- Top of page
- Supporting Information
There are several distinct facets of behavioural impulsivity, many of which overlap with subcomponents of executive (dys)function . For example, impulsive decision-making can be assessed with delay discounting procedures, in which participants make choices between small rewards that are available immediately versus larger rewards that are available after a delay . Disinhibition is assessed with computerized tasks such as the Stop-Signal  and Go/No-Go tasks, both of which establish a dominant motor response that participants are required to occasionally inhibit. Risk-taking can be assessed with tasks such as the Balloon Analogue Risk Task (BART) , in which participants attempt to win rewards by risking what they have accumulated up to that point. Dependent measures obtained from these tasks reflect distinct underlying concepts , which suggests that impulsivity is not a unitary construct . None the less, performance on each of these measures is associated with heavy drinking and alcoholism [7-9] and with other substance use disorders .
Initial experimentation with alcohol begins during adolescence. For example, in the United Kingdom approximately 24% of 12-year-olds report at least one episode of alcohol consumption, rising to 77% of 15-year-olds . Individual differences in behavioural impulsivity are associated with drinking behaviour and alcohol problems in adolescents . Theoretically, elevated impulsivity may pre-date alcohol involvement and serve as a risk factor for the development of heavy drinking and alcohol problems once individuals begin to experiment with alcohol . Consistent with this, longitudinal studies demonstrate that high levels of disinhibition are predictive of the development of heavy drinking and alcohol problems several years later [14-16]. Regarding other behavioural impulsivity measures, individual differences in the rate of increase in risk-taking during early adolescence (but not absolute levels of risk-taking) are predictive of subsequent alcohol involvement . Furthermore, individual differences in delay discounting predict the likelihood of starting smoking , although the relationship between delay discounting and subsequent heavy drinking in adolescents has not been investigated.
Adolescence is a critical stage of brain development, and the maturational changes that occur may render adolescents particularly sensitive to neuroadaptations that underlie development of alcohol dependence. For example, developmental brain changes that influence reward processing and impulse control are essential to the long-term development of self-regulation and adaptive decision-making [19, 20]. Neuroimaging studies have shown that adolescents, relative to adults and young children, show a heightened neural response to rewards in the nucleus accumbens [21, 22]. This heightened sensitivity occurs within the context of immature processing of reward and risk within the orbitofrontal cortex, a key region involved in inhibitory control [22, 23]. These features of brain development may render adolescents vulnerable to increased disinhibition, impulsive decision-making and risk-taking as consequences of heavy drinking. Consistent with this, studies with rodents demonstrate that the extent of neuronal loss following binge alcohol exposure is more pronounced in adolescents than in adults , as is increased probability discounting caused by heavy drinking . There is also some preliminary evidence for neurocognitive deficits arising from alcohol exposure in human adolescents [26, 27]. However, there is no direct evidence that heavy drinking during adolescence leads to increased impulsivity.
Our goal in the present study was to investigate the relationships between alcohol involvement and performance on behavioural impulsivity tasks among adolescents. We performed a cross-lagged prospective study involving a large sample of adolescents who were aged 12 or 13 years at the beginning of the study and tested each participant five times over 2 years (every 6 months). We recruited participants in this age range as UK government data  indicate that alcohol consumption in British adolescents tends to increase rapidly between the ages of 12 and 15 years. We hypothesized reciprocal prospective relationships between alcohol involvement and impulsivity. Specifically, we predicted that (i) individual differences in impulsivity would predict alcohol involvement at subsequent time-points, and (ii) individual differences in alcohol involvement would predict impulsivity at subsequent time-points.
- Top of page
- Supporting Information
Our cross-lagged models indicated that individual differences in performance on three behavioural impulsivity tasks each predicted a composite index of alcohol involvement 6 months later. These prospective relationships were consistent, as they were evident at three of the four 6-month intervals that we tested. However, we found no evidence for hypothesized alcohol-induced increases in behavioural impulsivity: individual differences in alcohol involvement did not predict subsequent impulsivity at any time-point.
As hypothesized, individual differences in disinhibition (as assessed with the Stop-Signal task) at the baseline assessment predicted alcohol involvement 6 months later. This finding is consistent with previous reports [14, 15], and to our knowledge it is the first such demonstration of this relationship in an adolescent sample from outside North America. Equally important, our study is the first to demonstrate that delay discounting and risk-taking also predict alcohol involvement after fairly short follow-up periods of 6 months. Although previous studies have demonstrated cross-sectional associations between alcohol involvement and delay discounting in adolescents , no previous studies have investigated whether individual differences in delay discounting are associated with subsequent changes in alcohol consumption and problems, and so our study contributes important new data. Regarding risk-taking, a previous report  found that the rate of increase in risk-taking predicted a very small increase in the likelihood of alcohol involvement at subsequent assessment points [odds ratio (OR) = 1.02] in a sample who were slightly younger (9–12 years of age) than our own sample at the beginning of the study. In contrast, our results show that the absolute level of risk-taking predicted alcohol involvement only 6 months later, a relationship that was seen across multiple time-points.
We did not detect any evidence of changes in behavioural impulsivity as a consequence of heavy drinking during adolescence. Our study was the first to investigate this issue directly, and this is an important finding. However, it is possible that a longer follow-up period, or a focus on adolescents ‘at risk’ for development of substance use disorders  rather than a random sample as in the present study, may have yielded a different outcome. On a related note, while the frequency of drinking alcohol increased over time, the majority of our participants were drinking infrequently even at the end of the study. These are limitations of our study and an important avenue for future research, but we highlight the high financial costs associated with conducting longitudinal research over such long periods of time.
The model fit for our cross-lagged models was not exceptional (ideally, both CFI and TLI should be 0.95 or above, and RMSEA should be 0.05 or below). However, our fit indices can be described as acceptable . While our sample size was large, model fit may have been better with an even larger sample. In the delay discounting and risk-taking tasks, participants were responding for hypothetical rather than real financial rewards. We opted to use hypothetical rewards for ethical and practical reasons, and on the basis of previous studies that obtained comparable results from delay discounting tasks when real versus hypothetical rewards were used [45-47]. In addition, other studies have shown that discounting rates for hypothetical monetary rewards are associated with alcohol use  and addictive behaviours more generally . Regarding the BART, although no previous studies have directly contrasted risk-taking behaviour when participants are responding for real versus hypothetical rewards, it is notable that risk-taking as measured by the BART is associated cross-sectionally with alcohol use regardless of whether hypothetical  or real  monetary rewards are used. However, one recent study suggests that the predictive validity of discounting tasks is superior when real rather than hypothetical rewards are used , and therefore it is important to replicate our findings using real financial rewards in the delay discounting and BART tasks.
Other limitations are our failure to record participant ethnicity, other drug use and trait (self-reported) impulsivity, so we were unable to evaluate the relative importance of behavioural measures of impulsivity after controlling for these other variables. Finally, we note some strengths of our study: the dropout rate was low, as was the percentage of missing data, thereby ensuring a high level of statistical power for all our primary analyses. No previous studies have tracked changes in performance on behavioural impulsivity tasks in relation to alcohol involvement over such an extended period of time, and we believe that the current study makes a very important contribution in this regard.
In summary, in the present study we explored longitudinal relationships between various indices of alcohol involvement and performance on behavioural impulsivity tasks. Across multiple time-points we found that disinhibition, delay discounting and risk-taking predicted alcohol involvement only 6 months later. Importantly, we found no evidence to suggest that heavy drinking had an impact on performance on any of these behavioural measures.