Social Learning and Differential Association

Edwin Sutherland presented differential association theory as a set of nine interlocking propositions in the 1947 edition of Principles of Criminology. The propositions are not casual observations; they are meant to constitute a formal theory from which specific hypotheses can be derived and tested. Understanding each proposition is the starting point for understanding every subsequent development in learning theories of crime.

• Criminal behaviour is learned, not inherited or invented independently.
• Criminal behaviour is learned in interaction with other persons, through a process of communication.
• The principal part of the learning of criminal behaviour occurs within intimate personal groups.
• The learning includes the techniques of committing the crime and the specific direction of motives, drives, rationalisations, and attitudes.
• The specific direction of motives and drives is learned from definitions of the legal codes as favourable or unfavourable.
• A person becomes delinquent because of an excess of definitions favourable to violation of law over definitions unfavourable to violation of law.
• Differential associations may vary in frequency, duration, priority, and intensity.
• The process of learning criminal behaviour by association with criminal and anti-criminal patterns involves all the mechanisms that are involved in any other learning.
• Criminal behaviour is an expression of general needs and values, but it is not explained by those general needs and values since non-criminal behaviour is an expression of the same needs and values.

Proposition six is the operational core. The key variable is the ratio of pro-criminal to anti-criminal definitions in a person's social environment. Proposition seven introduces important nuance: not all associations carry equal weight. An association formed early in life (priority) or in an emotionally intense relationship (intensity) carries more weight than a brief, impersonal contact. A teenager whose closest friends are deeply committed to offending is at higher risk than one who encounters criminal attitudes only in passing.

Proposition nine is sometimes overlooked but is theoretically significant. Sutherland rejected simple economic or psychological reductionism. Poverty does not cause crime directly; rather, it channels people into social environments where pro-criminal definitions are more prevalent. Two people with identical financial circumstances can follow very different criminal trajectories depending on whose definitions they are exposed to. This makes differential association a social-structural theory as much as an individual-level one.

In 1966, Ronald Akers and Robert Burgess published a restatement of differential association in the language of behaviourist psychology, specifically B.F. Skinner's operant conditioning. Their 1966 paper, 'Differential Association-Reinforcement Theory of Criminal Behaviour', proposed that Sutherland's propositions could be made more precise by specifying the reinforcement mechanisms through which definitions are acquired and maintained. Akers developed this into a fully articulated social learning theory in his 1973 book Deviant Behavior: A Social Learning Approach and refined it further in subsequent editions.

Akers argued that these four components operate sequentially but also interact in feedback loops. Initial exposure to pro-criminal definitions through differential association primes a person to attempt an act. Imitation may provide the initial template for how to perform it. If the act is then reinforced (by money, peer approval, or the absence of punishment), the behaviour is strengthened. Reinforcement also consolidates the definitions: if crime 'works', the rationalisations for it become more deeply held. This feedback loop explains why, in Akers's model, early intervention is more effective than late intervention.

Social learning theory sits within a crowded field of sociological and psychological crime theories. Understanding where it differs from strain theory, social bond theory, and labelling theory clarifies what is distinctive about the learning approach and where researchers have tried to integrate the perspectives.

The sharpest debate is with Gottfredson and Hirschi's self-control theory. They argue that the apparent effect of delinquent peers on individual offending is spurious: low self-control causes both the choice of delinquent peers and the commission of crime. On this account, social learning theory mistakes correlation for causation. Akers responded that the evidence for peer influence holds even when self-control is controlled for statistically, and that the two frameworks need not be mutually exclusive. Several researchers have developed integrated models that include both early socialisation (producing self-control levels) and ongoing peer learning (producing definitions).

Social learning theory has one of the largest bodies of supporting empirical research in criminology. Studies using self-report survey data consistently find that peer delinquency is one of the strongest predictors of individual delinquency, and that the relationship is mediated by definitions and reinforcement in the direction Akers's model predicts. The Iowa Youth and Families Project, the National Youth Survey in the United States, and the Edinburgh Study of Youth Transitions and Crime in Scotland all produced data consistent with social learning predictions.

The selection problem is the most persistent methodological challenge. People choose their peers partly on the basis of pre-existing attitudes and behaviour. A young person already inclined toward delinquency selects delinquent friends, making it difficult to determine whether peer influence caused the delinquency or whether the delinquency caused the peer selection. Longitudinal studies that track attitudes and peer associations over time have generally found evidence for both processes simultaneously: selection and socialisation both occur, and they amplify each other. The causal arrow points in both directions.

The theory also faces the question of who teaches the first criminal. If crime is always learned from others, there must be an initial source. Sutherland acknowledged this was a problem for individual-level explanation but noted that criminal patterns exist in cultural transmission across generations: techniques for committing fraud, for example, are passed from experienced offenders to new recruits in what is effectively an apprenticeship. Research on criminal networks in the United Kingdom and Netherlands has documented precisely this kind of structured knowledge transfer.

Social learning theory generates concrete prevention implications. If crime is learned through social interaction and maintained by reinforcement, then prevention can work by changing the social environment, altering the reinforcement structure, or replacing pro-criminal definitions with anti-criminal ones. These three levers map directly onto the three most evidence-based intervention types in contemporary practice.

Cognitive behavioural programmes (CBT) target definitions directly. Programmes such as Reasoning and Rehabilitation, developed in Canada and adopted in the United Kingdom, the United States, and Australia, train participants to identify pro-criminal attitudes, recognise how those attitudes support offending, and replace them with alternative reasoning patterns. Meta-analyses consistently find significant reductions in reoffending, with effects concentrated among medium-to-high risk offenders.

Peer-based and mentoring programmes work through the imitation and association mechanisms. By connecting at-risk individuals with pro-social peers or adult mentors, these programmes aim to shift the ratio of pro-criminal to anti-criminal definitions in the person's social environment. Big Brothers Big Sisters programmes in the United States and similar schemes in Australia, Canada, and across Europe have produced modest but consistent effects on delinquency.

In forensic assessment, social learning theory informs risk instruments that include peer association and pro-criminal attitudes as dynamic risk factors. The Level of Service Inventory (LSI-R) and its successors, used widely in North America, the United Kingdom, Australia, and increasingly in India under adapted protocols, score peer associations and antisocial attitudes directly. A high score on these items predicts reoffending and points toward peer-oriented or CBT interventions. Social learning theory thus connects academic criminology to day-to-day probation and parole practice.

Gang intervention programmes such as Cure Violence in the United States deploy credible messengers, people who have previously offended and carry social capital in high-crime communities, to interrupt the imitation and reinforcement mechanisms that spread violence. Evaluations in Chicago, New York, and Baltimore have found significant reductions in shootings in intervention areas. Similar mediator-based approaches have been piloted in the United Kingdom and Brazil.

Sutherland himself developed differential association partly to explain white-collar crime, which was anomalous for biological and economic deprivation theories. His 1949 book White Collar Crime documented systematic law-breaking by corporations and business professionals. He argued that corporate offenders learned pro-violation definitions through their occupational cultures: that bending regulations is normal, that enforcement is unlikely, and that competitors do the same. This made differential association the first major theory to treat corporate crime as requiring the same explanation as street crime.

Contemporary research on financial fraud, tax evasion, and corporate misconduct across the United States, European Union, and the United Kingdom continues to find evidence consistent with social learning mechanisms. Employees who work in organisations where fraud is normalised are more likely to participate in it. Prosecutions under the Sarbanes-Oxley Act in the United States and the Bribery Act 2010 in the United Kingdom both frequently involve co-offending networks where pro-criminal definitions were shared and reinforced across corporate hierarchies.

The theory has genuine limits. It struggles to explain solitary offending, particularly crimes committed by individuals with no clear criminal peer network. It does not easily account for the desistance process: if criminal behaviour is maintained by reinforcement, why do most offenders stop as they age even without formal intervention? Life-course criminologists such as Robert Sampson and John Laub argue that turning points, marriage, stable employment, military service, reorient social ties and reduce exposure to pro-criminal associations. This is consistent with social learning logic but requires the theory to be embedded in a broader life-course account.