Modeling Effects of SES and Working Memory on Developmental Trajectories of Impulsivity, Sensation Seeking, and Delay of Gratification from Early to Late Adolescence

Introduction: Working memory (WM) is a core executive cognitive function that may be impaired in children raised in families of low socioeconomic status (SES), potentially explaining one source of impulsivity in poor youth (Evans & Schamberg, 2009). Given the robust associations observed between impulsivity and risk-taking during adolescence (Romer et al, 2011), identifying the precursors of impulsivity is important from a prevention standpoint. Our goals in this study were to–1) model trajectories of different dimensions of impulsivity, including sensation seeking (SS), acting without thinking (AWT), and inability to delay gratification (IDG), from early to late adolescence; 2) identify their longitudinal associations with SES; and 3) assess how much of that covariance is due to weakness in WM.  

Method: Data from five annual assessments of an urban sample of adolescents aged 10-12 years (at baseline) were analyzed using latent growth curve modeling to identify trajectories of SS (Zuckerman, 1971), AWT (Eysenck & Eysenck, 1980), and IDG (Madden & Bickel, 2010). Baseline assessments of WM and SES (Hollingshead, 1975), coded such that higher scores implied better WM and higher SES, were used to predict individual differences in intercepts and slopes of SS, AWT and IDG trajectories. WM was assessed using four standardized tasks: Corsi block-tapping, letter two-back, digit span backwards, and a spatial WM task.

Results: Model fit criteria suggested a steady linear increase in SS trajectories from ages 10-17, whereas a quadratic function provided a better fit for AWT and IDG trajectories, with peaks around mid-late adolescence (15-16 years). WM emerged as a significant predictor of individual differences in the intercepts for all three dimensions, being positively related to SS, B(SE)=0.06 (.03), p<.05, but negatively related to AWT, B(SE)=-0.03 (.01), p<.01, and IDG, B(SE)=-2.76 (1.23), p<.05. SES was also related to the intercepts of AWT and IDG, but had no influence on SS trajectories. However, only some of the effect of SES on AWT (32%) and IDG (23%) intercepts was mediated by weak WM; the rest of its effect was direct, B_AWT(SE)=-0.002 (.001), p<.05,  and B_IDG(SE)=-0.28 (.09), p<.01.

Conclusions: Weakness in WM serves as one of the mechanisms by which poverty impacts two dimensions of impulsivity associated with adolescent risk-taking. The fact that WM was only associated with the intercepts of AWT and IDG is consistent with other research suggesting that weakness in WM develops early in poor families (Hackman et al, 2010). Nevertheless, it is possible to intervene preventatively by providing WM and self-control training to at-risk children, prior to their entry into adolescence (Diamond & Lee, 2011). The efficacy of such approaches needs to be tested.