We commend Redish et al. for attempting to bring a measure of theoretical order to the processes that underlie this important scientific and social problem: drug addiction. However, several aspects of their article warrant further discussion.

First, there are important disparities between the literature cited in the target article and other work in the human substance abuse literature. The article proposes an “explanation for addiction as ‘vulnerabilities’ in an established decision-making system” (sect. 2.1, para. 1). In essence, the claim is that individuals who engage in maladaptive substance use exhibit deficiencies in decision-making, and that this inability to make correct non-drug-using decisions is a main cause of continued substance use. However, many substance abusers have the capacity to (and do) make rational choices from available options. For example, Hart et al. (Reference Hart, Haney, Foltin and Fischman2000) compared self-administration of cocaine when either a $5 cash or merchandise voucher was available as an alternative reinforcer, and they found that near-daily cocaine users' choice to self-administer cocaine was significantly lower when cash was the alternative. These results are consistent with other human laboratory data evaluating the influence of alternative reinforcers on cocaine-taking behavior (e.g., Hatsukami et al. Reference Hatsukami, Thompson, Pentel, Flygare and Carroll1994; Higgins et al. Reference Higgins, Bickel and Hughes1994), and suggest that experienced current cocaine users are capable of making consistent reasonable choices. Similar results have been obtained in opioid abusers. Comer et al. (Reference Comer, Collins and Fischman1997) demonstrated that heroin-dependent individuals' self-administration of the drug was decreased as the alternative money amount increased. Several other investigators have reported similar findings (e.g., Comer et al. Reference Comer, Collins, Wilson, Donovan, Foltin and Fischman1998; Rosado et al. Reference Rosado, Sigmon, Jones and Stitzer2005; Stitzer et al. Reference Stitzer, McCaul, Bigelow and Liebson1983; Vandrey et al. Reference Vandrey, Bigelow and Stitzer2007). A discussion of how these and related findings apply to Vulnerabilities 7 and 9 would enhance the proposed theory.

Moreover, research participants in substance abuse studies are required to complete extensive screening procedures prior to study enrollment (see Hart et al. Reference Hart, Haney, Vosburg, Rubin and Foltin2008). Typically they undergo psychiatric and medical examinations and several days of training on study procedures, requiring multiple screening visits. Once enrolled in the study – which may consist of alternating in-patient and out-patient phases lasting a couple of months –demanding schedules are imposed, requiring participants to do considerable planning, inhibit behaviors that may be inconsistent with meeting study schedule requirements (e.g., drug use), and delay immediate gratification. All of these have been identified in the target article as potential “failure points.” The conclusion that substance abusers are handicapped by compromised decision-making skills that contribute to their addiction is inconsistent with the findings of human laboratory studies of substance abusers.

Second, most of the data supporting the proposed theory come from laboratory animal studies, with limited consideration of the social setting in which the drugs were administered. However, human drug-taking behavior is extremely sensitive to social context. Hart et al. (Reference Hart, Haney, Vosburg, Comer and Foltin2005), for example, found that experienced marijuana smokers self-administered significantly more delta-9-tetrahydrocannabinol (Δ⁹-THC) capsules (the primary psychopharmacologically active component of smoked marijuana) during social/recreational periods compared with non-social/recreational periods. This and other findings that drug self-administration is influenced by social factors (e.g., Doty and de Wit Reference Doty and de Wit1995; Foltin et al. Reference Foltin, Fischman, Brady, Capriotti and Emurian1989) raises questions about the applicability of the proposed theory to human substance abuse.

Third, the target article defines addiction “as the continued making of maladaptive choices” (sect. 1, para. 1). Appropriately, the definition is derived from the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) and International Classification of Diseases (ICD-10). However, both the DSM-IV-TR and the ICD-10 refer to the phenomenon of interest as substance dependence and not addiction. It is not clear how the proposed theoretical model deals with individuals who use “addictive” substances, but do not meet criteria for substance dependence. While this omission has implications for several aspects of the model, it is particularly relevant to the discussion of the mediating role of increased dopamine activity on the failure point of the habit system. Overwhelmingly, illicit drug users do not display symptoms associated with dependence (Substance Abuse and Mental Health Services Administration 2007), and many of the substances used by these individuals enhance dopaminergic activity acutely. In our experience with human cocaine and methamphetamine users, for example, many non-dependent individuals report using these drugs several times per week and, in some cases, for many years. In addition, many patients are maintained on daily doses of amphetamines, dopamine agonists, for medical reasons (e.g., to treat attention-deficit/hyperactivity disorder [ADHD]), yet do not meet criteria for a substance use disorder. In fact, some investigators have reported that stimulant medication treatment for ADHD is protective against subsequent development of substance abuse (Wilens et al. Reference Wilens, Faraone, Biederman and Gunawardene2003). Given the widespread licit and illicit use of dopamine-enhancing drugs, the model's failure to address the non-dependent use of so-called drugs of abuse is a serious omission.

Finally, Redish et al. contend that their theory has implications for prevention and treatment. We appreciate that an exhaustive consideration of research on substance abuse would be beyond the scope of the target article. However, we regret the authors' failure to discuss what is arguably the most successful substance abuse treatment strategy: contingency management (for review, see Higgins et al. Reference Higgins, Heil and Lussier2004). This omission might be related to the proposed theory's primary focus on neurobiological explanations of substance dependence, whereas contingency management (consistent with the human laboratory data cited earlier) views such behavior as sensitive to environmental consequences. It is clear that, under certain conditions, drug users can make rational choices that limit their drug intake. A theory that delineates the conditions under which drug users will and will not make maladaptive decisions about their drug use would be of enormous scientific and practical value.