nor a clinical diagnosis functions as a universal legal test. Expert opinions should be responsive to the narrow referral question and transparent about the difference between research and law.
Limitations of the social learning evidence
Many studies are observational and may be affected by selection, omitted variables, reciprocal influence, and temporal ambiguity. Even excellent longitudinal research is constrained by its measurement schedule and the network boundaries researchers can observe.
Measures of peer behavior range from perceptions to independently reported actions; favorable beliefs can be captured by different survey scales; offending can mean self-reported behavior, arrests, charges, or convictions. These differences affect results. Meta-analyses identify variation across study designs, rather than one universal effect.
A large share of influential research concerns adolescents and accessible school-based networks. Results may not transfer to adults, uncommon offenses, correctional populations, or countries with different legal and social systems. Cultural fairness and unequal exposure to legal enforcement warrant separate attention.
The theory must also account for resistance and change. Many people exposed to crime do not offend; many people who once offended cease. Explaining nonparticipation and desistance is essential to a complete account of social learning.
Common myths about learning and crime
Having delinquent friends means someone will offend
Research documents both social influence and self-selection into friendships. An association is not certainty, and membership in a network is not proof of individual misconduct.
Violent media exposure identifies a future violent offender
Observation alone cannot determine whether someone will imitate a model, commit an offense, or meet any legal standard. Offense-specific evidence and competing circumstances are essential.
Social learning is just copying others
Akers' account also includes differential association, learned definitions, and reinforcement. In the strongest meta-analytic tests, direct imitation generally had more modest associations than relationships and definitions.
Reward always causes later offending
A reward may help maintain behavior under some conditions, but observational correlations cannot establish a causal effect without identifying the timing and alternatives.
Social learning theory proves criminal intent
Psychological processes and criminal intent are different inquiries. A court applies relevant legal standards to case-specific evidence.
Group-based rehabilitation always spreads misconduct
Certain groups can amplify problematic behavior; other supervised interventions show benefits. Claims about harm or benefit must follow the outcomes of the actual program.
Frequently asked questions
What is a simple example of social learning and crime?
A young person hears peers justify shoplifting, sees a peer shoplift, then receives approval for participating. The illustration combines association, definitions, imitation, and reinforcement. It does not establish that all exposed people will act alike.
Who developed social learning theory in criminology?
Ronald L. Akers developed the influential framework by extending Sutherland's differential association approach and integrating behavioral learning mechanisms.
What are the four components of Akers' theory?
Differential association, definitions, differential reinforcement, and imitation.
What is the difference between differential association and social learning?
Sutherland emphasized learning definitions through social interaction. Akers expanded this account by adding explicit reinforcement and imitation processes.
Do bad influences cause juvenile delinquency?
Peer influence is one possible contribution; peer selection and shared circumstances also explain similarity. Researchers must distinguish these explanations rather than assume causation.
Is social learning theory only about adolescents?
The mechanisms can apply in adult settings, but much peer-network research uses youth populations. Generalization to adult or offense-specific samples requires dedicated research.
Does the theory explain gangs and co-offending?
It can frame hypotheses about group learning, approval, and norms, while explanations must also address opportunity, coordination, coercion, prior behavior, and legal evidence.
Does social learning explain cybercrime?
It supplies possible learning mechanisms for digital communities, but cybercrime-specific evidence is needed to demonstrate effects. Platform participation is not proof of an offense.
Can social learning help prevent repeat offending?
It motivates skill training, prosocial models, and behavioral supports. Whether any program prevents recidivism depends on well-designed outcome studies and follow-up.
Does a favorable attitude toward crime prove intention?
No. Survey attitudes, statements, acts, plans, and legal intent are distinct categories.
Can a psychologist predict a crime using social learning theory?
The theory does not itself provide an individually calibrated prediction. Risk estimates require independently validated instruments with documented limitations and errors.
Is social learning theory a mental disorder?
No. It is a scientific explanatory framework, not a clinical diagnosis.
Conclusion
Social learning theory identifies mechanisms by which associations, meanings, modeled behavior, and consequences can shape unlawful conduct and lawful alternatives. Its empirical foundation is strongest for the relationship of offending with peer associations and favorable definitions, alongside longitudinal evidence for both selection and influence. The direct measurement of imitation and reinforcement is more variable.
Forensic application becomes useful when these mechanisms are evaluated in context and kept distinct from individual prediction, legal responsibility, and proof of an offense. Prevention and rehabilitation can draw on learning principles while testing whether specific interventions improve outcomes fairly and reliably.
Related English Hub articles
Related Articles
References
Akers, R. L., & Jennings, W. G. (2009). The social learning theory of crime and deviance. In Handbook on Crime and Deviance. Springer. DOI.
Ayano, G., et al. (2024). Risk and protective factors of youth crime: An umbrella review of systematic reviews and meta-analyses. Clinical Psychology Review, 113, 102479. DOI.
Correctional Service of Canada. (2024). Correctional programs. Government source.
Dishion, T. J., & Tipsord, J. M. (2011). Peer contagion in child and adolescent social and emotional development. Annual Review of Psychology, 62, 189–214. DOI.
Dishion, T. J., McCord, J., & Poulin, F. (1999). When interventions harm: Peer groups and problem behavior. American Psychologist, 54(9), 755–764. PubMed.
Fazel, S., Hurton, C., Burghart, M., DeLisi, M., & Yu, R. (2024). An updated evidence synthesis on the risk–need–responsivity model: Umbrella review and commentary. Journal of Criminal Justice, 92, 102197. DOI.
Gallupe, O., McLevey, J., & Brown, S. (2019). Selection and influence: A meta-analysis of the association between peer and personal offending. Journal of Quantitative Criminology, 35, 313–335. DOI.
McGloin, J. M., & Thomas, K. J. (2019). Peer influence and delinquency. Annual Review of Criminology, 2, 241–264. DOI.
Pratt, T. C., Cullen, F. T., Sellers, C. S., Winfree, L. T., Jr., Madensen, T. D., Daigle, L. E., Fearn, N. E., & Gau, J. M. (2010). The empirical status of social learning theory: A meta-analysis. Justice Quarterly, 27(6), 765–802. DOI.
Salisbury, E. J. (2012). Social learning and crime. In The Oxford Handbook of Criminological Theory (pp. 115–130). Oxford University Press. DOI.
U.S. Courts. (n.d.). Evidence-based practices. U.S. federal courts.
Wang, X., et al. (2023). Are risk–need–responsivity principles golden? A meta-analysis of randomized controlled trials of community correction programs. Journal of Experimental Criminology. PubMed.