In the target article, Redish et al. characterize drug use in terms of a constellation of vulnerabilities inherent in the brain's decision-making system. Unlike previous approaches to addiction, this characterization is flexible enough to deal with multiple features of addictive drugs and addicted individuals. We maintain, however, that the authors' sharp separation between the appraisal of situational cues and expected outcomes in motivating drug use fails to consider how cues can play more than a signaling role in habit formation or the reward system. The phenomenon of cue fascination suggests that any description of decision-making vulnerabilities in drug addiction must address how an addict's interest in drug cues can influence habit formation or the calculation of expected benefits.Footnote ¹ We believe that the authors' attempt to incorporate over-attention to drug cues within the tenth vulnerability, where they note that over-attention to drug cues may be related to changes in learning processes, substantially mischaracterizes this phenomenon, especially when paired with their model of robotic drug use. A fuller and more robust account of this key feature of addiction, when incorporated with the rest of the decision-making system, will strengthen the usefulness of this model.

Over-attention to drug cues has often been characterized largely in terms of classical conditioning. The drug cue is a conditioned stimulus that becomes paired with drug use and consequently acquires physiological and behavioral relevance as a precursor to drug ingestion (see, for example, Carter & Tiffany Reference Carter and Tiffany1999). We suggest that this model is incomplete. In many cases, the addict's interest in drug cues reflects an intrinsic fascination with the circumstances of the drug use itself. A simple example illustrates this point. Smokers attempting to quit often find themselves staring at packages of cigarettes. In some cases, smokers express an urge merely to touch or look at a cigarette. When reminded of their resolve to quit, these smokers might say, “But I don't want to smoke a cigarette, I just want to hold one.” While this phenomenon remains understudied, and the typical response to such statements is skepticism, accounts of drug use and addiction in humans and animals reveal similar episodes of intense interest in drug cues, often without corresponding expectations of drug rewards. In these circumstances, drug-addicted individuals experience a persistent absorption in environments or cues associated with past drug use. Rosse et al. (Reference Rosse, Fay-McCarthy, Collins, Risher-Flowers, Alim and Deutsch1993) describe compulsive crack cocaine foraging behavior (“chasing” or “geeking”) in addicts who have no expectation of finding lost fragments or pieces of the drug. A study of drug-use rituals similarly suggests that drug cues and paraphernalia take on added significance or meaning for drug users even in instances where those rituals delay or obscure the impact of taking the drug itself (see Zinberg Reference Zinberg1984). We suggest a model in which drug-related cues have an attraction for addicts that is in important respects independent from both a conscious desire to use drugs and what Redish et al. call a “robotic,” automatic, habitual response to cues.

This characterization of over-attention to drug cues corresponds to neurophysiological studies that pair drug cues with increased levels of dopamine in the mesocortical limbic system. The theory of incentive salience (Robinson & Berridge Reference Robinson and Berridge1993) suggests that the relation between drug cues and drug use is given not merely by their relation to hedonic response, in which cues become associated with the pleasure derived from the drug itself, but rather by their influence on core wanting (Berridge & Robinson Reference Berridge and Robinson2003). Although folk psychological accounts of motivation have typically paired “liking” with “wanting,” Berridge and Robinson (1993) suggest that these processes correspond to distinct neural substrates that are, in fact, dissociable in the laboratory. Increases in dopamine signaling have been traced not only to the ingestion of drugs, but also to the presence of drug cues. In both cases, there is a corresponding increase in incentive salience, or core wanting, even in the absence of increases in hedonic rewards. Although the theory of incentive salience has been enormously influential, exactly what feature of addicts' experience is referred to by “wanting” has been unclear. Robinson and Berridge (Reference Robinson and Berridge2004) have attempted to clarify “wanting” in terms that are useful for our present purposes. “[T]he process of incentive salience attribution … transforms the sensory features of ordinary stimuli or, more accurately, the neural and psychological representations of stimuli, so that they become especially salient stimuli, stimuli that ‘grab the attention’” (p. 352, emphasis added).

Drug cues “grab the attention” because they release dopamine. Indeed, Kotler et al. (Reference Kotler, Cohen, Segman, Gritsenko, Nemanov, Lerer, Kramer, Zer-Zion, Kletz and Ebstein1997) report that, for cocaine addicts, “cocaine cues by themselves increase dopamine release” (p. 251, emphasis added). We believe, however, that another well-documented feature of dopamine release offers insight into the psychological processes that underlie addicts' experience of drug cues. We refer to dopamine's association with the experience of novelty (see Freeman et al. Reference Freeman, Meltzer and Bunney1985; Garris et al. Reference Garris, Kilpatrick, Bunin, Michael, Walker and Wightman1999; Schultz Reference Schultz1998). Studies of paranoid schizophrenic patients note that “the dopamine system which under normal conditions is a mediator of context-driven novelty/salience in the psychotic state becomes a creator of aberrant novelty and salience” (Kapur et al. Reference Kapur, Mizrahi and Li2005, p. 61). This inappropriate marking results in patients reporting a subjective state characterized by fascination. This link between dopamine and novelty suggests that the experience of the drug cue for the addict is a very particular kind of experience: the fascination of a novel object. If this is true, then the addict is someone for whom a certain class of objects in the world – objects associated with drug cues – never becomes familiar, never becomes dull, never loses the fascination that accrues to what is new.

The link between dopamine release and drug cues is well established, as is the association between dopamine release and novelty. We suggest that putting these pieces together will bring out some important implications of Berridge and Robinson's work on drug cues. In the target article, Redish et al. consign the phenomenon of “over-attention” to cues to the catch-all “tenth vulnerability,” where they connect it to increased learning effects (see sect. 3.6). This is inadequate. Over-attention to drug cues, the phenomena that we call cue fascination, offers a new way to characterize susceptibility to drug use as well as new avenues for drug treatment.