The neurocircuit signature of retaliation in adolescents with alcohol problems

Alcohol use disorder (AUD) exacts a devastating medicaland psychosocial toll worldwide. Chief among these impacts is violence while intoxicated. Compared to other substances of abuse, alcohol shows by far the clearest linkage to aggression, both anecdotally and in the laboratory. Blair and colleagues (1) scanned a sample of adolescents enriched for high levels of behavior, alcohol and cannabis-related problems and found that symptom severity of alcohol use disorder (AUD) but not cannabis use disorder (CUD) correlated with exaggerated neurocircuit responses during aggressive retaliation against unfair offers in the ultimatum game, a social-interaction economic task.

This finding provides a mechanistic account that explains in part the association between aggressive behavior and adolescent substance use and substance use disorders (SUD). On a psychometric level, AUD symptom severity on the AUDIT and CUD symptom severity on the CUDIT correlated with parent-rated aggression. On a behavioral level, youth with high CUD symptoms (and to some extent AUD symptoms) retaliated more aggressively against mildly unfair offers. These cross sectional findings reflect those of cohort studies that have long linked childhood syndromes of poor behavior control to increased risk for SUD (2). They also reflect longitudinal findings that chronic substance use in adolescence reduces executive functions (3). Most participants were recruited from a residential behavioral treatment program after weeks of supervised abstinence from all substances, thus minimizing acute detoxification effects on irritability. Their use of a monetary aggression paradigm also avoided potential sex differences in willingness to deliver physical aggression.

Although Blair et al (1) did not probe brain responses to provocation and retaliation while participants were actually intoxicated, readers might consider how their findings also provide a mechanistic account of alcohol-intoxicated aggression in youth most likely destined for it. Anecdotal lore of drunken individuals is replete with various typologies of them, such as online guides to “The (x) kinds of drunk people you meet in a bar.” A key subtype of intoxicated drinker that pervades these schemes is the “nasty drunk,” a volatile and aggressive individual. Only somewhat recently have typologies of real-life intoxicated behavior been formally probed with psychometric instruments. Winograd et al(4) surveyed drinking-buddy dyads on personality and behavior while sober vs intoxicated. Their four-factor solution yielded “The Hemingway” (largely resistant to acute alcohol effects on personality), “The Mary Poppins” (resistant but starting from a higher level of agreeableness while sober), “The Nutty Professor” (The introvert who comes out of his shell while inebriated), but also “The Mr. Hyde” (an individual who becomes less conscientious and agreeable while intoxicated).

Importantly, even in that ostensibly neurotypical sample, it was the Mr. Hyde subtype that showed elevated alcohol-related problems. This is not surprising in light of laboratory studies of acute effects of alcohol on aggressive behavior. These have shown that increased aggression while intoxicated tends to be specific to participants with higher levels of hostility, impulsivity, aggressiveness, or lower cognitive ability while sober. For example, delivery of more severe electric shocks while intoxicated was linked to both diminished consideration of future consequences (5) and lower baseline executive function (6). Similarly, delivery of money subtractions to a fictitious opponent was increased by acute alcohol only in participants with supramedian levels of aggressive responses during pre-alcohol baseline (7). The Blair et al finding (1) provides a neurocircuit account for why persons with elevated aggressiveness while sober are at risk of becoming Mr. Hyde when drunk.

As their primary finding, Blair and colleagues reported an interaction between AUD symptoms and severity of retaliatory aggressive response to increased activation of insula, dorsal anterior cingulate cortex (dACC), caudate and dorsolateral prefrontal cortex (DLPFC), and to a lesser extent periaqueductal gray (PAG). Put differently, youth with higher alcohol use and related symptoms activated these structures more robustly as a function of how much punishment they were dealing to the (fictitious) unfair opponent. These effects were robust amid follow-on analyses that controlled for medication usage and presence of conduct disorder. In contrast, neither AUD nor CUD symptomatology related to brain activation by initial perception of unfairness of ultimatum game offers themselves (i.e. activation time-locked to viewing of the fair or unfair offers).

In light of the contemporary concern over reproducibility of functional magnetic resonance imaging (fMRI) linkages to other phenotypes, it is notable that these are the same brain regions whose activation scaled with retaliation severity in a completely separate sample of neurotypical youth tested with that paradigm. These regions also include those that activate in response to violations of social norms in other economic social-exchange fMRI) paradigms (e.g. insula, caudate, midbrain), where these norm violations could be perceived as a type of negative reward prediction error (discussed in (9)). In this framing, exaggerated neurocircuit responses to progressively unfair offers in youth with more AUD symptoms also suggests they have an abnormality in the “frustrative nonreward” construct of the National Institute of Mental Health Research Domain Criteria (RDoC).

Moreover, retaliation intensity correlated with deactivation of the ventromesial prefrontal cortex (VMPFC) ((1) Table S1) irrespective of AUD or CUD symptomatology. The VMPFC shows connectivity with both the reward circuitry of the nucleus accumbens and with the behavior-control circuitry of the DLPFC, and so is thought to regulate behavior by integrating reward or loss prospects of a potential action or incentive to determine its net subjective value (8). By extension, its deactivation could be a reflection of blunted frontocortical regulatory activity over aggressive behavior, which usually entails prospects of both rewards and punishments for the behavior.

Blair et al interpret exaggerated recruitment of insula, dACC, DLPFC and other structures time-locked to retaliation selection in high AUD symptom participants as being in service of “an organized angry response,” or cognitive control. This account is plausible in light of the role of the dACC and DLPFC in behavior monitoring. Moreover, the relationship between AUD and task behavior was specific to the phase of the task when participants were choosing among retaliatory options. AUD symptoms did not relate to activation in the preceding task phase when the participant was initially presented with the (unfair) offers. However, the time interval between the offset of each offer and the onset of the retaliation-choice epoch ranged from only 0.5 to 2.5 seconds. Also, retaliation settings showed an orderly relationship to degree of economic unfairness of the offers. This linkage raises the possibility that some component this activation nevertheless indexed protracted neural activity elicited by the salience (“umbrage”) of an unfair offer itself, due to minor computational processing demands.

The dACC along with anterior insula are the primary nodes of the salience network that initially infuses environmental inputs with actionable meaning(10). As Blair et al allude in the introduction, an alternative account for the findings would be that the increased activation of these structures indexed the motivational salience of either the retaliation they were selecting, the provocation that elicited it, or more likely a combination of both. Importantly, at a relaxed threshold, AUD scores also correlated with retaliation-amplitude-modulated recruitment of PAG, which is well-understood to be a key structure that helps govern alertness and physiological responses to threat. Moreover, as the authors note, VMPFC regulatory circuitry de-activated with increasing retaliation rather than activated. Activation might be expected if participants were weighing costs vs benefits of sacrificing task earnings in order to retaliate. Finally, that high AUD-symptom participants would have more frontocortical regulation of their behavior stands in contrast to a large body of behavioral, electroencephalographic and neuroimaging findings of reduced frontocortical control in problem behavior youth.

If instead AUD symptomatology was linked more with exaggerated salience of unfair offers in these youth, it would not only comport with a wealth of clinical (and parental) lore of an exaggerated sense of “unfairness” and entitlement in adolescents with behavior problems, it would also provide a clear account of enhanced potential for aggression during alcohol intoxication in individuals who are already aggressive or hostile as a trait characteristic. Various theories of why acute alcohol intoxication increases aggression (or even risky sex) generally posit that alcohol intoxication increases the salience of and focus on immediate environmental stimuli while downregulating processing of deferred or non-salient contingencies. By extension, if provocations are unusually motivationally salient in adolescents with AUD symptomatology even while sober, the further enhancement of the salience of provocations during alcohol intoxication, coupled with impaired behavior control, could result in the intoxicated youth becoming highly aggressive in response to any affront (Mr Hyde).

In conclusion, the well-designed experiment of Blair et al (1) offers a window into the neurocircuit features of aggressiveness in adolescents who have already developed problems with alcohol. The findings at very least suggest an aberrant response during aggression in structures linked to both behavior control and to salience processing as a function of SUD symptomatology. Future work could perhaps include trials akin to the “dictator” game, where the participant simply sees unfair shares of what he or she is getting from the pot, and has no recourse. This would enable some disentanglement of whether high levels of SUD symptomatology or reactive aggression relates more specifically to the salience of provocation alone.