The model proposed by Redish et al. considers vulnerabilities within decision systems based on expectancy-value assumptions. Further understanding of processes leading to addiction can be gained by considering other inputs to decision-making, particularly affective associations with behaviors. This consideration suggests additional decision-making vulnerabilities that might explain addictive behaviors.

Decision-making is a complex activity for which emotions and affects are essential. Maladaptive choices depend on negative affects. Vulnerabilities to drug or non-drug objects depend on previous psychopathological comorbidities. Premorbid individual characteristics allow us to understand why some individuals – and not others – enter into the addiction cycle. Moreover, plasticity of reward neurocircuitry is, at least in past, responsible for these vulnerabilities leading to compulsive drug use.

Redish et al. present a compelling, interdisciplinary, unified framework of addiction. The effort to integrate pathological gambling is especially important, but only the vulnerability of misclassifying situations is described in detail as being linked directly to this disorder. This commentary focuses on further developing the comprehensiveness of this framework for pathological gambling using over-fast discounting as an illustrative example.

Redish et al. rely too much on a rational and innate view of decision-making, when their emphasis on variation, their integrative spirit, and their neuroscientific insights point towards a broader view of why addiction is such a tenacious problem. The integration of subjective, sociocultural, and evolutionary factors with cognitive neuroscience advances our understanding of addiction and decision-making.

For all its problems, the microeconomic “rational addiction” theory had the appeal of making clear predictions about the effects of various drug policies. The emerging picture of the “what” and “how” of addiction is far more complex. Addiction scientists might help bridge the science–policy gap by devoting more attention to the “whom” and “when” of addiction.

Redish et al. trace vulnerabilities in habit and planning systems almost exclusively to pharmacological effects of addictive substances on underlying brain systems. As we discuss, however, these systems also can be disrupted by purely psychological factors inherent in normal decision-making and everyday behavior. A truly unified model must integrate the contribution of both sets of factors in driving addiction.

A central theme of the unified framework for addiction advanced by Redish et al. is that there exists a common value or incentive process controlling Pavlovian and instrumental conditioning. Here we briefly review evidence from a variety of sources demonstrating that these incentive processes are in fact independent. Clearly the influence of Pavlovian predictors and goal values on choice offer distinct potential targets for pathologies of decision-making.

People become addicted in different ways, and they respond differently to different interventions. There may nevertheless be a core neural pathology responsible for all distinctively addictive suboptimal behavioral habits. In particular, timing models of reward learning suggest a hypothesis according to which all addiction involves neuroadaptation that attenuates serotonergic inhibition of a mesolimbic dopamine system that has learned that cues for consumption of the addictive target are signals of a high-reward-rate environment.

Redish et al. propose a constellation of vulnerabilities inherent in the brain's decision-making system. They allow over-attention to cues a minor role in drug addiction. We think this is inadequate. Using the established links among drug cues, dopamine, and novelty, we propose a fuller account of this key feature of addiction, which we call the phenomenon of cue fascination.

Neuroscientists and psychologists have proposed a variety of well-supported theories to explain addiction. Many of these theories suggest that addiction results from a single process or dysfunction across all of its forms. The authors of the current review, in contrast, have used a well-defined theoretical account of decision-making to outline the variety of dysfunctions that could account for addictive behavior.

The model of addiction proposed by Redish et al. shows a lack of fit with recent data and models in psychological studies of addiction. In these dual process models, relatively automatic appetitive processes are distinguished from explicit goal-directed expectancies and motives, whereas these are all grouped together in the planning system in the Redish et al. model. Implications are discussed.

By summarizing, in a single piece, various current perspectives on addiction, Redish et al. have performed a useful service to the field. Their central message is that addiction comprises many vulnerabilities rather than a single vulnerability. Such a message may not be new, but it is worth repeating.

In our target article, we proposed that addiction could be envisioned as misperformance of a decision-making machinery described by two systems (deliberative and habit systems). Several commentators have argued that Pavlovian learning also produces actions. We agree and note that Pavlovian action-selection will provide several additional vulnerabilities. Several commentators have suggested that addiction arises from sociological parameters. We note in our response how sociological effects can change decision-making variables to provide additional vulnerabilities. Commentators generally have agreed that our theory provides a framework within which to site addiction and treatment, but additional work will be needed to determine whether our taxonomy will help identify and treat subpopulations within the addicted community.