Criminal Justice

iResearchNet




Criminal Justice > Criminology Theories > Social Learning Theory

Social Learning Theory




Social learning theory is one of the most influential and empirically supported frameworks in criminology, offering a general explanation for the acquisition, maintenance, and cessation of criminal and deviant behavior through the mechanisms of differential association, definitions, differential reinforcement, and imitation. Developed by Ronald L. Akers as a reformulation and extension of Edwin Sutherland’s differential association theory, social learning theory within Criminology Theories integrates sociological propositions about the social context of crime with principles drawn from behavioral and cognitive psychology. This article traces the theoretical origins of social learning theory in Sutherland’s differential association and in the behavioral psychology of B. F. Skinner and Albert Bandura, examines the four core concepts that constitute the theory’s explanatory apparatus, reviews the empirical evidence that has accumulated over more than five decades of testing, discusses Akers’s macrolevel extension — social structure and social learning — and assesses the theory’s contemporary applications, criticisms, and future directions.

Outline

I. Introduction

II. Theoretical Origins




  • a. Sutherland’s Differential Association Theory
  • b. Burgess and Akers: The Differential Association-Reinforcement Theory
  • c. From Behavioral to Cognitive Learning: Bandura’s Influence

III. The Four Core Concepts

  • a. Differential Association
  • b. Definitions
  • c. Differential Reinforcement
  • d. Imitation

IV. Empirical Status

  • a. The Classic Studies: Boys Town, Iowa, and Elderly Drinking
  • b. The Pratt et al. Meta-Analysis
  • c. Cross-Cultural and Cross-National Tests

V. Social Structure and Social Learning

  • a. The SSSL Model: Four Structural Domains
  • b. The Mediation Hypothesis
  • c. Preliminary Evidence

VI. Contemporary Applications

  • a. Cybercrime and Online Deviance
  • b. Terrorism and Radicalization
  • c. Domestic Violence and Intimate Partner Violence

VII. Criticisms and Limitations

  • a. The Peer Selection Problem
  • b. Temporal Ordering and Causal Ambiguity
  • c. The Relative Weakness of Reinforcement and Imitation

VIII. Comparison with Competing Theories

  • a. Social Learning and Social Control Theory
  • b. Social Learning and Strain Theory
  • c. Social Learning and Self-Control Theory

IX. Policy Implications

  • a. Cognitive-Behavioral Interventions
  • b. Prevention and Early Intervention Programs

X. Conclusion

XI. References

Introduction

Why do individuals engage in criminal and deviant behavior? What social processes lead some persons to adopt values, attitudes, and behavioral patterns that support illegal conduct while others adopt conforming orientations? And what mechanisms sustain criminal involvement over time or lead individuals to desist from offending? Social learning theory addresses these questions by proposing that criminal behavior is learned through the same processes by which all behavior is learned — through social interaction, observation, and the experience of rewards and punishments — and that the social context in which learning occurs determines whether the behavioral patterns that are acquired are criminal or conforming.

Social learning theory, as developed by Ronald L. Akers, represents one of the most extensively tested and broadly supported theoretical frameworks in the history of criminology. It has been applied to explain a wide range of criminal and deviant behaviors — substance use, violent crime, property offenses, sexual coercion, white-collar crime, gang membership, terrorism, and cybercrime — across diverse samples, settings, and cultural contexts. The theory occupies a central position within Criminology Theories alongside social control and strain theories as one of the three core paradigms in contemporary criminological thought (Cullen, Wright, & Blevins, 2006). Its strength lies in its capacity to specify the concrete mechanisms through which social influences translate into individual behavior — mechanisms that are measurable, testable, and amenable to intervention.

The theory’s intellectual roots lie in two traditions: the sociological tradition of Edwin Sutherland’s differential association theory, which located the causes of crime in the social relationships and cultural environments to which individuals are exposed, and the psychological tradition of behavioral and cognitive learning theory, which specified the mechanisms — operant conditioning, reinforcement, and observational learning — through which behavior is acquired and maintained. By integrating these traditions, Akers created a theory that is both sociologically grounded and psychologically precise, capable of explaining not only why individuals begin to engage in crime but also why they continue, escalate, specialize, or desist.

Theoretical Origins

Sutherland’s Differential Association Theory

The sociological foundation of social learning theory is Edwin Sutherland’s differential association theory, first presented in 1939 and refined in the fourth edition of his Principles of Criminology (Sutherland, 1947). Sutherland’s theory, expressed in nine propositions, advanced the fundamental claim that criminal behavior is learned through social interaction in intimate personal groups. The theory’s central proposition — the principle of differential association — stated that an individual becomes delinquent when definitions favorable to violation of law exceed definitions unfavorable to violation of law. The learning process involves not only the techniques of committing crime but also the motives, rationalizations, and attitudes that support criminal conduct. The associations that exert the greatest influence are those that occur earliest in life (priority), persist longest (duration), occur most frequently (frequency), and involve the most intimate and important partners (intensity).

Sutherland’s theory represented a radical departure from the biological and psychological explanations of crime that dominated criminology in the early twentieth century. By locating the causes of crime in social learning processes rather than in individual pathology, Sutherland established the sociological study of criminal behavior on a firm theoretical foundation. His theory also explained a puzzle that individual-level theories could not: why crime rates vary systematically across social groups, neighborhoods, and cultures. If crime is learned through differential association, then groups and communities that expose their members to an excess of pro-criminal definitions will produce higher rates of criminal behavior than groups that expose their members to predominantly conforming definitions.

Despite its theoretical elegance, differential association theory faced persistent criticism. Critics argued that the theory was difficult to operationalize — concepts such as “definitions favorable to violation of law” and the “excess” of such definitions were vague and hard to measure. The theory did not specify the learning mechanisms through which definitions are acquired, and it could not explain why individuals exposed to the same associations sometimes responded differently. These limitations set the stage for the reformulation that Burgess and Akers would undertake in the 1960s.

Burgess and Akers: The Differential Association-Reinforcement Theory

In 1966, Robert L. Burgess and Ronald L. Akers published their differential association-reinforcement theory, which reformulated Sutherland’s nine propositions by integrating them with the principles of operant conditioning developed by B. F. Skinner. Burgess and Akers (1966) argued that Sutherland had correctly identified the social context in which criminal behavior is learned but had failed to specify the learning mechanisms through which this process occurs. By incorporating the concepts of reinforcement, punishment, and discriminative stimuli from behavioral psychology, they provided the mechanistic foundation that Sutherland’s theory lacked.

Burgess and Akers condensed Sutherland’s nine propositions into seven reformulated statements that specified how the learning of criminal behavior operates through operant conditioning. The first principle stated that criminal behavior is learned according to the principles of operant conditioning — replacing Sutherland’s vague reference to “all of the mechanisms that are involved in any other learning.” The sixth principle — corresponding to Sutherland’s principle of differential association — stated that criminal behavior is a function of norms that are discriminative for criminal behavior, the learning of which takes place when such behavior is more highly reinforced than noncriminal behavior. The seventh principle stated that the strength of criminal behavior is a direct function of the amount, frequency, and probability of its reinforcement.

The reformulation addressed the operationalization problem that had plagued differential association theory by specifying the concrete behavioral mechanisms — reinforcement contingencies, discriminative stimuli, schedules of reinforcement — through which definitions are acquired, maintained, and extinguished. It also provided a framework for understanding individual variation in response to similar social environments: individuals exposed to the same associations may respond differently because of differences in their reinforcement histories, the schedules and values of reinforcement they experience, and the discriminative stimuli present in their immediate situations.

From Behavioral to Cognitive Learning: Bandura’s Influence

Akers (1973, 1977, 1985, 1998) continued to develop the theory over subsequent decades, and in later formulations he increasingly acknowledged the influence of Albert Bandura’s social cognitive theory. While Burgess and Akers (1966) had originally subsumed imitation under the broader umbrella of operant conditioning — treating modeling as simply one kind of behavior shaped through successive approximation — Akers came to accept the distinctiveness of observational learning as a separate behavioral mechanism. Bandura (1977) had demonstrated that individuals can acquire new behavioral patterns simply by observing the behavior of models and the consequences that models experience, without direct reinforcement of their own behavior — a process Bandura termed vicarious reinforcement.

This shift from a purely operant framework to one that incorporated cognitive learning processes broadened the theory’s explanatory scope. It allowed social learning theory to account for the acquisition of complex behavioral patterns — including criminal techniques, rationalizations, and subcultural values — that would be difficult to explain through direct operant conditioning alone. It also provided a mechanism for explaining the influence of indirect or symbolic models, including media representations of criminal behavior, on individual conduct. Akers (1998) emphasized that social learning theory is not a rival of Sutherland’s differential association theory but a broader framework that modifies, builds on, and integrates Sutherland’s propositions with principles from behavioral and cognitive learning theory. Support for differential association provides support for social learning theory, and findings that support social learning do not negate Sutherland’s original contribution.

The Four Core Concepts

Differential Association

Differential association is the most central concept in Akers’s social learning theory. It refers to the process by which individuals are exposed to normative definitions, behavioral models, and reinforcement patterns through their interactions with others — both directly, through face-to-face relationships, and indirectly, through reference groups, media, and symbolic communication. The groups with which an individual differentially associates provide the primary social context in which all of the other mechanisms of social learning operate: they expose the individual to definitions favorable or unfavorable to crime, present models of criminal or conforming behavior to imitate, and provide the reinforcement (or punishment) that shapes the individual’s behavioral choices.

Following Sutherland, Akers identified family and peers as the primary groups that are most salient for the social learning process. In early childhood, the family is typically the dominant source of behavioral learning; during adolescence, peer groups, school associations, and leisure groups assume increasing importance; and in adulthood, spouses, work groups, and friendship networks become the primary contexts for learning. Secondary and reference groups — including neighbors, religious leaders, teachers, and what Mark Warr (2002) termed “virtual groups” such as mass media and Internet communities — can also influence learning indirectly through the normative definitions and behavioral models they transmit.

The theory predicts that the associations exerting the greatest effect on behavior are those characterized by Sutherland’s four modalities: priority (associations formed earlier in life exert greater influence), duration (longer-lasting associations are more influential), frequency (more frequent interactions have greater effect), and intensity (associations with closer, more important partners carry more weight). The more an individual is differentially associated with persons who engage in and espouse definitions favorable to criminal behavior, the greater the probability that the individual will engage in criminal behavior.

Definitions

Definitions are the attitudes, orientations, rationalizations, and moral evaluations that an individual holds toward a given behavior. They represent the cognitive dimension of social learning theory — the internal system of beliefs that either restrains or motivates criminal conduct. Akers distinguished between general definitions, which are broad moral and religious beliefs about right and wrong, and specific definitions, which orient the individual toward committing or refraining from particular criminal or deviant acts. An individual may hold a general belief that violence is wrong while simultaneously holding a specific definition that smoking marijuana is harmless and acceptable.

Akers further classified definitions as conventional (negative toward crime), positive (favorable toward crime), or neutralizing (justifications and excuses that make criminal behavior permissible in specific circumstances). Conventional definitions restrain criminal behavior by defining it as morally wrong or legally impermissible. Positive definitions actively motivate criminal behavior by framing it as desirable, admirable, or wholly acceptable — as when a subcultural norm defines physical aggression as a sign of masculine strength. Neutralizing definitions operate by providing rationalizations that remove the moral restraint that would otherwise inhibit criminal conduct — justifications such as “the company can afford it,” “he had it coming,” or “everyone does it.”

Akers’s conceptualization of neutralizing definitions incorporated insights from Gresham Sykes and David Matza’s (1957) techniques of neutralization, Donald Cressey’s (1953) analysis of the verbalizations used by embezzlers to justify their conduct, and Bandura’s (1990) concept of moral disengagement. These neutralizing definitions serve both a cognitive function — providing a readily accessible system of justifications that lowers the threshold for criminal action — and a behavioral function — serving as discriminative stimuli that signal that criminal behavior is appropriate or justified in a particular situation.

Differential Reinforcement

Differential reinforcement refers to the balance of anticipated and experienced rewards and punishments associated with a given behavior. The concept draws directly from Skinner’s operant conditioning framework: behaviors that are reinforced (rewarded) are more likely to be repeated, while behaviors that are punished are less likely to be repeated. The differential reinforcement process operates through four modes. Positive reinforcement occurs when a behavior produces a rewarding consequence — peer approval, financial gain, pleasurable sensations, or elevated status. Negative reinforcement occurs when a behavior allows the individual to escape or avoid an aversive stimulus — as when drug use alleviates anxiety or when truancy allows a student to avoid a hostile school environment.

Positive punishment occurs when a behavior produces an aversive consequence — arrest, injury, social disapproval, or loss of reputation. Negative punishment occurs when a behavior results in the removal of a valued stimulus — as when a parent revokes driving privileges in response to rule-breaking. The theory predicts that individuals whose criminal behavior has been positively reinforced and whose conforming behavior has been inadequately reinforced will be more likely to continue offending, while individuals whose criminal behavior has been punished and whose conforming behavior has been rewarded will be more likely to desist.

Akers emphasized that the reinforcement process does not operate in an either-or fashion but follows a quantitative law of effect: the behavior that yields the greatest net balance of reinforcement relative to alternative behaviors is the behavior most likely to be performed and repeated. Rewards that are higher in value, more certain, and more immediate exert stronger effects than rewards that are modest, uncertain, or delayed. This principle helps explain why criminal behavior persists despite the threat of formal punishment: the rewards of crime (immediate gratification, peer approval, material gain) are often more salient and certain than the punishments (which are delayed, uncertain, and intermittent).

Imitation

Imitation — also termed modeling or observational learning — is the process by which an individual engages in behavior that replicates or follows behavior observed in others. An individual can observe potential models either directly, through personal interaction, or indirectly, through media representations, online content, or symbolic communication. The probability that observed behavior will be imitated depends on the characteristics of the model (whether the model is liked, respected, or similar to the observer), the characteristics of the behavior itself (its complexity, salience, and perceived utility), and the observed consequences of the behavior (whether the model is seen to be rewarded or punished).

Akers argued that imitation is most important in the initial acquisition of novel behavior — the first time an individual tries a drug, commits a theft, or engages in aggressive conduct after observing others doing so. Once the behavior has been initiated, its continuation or cessation is governed primarily by the reinforcement contingencies the individual experiences directly. This hypothesis explains why the empirical effect of imitation on criminal behavior is typically more modest than the effects of differential association and definitions — imitation triggers the behavior, but reinforcement and association sustain it.

Table 1. The Four Core Concepts of Social Learning Theory

Concept Definition Primary Mechanism Empirical Support
Differential association Exposure to pro-criminal or pro-social norms and models through social interaction Direct and indirect social interaction with primary and reference groups Strong and consistent across studies
Definitions Personal attitudes, rationalizations, and moral evaluations toward a behavior Cognitive orientation shaped by socialization and association Strong; second strongest predictor
Differential reinforcement The balance of anticipated and experienced rewards and punishments for a behavior Operant conditioning: positive/negative reinforcement and punishment Moderate; measurement challenges
Imitation Replication of behavior observed in models Observational/vicarious learning Modest; strongest for behavioral initiation

Empirical Status

The Classic Studies: Boys Town, Iowa, and Elderly Drinking

Social learning theory has been subjected to extensive empirical testing across more than five decades. Akers and his associates conducted four landmark research projects that established the theory’s empirical credentials. The Boys Town study (Akers, Krohn, Lanza-Kaduce, & Radosevich, 1979) — the first full test of all four social learning components — surveyed approximately 3,000 students in grades 7 through 12 across eight Midwestern communities. The social learning variables explained 54 percent of the variance in the frequency of drinking alcohol and 68 percent of the variance in marijuana use — effect sizes that far exceeded those typically reported for competing criminological theories. Each of the four elements exerted a substantial independent effect, with the exception of imitation, whose more modest contribution was consistent with its hypothesized role in behavioral initiation rather than maintenance.

The Iowa study, a five-year longitudinal examination of adolescent smoking in Muscatine, Iowa, replicated the Boys Town findings with nearly identical results (Spear & Akers, 1988). The social learning variables explained over 50 percent of the variance in self-reported smoking in the cross-sectional analysis. Longitudinal analysis using data from all five waves demonstrated that the social learning variables predicted both the continuation and cessation of smoking, with approximately 41 percent explained variance in predicting smoking status by the third year among those who had been smokers in prior waves (Krohn, Skinner, Massey, & Akers, 1985). Akers and Lee (1996) provided further longitudinal support using the complete five-year dataset, revealing reciprocal effects between smoking behavior and the social learning variables.

The third project examined the frequency of alcohol use and problem drinking among a large sample of elderly respondents in Florida and New Jersey (Akers, 1992). The social learning variables accounted for more than 50 percent of the explained variance in elderly alcohol use and abuse — demonstrating that social learning processes operate across the lifespan, not only during adolescence. The fourth project examined rape and sexual coercion among college men (Boeringer, Shehan, & Akers, 1991), finding that the social learning variables exerted moderate to strong effects on self-reported use of nonphysical coercion and on rape proclivity. Lanza-Kaduce, Akers, Krohn, and Radosevich (1984) also demonstrated that social learning variables predicted the cessation of substance use — showing that desistance, like initiation, is governed by changes in associations, definitions, and reinforcement contingencies.

The Pratt et al. Meta-Analysis

The most systematic assessment of social learning theory’s empirical status was the meta-analysis conducted by Pratt, Cullen, Sellers, Winfree, Madensen, Daigle, Fearn, and Gau (2010), published in Justice Quarterly. Pratt and colleagues systematically identified all studies measuring social learning variables published in leading criminology and criminal justice journals and calculated effect size estimates across studies. Their results revealed that differential association and definitions (antisocial attitudes) produced strong and consistent effect sizes across studies, confirming these as the most empirically validated components of the theory. Differential reinforcement and imitation produced more modest effect sizes, with considerable variation across studies depending on how these concepts were operationalized. Effect sizes for all four variables differed according to variations in model specification, dependent variable type, and sample characteristics — suggesting that the strength of social learning effects is partially contingent on methodological choices.

The Pratt et al. (2010) meta-analysis confirmed what the individual studies had consistently shown: social learning theory is among the most empirically supported theories in criminology, with peer association (differential association) and antisocial attitudes (definitions) ranking among the strongest and most consistent predictors of criminal and deviant behavior across the criminological literature.

Cross-Cultural and Cross-National Tests

A growing body of research has tested social learning theory outside the American context. Hwang and Akers (2003) tested the theory using data on substance use among Korean adolescents, finding results consistent with the American studies. Meneses and Akers (2011) compared marijuana use among American and Bolivian university students and found that the social learning variables operated similarly across both cultural contexts. Miller, Jennings, Alvarez-Rivera, and Miller (2008) tested the theory among Puerto Rican adolescents, and Jensen and Akers (2003) applied social learning principles to explain cross-national variation in homicide rates. These cross-cultural replications support the theory’s claim to generality — the proposition that the same learning processes govern criminal behavior regardless of cultural context — although they also reveal variations in the relative strength of specific variables across different societies.

Table 2. Key Empirical Tests of Social Learning Theory

Study Sample/Setting Dependent Variable Key Findings
Akers et al. (1979) — Boys Town 3,000 adolescents, 8 Midwestern communities Alcohol and marijuana use R² = .54 (alcohol), .68 (marijuana); all four SL variables significant except imitation
Spear & Akers (1988) — Iowa Adolescents, Muscatine, Iowa Smoking Cross-sectional replication of Boys Town; R² > .50 for smoking
Krohn et al. (1985) — Iowa longitudinal 5-year panel, adolescents Smoking initiation and cessation 41% explained variance in predicting continuation/cessation by Year 3
Akers (1992) — Elderly drinking Elderly adults, FL and NJ Alcohol use and abuse R² > .50; social learning operates across the lifespan
Boeringer et al. (1991) College men Rape and sexual coercion Moderate to strong effects of SL variables on coercion and rape proclivity
Pratt et al. (2010) — Meta-analysis 133 effect size estimates across studies Multiple crime/deviance types Strong effects for differential association and definitions; modest for reinforcement and imitation
Hwang & Akers (2003) Korean adolescents Substance use Cross-cultural support; results consistent with American studies
Miller et al. (2008) Puerto Rican adolescents Substance use Partial test supports SL theory across cultural contexts

Social Structure and Social Learning

The SSSL Model: Four Structural Domains

Akers (1998) extended social learning theory to the macrolevel with his social structure and social learning (SSSL) model, which proposes that social structural factors affect individual behavior indirectly through their influence on the social learning variables. The SSSL model identifies four domains of social structure within which the social learning process operates. Differential social organization refers to structural characteristics of communities and societies — age composition, population density, urbanization — that produce different aggregate rates of crime. Differential location in the social structure refers to the sociodemographic characteristics — class, gender, race, age, marital status — that position individuals within the larger social structure and thereby shape their exposure to pro-criminal and pro-social associations. Theoretically defined structural variables refer to concepts drawn from other criminological theories — anomie, social disorganization, class oppression, patriarchy — that identify criminogenic conditions in societies and communities. Differential social location refers to individuals’ membership in primary, secondary, and reference groups — family, peer groups, work groups, religious organizations — that provide the immediate social context for learning.

The Mediation Hypothesis

The central proposition of the SSSL model is the mediation hypothesis: the effects of social structural factors on individual criminal behavior are mediated by the social learning variables. Social structure does not directly cause crime; rather, structural conditions shape the associations individuals form, the definitions they are exposed to, the reinforcement contingencies they experience, and the models they observe — and these social learning processes, in turn, directly affect the individual’s probability of engaging in criminal behavior. If the mediation hypothesis is correct, then the social structural variables should exhibit direct effects on both the social learning variables and the dependent variable, but once the social learning variables are included in the model, the direct effects of the structural variables should be substantially reduced or eliminated.

This formulation positions social learning theory as an integrative framework capable of absorbing insights from macro-level theories — strain theory, social disorganization theory, conflict theory — by specifying the individual-level mechanisms through which structural conditions translate into criminal behavior. It offers an answer to a question that macro-level theories have traditionally struggled with: why do some individuals exposed to criminogenic structural conditions commit crimes while others in similar conditions do not? The answer, according to SSSL, lies in the variation in social learning processes that individuals experience within structurally similar environments.

Preliminary Evidence

The empirical research on SSSL remains relatively limited compared to the extensive literature testing the micro-level theory, but the preliminary findings are encouraging. Studies have demonstrated support for the mediation hypothesis across a range of dependent variables, including adolescent delinquency and substance use, elderly alcohol abuse, sexual coercion, violence, and binge drinking among college students (Akers & Jensen, 2006). Jensen and Akers (2003) applied the SSSL model to explain cross-national variation in homicide rates, finding that structural factors operated through social learning mechanisms to produce differential rates of lethal violence. However, the mediation is not always complete — some studies find that structural variables retain modest direct effects on crime even after the social learning variables are included, suggesting that additional mediating processes may be at work beyond those specified by social learning theory.

Contemporary Applications

Cybercrime and Online Deviance

One of the most productive areas of contemporary application has been the extension of social learning theory to cybercrime and online deviance. Holt, Burruss, and Bossler (2010) used structural equation modeling to assess the full social learning process in relation to cyber-deviance, finding that social learning functioned as a second-order latent construct that explained a large proportion of the variation in online offending. Their results also demonstrated that the social learning process mediated the effects of race and sex on cyber-deviance, supporting the SSSL model in the digital context. Subsequent research has examined the relative influence of offline versus online peer associations on cybercrime, finding that offline deviant peers typically exert stronger effects than virtual peers (Dearden & Parti, 2021), though online associations — including those formed in hacker forums, extremist communities, and social media platforms — also exert significant independent effects. The application of social learning theory to cybercrime has extended the theory’s domain beyond traditional street crime and demonstrated its relevance to the rapidly evolving landscape of digital offending.

Terrorism and Radicalization

Akers and Silverman (2004) proposed a social learning model of terrorism, arguing that the same mechanisms — differential association with radical groups, acquisition of definitions favorable to political violence, reinforcement through group approval and ideological commitment, and imitation of martyrdom models — that explain ordinary criminal behavior also explain the recruitment and radicalization of terrorists. Mills, Freilich, Chermak, Holt, and LaFree (2019) tested this framework using qualitative case studies of hate crime and terrorist offenders, finding support for an integrated social control–social learning model in explaining radicalization pathways. Wolfowicz, Litmanovitz, Weisburd, and Hasisi (2021) compared the Facebook profiles of Palestinian terrorists with those of nonviolent radicals in the 100 days before attacks, finding that variables from each of the four social learning domains — differential association, definitions, differential reinforcement, and imitation — significantly differentiated terrorists from nonviolent radicals. These findings demonstrate that social learning processes operate in online environments and that behavioral-level metrics derived from the theory can contribute to risk assessment and threat identification.

Domestic Violence and Intimate Partner Violence

Social learning theory has been applied extensively to explain domestic violence and intimate partner violence. The theory predicts that individuals who observe violence between parents during childhood — a form of differential association and imitation — are more likely to adopt definitions favorable to the use of violence in intimate relationships and to employ violence themselves in adulthood. Research has consistently supported this intergenerational transmission hypothesis, finding that exposure to parental violence is among the strongest predictors of subsequent perpetration and victimization. Social learning theory provides a more complete account of this process than simple intergenerational cycling models because it specifies the mediating mechanisms — the acquisition of pro-violent definitions, the reinforcement of aggressive behavior through partner compliance or peer approval, and the modeling of violent conflict resolution strategies — through which childhood exposure translates into adult perpetration.

Criticisms and Limitations

The Peer Selection Problem

The most persistent criticism of social learning theory concerns the temporal ordering of the relationship between peer association and criminal behavior. Critics argue that the strong correlation between having delinquent peers and engaging in delinquency may reflect a selection effect rather than a socialization effect — individuals who are already predisposed to crime may seek out like-minded peers rather than being socialized into crime by their associations. This “birds of a feather” critique suggests that delinquent peer association is a consequence rather than a cause of criminal behavior. Longitudinal research has produced mixed results on this question, with most studies finding evidence for both selection and socialization effects operating simultaneously. Gallupe, McLevey, and Brown (2018) conducted a meta-analysis of 19 longitudinal studies using stochastic actor-oriented models to examine peer effects on crime, finding evidence for both peer influence and peer selection but with socialization effects generally exceeding selection effects. Social learning theorists have responded that the theory does not claim that association is the sole cause of criminal behavior — rather, it is one component of a reciprocal process in which associations influence behavior and behavior, in turn, influences subsequent associations.

Temporal Ordering and Causal Ambiguity

A related criticism concerns the difficulty of establishing the causal direction of the relationships specified by the theory. Cross-sectional studies, which constitute the majority of the empirical literature, cannot determine whether delinquent associations precede delinquent behavior or vice versa. Even longitudinal studies face challenges in disentangling the reciprocal effects of association, definitions, reinforcement, and behavior. Akers (1998) acknowledged the reciprocal nature of these relationships and argued that the theory explicitly predicts feedback loops in which behavior affects subsequent associations and definitions — making reciprocal causation a feature of the theory rather than a disconfirmation of it. Nevertheless, the difficulty of establishing clear temporal ordering remains a methodological limitation that weakens causal claims.

The Relative Weakness of Reinforcement and Imitation

The Pratt et al. (2010) meta-analysis highlighted a significant asymmetry in the empirical support for the four core concepts: differential association and definitions receive strong and consistent support, while differential reinforcement and imitation receive only modest support with considerable cross-study variation. This finding raises the question of whether social learning theory is primarily a theory about peer influence and attitudes — which would make it substantially similar to differential association theory — or whether the reinforcement and imitation components add genuine explanatory power. Akers and Jennings (2016) attributed the weaker findings for reinforcement and imitation to measurement difficulties — these concepts are inherently harder to operationalize in survey research than peer association and attitudes — rather than to theoretical inadequacy. The challenge of adequately measuring the full range of reinforcement contingencies that individuals experience remains a significant methodological obstacle.

Comparison with Competing Theories

Social Learning and Social Control Theory

Social learning theory and social control theory (Hirschi, 1969) offer contrasting accounts of criminal behavior. Social control theory asks not why individuals commit crime but why they conform — proposing that strong bonds to conventional society (attachment, commitment, involvement, belief) restrain individuals from acting on their natural inclinations toward self-interest. Social learning theory, by contrast, proposes that criminal behavior must be learned through exposure to pro-criminal associations and definitions. The theories make opposing predictions about the role of peers: social control theory predicts that weak social bonds lead to crime, with delinquent peer association being a consequence rather than a cause of delinquency, while social learning theory predicts that delinquent peer association is a primary cause of criminal behavior. Empirical tests comparing the two theories have generally found that peer variables (consistent with social learning) are stronger predictors of delinquency than bonding variables (consistent with social control), although both sets of variables typically make independent contributions.

Social Learning and Strain Theory

Strain theory (Merton, 1938; Agnew, 1992) proposes that criminal behavior results from the frustration, anger, or negative affect produced by the inability to achieve valued goals through legitimate means. Social learning theory and strain theory are not necessarily incompatible — the experience of strain may increase exposure to delinquent peers, strengthen definitions favorable to crime, and alter reinforcement contingencies in ways that make criminal behavior more attractive. The SSSL model explicitly identifies strain-related structural variables as potential distal causes of crime that operate through social learning mechanisms. Empirical tests comparing the two theories have found that social learning variables typically account for more variance in crime than strain variables, but that both theories contribute independent explanatory power.

Social Learning and Self-Control Theory

Gottfredson and Hirschi’s (1990) self-control theory proposes that low self-control — established through ineffective parenting during childhood — is the primary cause of criminal behavior across all times, places, and contexts. Self-control theory explicitly challenges social learning theory by arguing that the apparent effect of peer association on crime is spurious — a product of the selection of like-minded, low-self-control individuals into delinquent peer groups. Empirical tests comparing the two theories have generally found that both self-control and social learning variables make significant independent contributions to explaining crime, with peer association effects persisting after controlling for self-control and vice versa (Pratt & Cullen, 2000). This suggests that the two theories capture different aspects of the causal process and that neither alone provides a complete account.

Table 3. Social Learning Theory Compared with Competing Frameworks

Dimension Social Learning Theory Social Control Theory Strain Theory Self-Control Theory
Central question How is criminal behavior learned? Why do people conform? Why do strained individuals turn to crime? Why do some individuals lack restraint?
Key causal variable Differential association, definitions, reinforcement Social bonds (attachment, commitment, belief) Blocked goals, negative affect Low self-control
Role of peers Primary cause (socialization) Consequence of weak bonds (selection) Source of delinquent coping strategies Spurious (selection by self-control level)
Level of analysis Individual (micro); extended to macro via SSSL Individual (micro) Individual and structural Individual (micro)
Empirical support Strong, especially for association and definitions Moderate; attachment and commitment strongest Moderate; general strain theory well supported Strong for self-control; peer effects persist

Policy Implications

Cognitive-Behavioral Interventions

Social learning theory provides the theoretical foundation for cognitive-behavioral treatment and intervention programs in corrections — the approach that the “what works” literature has consistently identified as the most effective strategy for reducing recidivism. Cognitive-behavioral programs target the social learning mechanisms that sustain criminal behavior: they work to change the pro-criminal definitions (antisocial attitudes, rationalizations, cognitive distortions) that motivate offending, to restructure the reinforcement contingencies that reward criminal conduct, and to build prosocial skills and associations that provide alternative sources of reinforcement. The risk-need-responsivity (RNR) model developed by Andrews and Bonta (2010) — the most widely adopted framework for offender assessment and treatment — draws directly on social learning principles in identifying antisocial attitudes, antisocial associates, and antisocial behavioral patterns as the primary criminogenic needs that treatment should target. Meta-analyses of correctional treatment programs have consistently found that programs based on cognitive-behavioral and social learning principles are significantly more effective at reducing recidivism than programs based on other theoretical approaches or no theoretical framework at all (Cullen, Wright, Gendreau, & Andrews, 2003).

Prevention and Early Intervention Programs

At the prevention level, social learning theory supports programs that seek to reduce exposure to pro-criminal associations and definitions during the critical developmental periods of childhood and adolescence. Family-based interventions — such as functional family therapy and multisystemic therapy — target the family environment in which early social learning occurs, working to improve parental supervision, reduce exposure to deviant models within the household, and strengthen prosocial definitions and reinforcement contingencies. School-based programs that teach resistance skills, prosocial decision-making, and critical evaluation of media messages draw on the social learning insight that definitions favorable to crime can be countered by exposure to definitions unfavorable to crime. Community-based programs that provide structured prosocial activities and mentoring relationships for at-risk youth seek to shift the balance of differential association toward conforming rather than deviant associations.

Conclusion

Social learning theory has established itself as one of the foundational paradigms in criminological thought. Across more than five decades of empirical testing — spanning substance use, violence, sexual coercion, white-collar crime, gang membership, terrorism, and cybercrime — the theory has demonstrated a breadth and consistency of empirical support that few criminological theories can match. The Pratt et al. (2010) meta-analysis confirmed that differential association and definitions rank among the strongest and most consistent predictors of criminal behavior in the empirical literature, and the theory’s four-component framework provides a more complete account of the learning process than Sutherland’s original formulation.

The SSSL extension represents an ambitious effort to bridge the gap between individual-level and structural-level explanations of crime by specifying the mechanisms through which social structure affects individual behavior. While the evidence on SSSL remains preliminary, the mediation hypothesis has received consistent support across the studies that have tested it, and the framework provides a principled basis for integrating insights from macro-level theories into a unified explanatory model.

The future of social learning theory lies in continued refinement of measurement approaches — particularly for the differential reinforcement and imitation components that have proven difficult to operationalize — in further testing of the SSSL model across diverse structural contexts, and in the application of social learning principles to emerging forms of crime in digital environments. As Akers (1998) argued, if social learning is truly a general theory, then it should account for criminal behavior not only in American and Western societies but across cultures, historical periods, and technological contexts. The cross-cultural studies conducted to date have largely supported this claim, and the theory’s successful application to cybercrime and online radicalization suggests that the mechanisms of social learning operate in digital environments as they do in physical ones — through associations, definitions, reinforcement, and modeling that shape behavior regardless of the medium through which they are transmitted.

References

  1. Agnew, R. (1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30(1), 47–87. https://doi.org/10.1111/j.1745-9125.1992.tb01093.x
  2. Akers, R. L. (1973). Deviant behavior: A social learning approach. Wadsworth.
  3. Akers, R. L. (1985). Deviant behavior: A social learning approach (3rd ed.). Wadsworth.
  4. Akers, R. L. (1992). Drugs, alcohol, and society: Social structure, process, and policy. Journal of Drug Issues, 22(3), 735–752.
  5. Akers, R. L. (1998). Social learning and social structure: A general theory of crime and deviance. Northeastern University Press.
  6. Akers, R. L., & Jensen, G. F. (2006). The empirical status of social learning theory of crime and deviance: The past, present, and future. In F. T. Cullen, J. P. Wright, & K. R. Blevins (Eds.), Taking stock: The status of criminological theory (pp. 37–76). Transaction.
  7. Akers, R. L., & Jennings, W. G. (2016). Social learning theory. In A. R. Piquero (Ed.), The handbook of criminological theory (pp. 230–240). Wiley Blackwell.
  8. Akers, R. L., Krohn, M. D., Lanza-Kaduce, L., & Radosevich, M. (1979). Social learning and deviant behavior: A specific test of a general theory. American Sociological Review, 44(4), 636–655. https://doi.org/10.2307/2094592
  9. Akers, R. L., & Lee, G. (1996). A longitudinal test of social learning theory: Adolescent smoking. Journal of Drug Issues, 26(2), 317–343. https://doi.org/10.1177/002204269602600203
  10. Akers, R. L., & Sellers, C. S. (2013). Criminological theories: Introduction, evaluation, and application (6th ed.). Oxford University Press.
  11. Akers, R. L., & Silverman, A. (2004). Toward a social learning model of violence and terrorism. In M. A. Zahn, H. H. Brownstein, & S. L. Jackson (Eds.), Violence: From theory to research (pp. 19–35). LexisNexis–Anderson.
  12. Anderson, E. (1999). Code of the street: Decency, violence, and the moral life of the inner city. W. W. Norton.
  13. Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct (5th ed.). Anderson.
  14. Bandura, A. (1977). Social learning theory. Prentice Hall.
  15. Bandura, A. (1990). Mechanisms of moral disengagement. In W. Reich (Ed.), Origins of terrorism: Psychologies, ideologies, theologies, and states of mind (pp. 161–191). Cambridge University Press.
  16. Boeringer, S., Shehan, C. L., & Akers, R. L. (1991). Social contexts and social learning in sexual coercion and aggression: Assessing the contribution of fraternity membership. Family Relations, 40(1), 558–564. https://doi.org/10.2307/585641
  17. Burgess, R. L., & Akers, R. L. (1966). A differential association-reinforcement theory of criminal behavior. Social Problems, 14(2), 128–147. https://doi.org/10.2307/798612
  18. Cressey, D. R. (1953). Other people’s money. Free Press.
  19. Cullen, F. T., Wright, J. P., & Blevins, K. R. (Eds.). (2006). Taking stock: The status of criminological theory. Transaction.
  20. Cullen, F. T., Wright, J. P., Gendreau, P., & Andrews, D. A. (2003). What correctional treatment can tell us about criminological theory: Implications for social learning theory. In R. L. Akers & G. F. Jensen (Eds.), Advances in criminological theory: Vol. 11. Social learning theory and the explanation of crime (pp. 339–362). Transaction.
  21. Dearden, T. E., & Parti, K. (2021). Cybercrime, online peer association, and social learning. American Journal of Criminal Justice, 46(3), 442–460. https://doi.org/10.1007/s12103-020-09582-8
  22. Gallupe, O., McLevey, J., & Brown, S. (2018). Selection and influence: A meta-analysis of the association between peer and personal offending. Journal of Quantitative Criminology, 35(2), 313–335. https://doi.org/10.1007/s10940-018-9384-y
  23. Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford University Press.
  24. Hirschi, T. (1969). Causes of delinquency. University of California Press.
  25. Holt, T. J., Burruss, G. W., & Bossler, A. M. (2010). Social learning and cyber-deviance: Examining the importance of a full social learning model in the virtual world. Journal of Crime and Justice, 33(2), 31–61. https://doi.org/10.1080/0735648X.2010.9721287
  26. Hwang, S., & Akers, R. L. (2003). Substance use by Korean adolescents: A cross-cultural test of social learning, social bonding, and self-control theories. In R. L. Akers & G. F. Jensen (Eds.), Advances in criminological theory: Vol. 11. Social learning theory and the explanation of crime (pp. 39–64). Transaction.
  27. Jensen, G. F., & Akers, R. L. (2003). Taking social learning global: Micro-macro transitions in criminological theory. In R. L. Akers & G. F. Jensen (Eds.), Advances in criminological theory: Vol. 11. Social learning theory and the explanation of crime (pp. 9–37). Transaction.
  28. Krohn, M. D., Skinner, W. F., Massey, J. L., & Akers, R. L. (1985). Social learning theory and adolescent cigarette smoking. Social Problems, 32(5), 455–473. https://doi.org/10.2307/800775
  29. Lanza-Kaduce, L., Akers, R. L., Krohn, M. D., & Radosevich, M. (1984). Cessation of alcohol and drug use among adolescents: A social learning model. Deviant Behavior, 5(1–4), 79–96. https://doi.org/10.1080/01639625.1984.9967596
  30. Meneses, R. A., & Akers, R. L. (2011). A comparison of four general theories of crime and deviance: Marijuana use among American and Bolivian university students. International Criminal Justice Review, 21(3), 263–283. https://doi.org/10.1177/1057567711408302
  31. Merton, R. K. (1938). Social structure and anomie. American Sociological Review, 3(5), 672–682. https://doi.org/10.2307/2084686
  32. Miller, H. V., Jennings, W. G., Alvarez-Rivera, L. L., & Miller, J. M. (2008). Explaining substance use among Puerto Rican adolescents: A partial test of social learning theory. Journal of Drug Issues, 38(1), 261–284. https://doi.org/10.1177/002204260803800112
  33. Mills, C. E., Freilich, J. D., Chermak, S. M., Holt, T. J., & LaFree, G. (2019). Social learning and social control in the off- and online pathways to hate crime and terrorist violence. Studies in Conflict and Terrorism, 44(9), 701–725. https://doi.org/10.1080/1057610X.2019.1585628
  34. Pratt, T. C., & Cullen, F. T. (2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38(3), 931–964. https://doi.org/10.1111/j.1745-9125.2000.tb00911.x
  35. Pratt, T. C., Cullen, F. T., Sellers, C. S., Winfree, L. T., 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. https://doi.org/10.1080/07418820903379610
  36. Spear, S., & Akers, R. L. (1988). Social learning variables and the risk of habitual smoking among adolescents: The Muscatine study. American Journal of Preventive Medicine, 4(6), 336–348. https://doi.org/10.1016/S0749-3797(18)31205-3
  37. Sutherland, E. H. (1947). Principles of criminology (4th ed.). Lippincott.
  38. Sykes, G., & Matza, D. (1957). Techniques of neutralization: A theory of delinquency. American Sociological Review, 22(6), 664–670. https://doi.org/10.2307/2089195
  39. Warr, M. (2002). Companions in crime: The social aspects of criminal conduct. Cambridge University Press.
  40. Wolfowicz, M., Litmanovitz, Y., Weisburd, D., & Hasisi, B. (2021). Faces of radicalism: Differentiating between violent and non-violent radicals by their social media profiles. Computers in Human Behavior, 116, Article 106646. https://doi.org/10.1016/j.chb.2020.106646