“Drugs are bad.” “Addiction is a disease.” These claims can polarize popular thinking about drug consumption and tacitly influence research. Against this background, Müller & Schumann's (M&S's) proposal should be welcomed for its good sense. Most drug use never meets diagnostic criteria for addiction. M&S offer an instrumental framework underpinned by an information-processing model for explaining non-addictive consumption. They argue that drugs alter mental states – a fact that once learnt can be instrumentalized. Given that altered mental states may help to achieve desired ends, drugs can be purposively consumed because they are reliable means to those ends. M&S suggest eight ends served by consumption: (1) improved social interaction; (2) facilitated sexual behaviour; (3) improved cognitive performance; (4) coping with stress; (5) alleviating psychiatric symptoms; (6) novel perceptual and sensory experiences; (7) hedonia or euphoria; and (8) improved physical and sexual appearance. Finally, M&S argue that non-addictive drug consumption is both an adaptation and adaptive in modern society. Despite the risk of consumption becoming addictive and so plausibly maladaptive, non-addictive drug use potentially enhances survival and reproduction in virtue of its instrumental effects.

It is crucial to distinguish three claims. First, that non-addictive drug use is instrumental; second, that it is evolutionarily adaptive; and third, that its desired ends are easy to comprehend. Use can be instrumental without being adaptive or even easily comprehensible. The philosopher G. E. M. Anscombe (Reference Anscombe1957) famously wondered whether we can ever make sense of someone wanting a saucer of mud. Wanting a saucer of mud is unlikely to be adaptive. Without further explanation, it is not easy to comprehend. But if a saucer of mud is what you want, you display instrumental rationality if you mix dirt with water and dump it in a saucer: That behaviour is an instrumental means to the desired end.

Facilitated sexual behaviour may strike us as evidently adaptive. The adaptive case for hedonia may prove to be harder to establish. But evolutionary considerations aside, all of the ends M&S identify are intelligible human goods: sex, social interaction, cognitive capacity, new experiences, pleasure, and relief from stress and distressing symptoms. On the whole, it is natural for people to use available means, including drugs, to achieve these ends. Moreover, as with all instrumentally learned behaviour, we should expect there to be an account of the information-processing, learning, and memory underpinning it. Hence, the heart of M&S's framework boils down to this: non-addicts consume drugs for good reasons, that is, in order to achieve desired ends that are intelligible human goods, and, given this, we need an information-processing account of how knowledge of the effects of drugs is acquired, stored, and used to drive context- and end-specific consuming behaviour. This is indeed good sense. But we should wonder why research on drug consumption is such that it needs to be said.

Aside from the polarization of popular thinking, one likely reason is that much research on drug consumption focuses on addiction: where use seems maladaptive, and where it is hard to comprehend how anyone could desire such ends. Alongside tolerance and withdrawal, diagnostic criteria for addiction include increasing focus on use at the expense of other goods, continued use despite desire for control and efforts to achieve it, and use-consequent physical and psychological harm (APA 2000). M&S suggest that their framework may help to develop clinical interventions that reduce the likelihood of instrumental use becoming addictive, by suggesting the need for education, identification of high-risk individuals, and understanding of an individual's pattern of instrumentalization so that alternative means to the ends served by drugs can be learned. These clinical interventions are already routine (Petersen & Mcbride Reference Petersen and Mcbride2002). The bolder question that M&S's framework invites, but which they do not pursue, is whether addictive consumption is also instrumental behaviour. It may not be adaptive, and it may seem from the outside no more comprehensible than wanting a saucer of mud, but it may yet be an instrumental means to desired ends.

To deflect this question, M&S gesture at research showing that as use escalates, control devolves from the prefrontal cortex to the striatum, in line with a shift from action-outcome to stimulus-response learning (Everitt & Robbins Reference Everitt and Robbins2005). In rats, drug use that is initially goal-directed and sensitive to devaluation of outcome becomes increasingly habitual: triggered automatically and insensitive to mild devaluation. But, in humans, behaviourally complicated and temporally extended habits that have developed out of action-outcome learning are typically still subject to some executive control, through the formation of decisions and the exercise of will. Habits make control hard, but they do not extinguish it.

As Heyman (Reference Heyman2009) emphasises, large-scale survey data suggest that addiction peaks in adolescence and early adulthood but, in the majority of cases, has resolved permanently, without clinical intervention, by the early thirties (Anthony & Helzer Reference Anthony, Helzer, Robins and Regier1991; Kessler et al. Reference Kessler, Berglund, Demler, Jin, Merikangas and Walters2005a; Reference Kessler, Chiu, Demler, Merikangas and Walters2005b; Stinson et al. Reference Stinson, Grant, Dawson, Ruan, Huang and Saha2005; Warner et al. Reference Warner, Kessler, Hughes, Anthony and Nelson1995). Addicts tend to “mature out.” The exceptions are addicts who suffer from additional psychiatric disorders: Chronic, relapsing addiction is associated with psychiatric comorbidity (Regier et al. Reference Regier, Farmer, Rae, Locke, Keith, Judd and Frederick1990). These data, in combination with M&S's good sense, suggest a reason for thinking that even the worst cases of addiction are instrumental. Such drug consumption is a habitual means to desired ends (4) and (5), namely, coping with high levels of stress and alleviating psychiatric symptoms (Pickard & Pearce, forthcoming/anticipated Reference Pickard, Pearce and Levy2012). Until these underlying problems are addressed and/or alternative but equally effective means of achieving these ends secured, addicts may have little incentive to resolve to control their drug habit, despite consequent harm. M&S are plainly right that non-addictive use is instrumental. The harder question is whether addiction is as well.