The genetic architecture of reactive and proactive aggression: relations to disruptive behaviour problems through development

Abstract

Over the past two decades there has been increasing interest in the distinction between reactive (RA) and proactive aggression (PA; Card & Little, 2006; Polman, Orobio de Castro, Koops, van Boxtel & Merk, 2007). RA describes aggression that is defensive, impulsive and affect-laden, while PA comprises instrumental, calculated and typically unemotional aggressive behaviours (Vitaro, Brendgen & Tremblay, 2002). There is growing consensus that developmental models of RA and PA may help clarify risk pathways associated with the three disruptive behaviour disorders (DBD), attention-deficit hyperactivity disorder, oppositional defiant disorder and conduct disorder (Kempes, Matthys, de Vries & van Engeland, 2005; Raine et al., 2006; Waschbusch, Willoughby & Pelham, 1998). However, some confusion remains regarding etiological influences responsible for the differential shaping of aggression subtypes (Baker, Raine, Liu & Jacobson, 2008).Contributing the first elucidation of developmental relations between aggression subtypes and DBDs, caregiver ratings were ascertained from a large community sample of families of twins (aged 6-18 years) using the Australian Twin Behaviour Rating Scales (ATBRS, n=2082), and at 9-month follow-up via an online electronic version of the ATBRS (n=511). These data were partitioned according to two age cohorts (6-10 years, and 10-18 years) and subsequently submitted to a series of univariate and multivariate cross-sectional and longitudinal analyses.Consistent with previous research (e.g., Baker et al., 2008), cross-sectional models indicated strong influence of genes on both RA and PA—with genes showing greater effects on PA compared to RA, particularly in the older cohort. Multivariate longitudinal analyses revealed a substantial differential role for genes in the continuity of aggression subtypes over time for both RA and PA in the younger cohort, and for PA in the older cohort. At odds with etiological models that posit differential psychosocial factors underpinning RA and PA (Crick & Dodge, 1996; Dodge, 1991), no evidence was found for shared environmental effects on persistence in RA over time. Conversely, shared environmental effects explained a significant portion of covariance in childhood PA across time points. However, contrary to predictions derived from psychosocial explanations (e.g., Dodge, 1991), these shared environmental influences were mediated through a general risk for (undifferentiated) overt physical aggression.Data provide preliminary, albeit modest, evidence for a recent sequential model of RA and PA (i.e., Vitaro & Brendgen, 2005) which implicates RA as an early developmental precursor to PA. This model reformulates the role of social learning mechanisms invoked by seminal psychosocial formulations (i.e., Dodge, 1991). Specifically, the sequential hypothesis postulates that early RA which is reinforced in childhood, may increasingly come to be used operantly (i.e., instrumentally). In this way, RA is effectively converted to PA-type aggression. As predicted by this model, current data revealed the longitudinal relationship from RA to PA was predominantly explained by shared environmental influences (over and above a general risk for overt aggression) in the younger cohort. By contrast, the pathway from RA to PA in the older cohort was explained mainly by the influence of genes. These data provide some support for the suggestion that putative social learning mechanisms impact on the sequential pathway in childhood but not adolescence.The expected high levels of phenotypic overlap between aggression subtypes and DBDs were consistently explained by genes and nonshared environmental influences with the former accounting for the majority of covariance in most bivariate models. Age cohort differences in multivariate models of aggression subtypes and DBDs were in line with the smorgasbord hypothesis suggesting the effects of genes generally increase, while the effects of shared environment generally decrease, as a function of age. This pattern was most consistent and pervasive in RA-related models. Notable exceptions to this pattern occurred in bivariate models involving PA on the one hand, and hyperactivity or CD on the other—with these models showing greater influence of shared environment in the older versus younger age cohort. Additionally, the current research indicates a trend towards greater segregation of genetic effects across aggression subtypes and DBDs as a function of age, while conversely, shared environmental effects were more likely to simultaneously affect multiple syndromes in the older versus younger cohort.In regards to clinically specified risk pathways, relevant longitudinal analyses suggested that impulsivity confers only limited risk for future RA, while partial support was obtained for recent suggestions that both hyperactivity and PA are important cofactors in risk pathways associated with ODD and CD. Overall, the data that include DBDs broadly support RA and hyperactivity as key early markers of long-term risk.The online component of the study also included two neuropsychological tasks adapted for the internet and completed by 310 twin siblings. Representing an attempt to integrate multiple explanatory frameworks, this latter study evaluated differerential putative neuro-biogenic mechanisms underpinning RA and PA. Results from this study were inconclusive and issues pertaining to the delivery of neuropsychological tasks online are considered. In contrast to results from the adapted neuropsychological tasks, the online electronic ATBRS yielded higher data integrity and higher scale reliabilities than its original paper-and-pencil counterpart.All results are discussed and implications for future research and clinical practice relating to childhood and adolescent aggression are considered. Finally, limitations of the current research project are examined.