Cue‐Exposure Treatment: Time for Change

S ir—We thank Drs Niaura, Drummond and Hughes ( Drummond 2002 ; Hughes 2002 ; Niaura 2002 ) for their comments and appraisals of our manuscript, 'Applying extinction research and theory to cue-exposure addiction treatments' ( Conklin & Tiffany 2002 ). Although we agree with many of their points, we wish to clarify our position and comment on where we disagree with certain of their conclusions and suggestions. In our paper, we reviewed the cue-exposure treat­-ment literature and made recommendations for how extinction-based treatments might be enhanced through the application of contemporary animal research and models of learning. In order to establish the overall efficacy of this treatment, we conducted a meta-analytical review that revealed, much as many in the field would suspect, that it is not particularly effective in treating substance abuse. Given this, we wanted to determine possible reasons for the failure of this potentially promising treatment and attempt to find ways in which it might be improved. We decided to return to cue-exposure's roots, namely, animal extinction work, and see what has been discovered since this treatment's inception. We hoped that contemporary animal research might inform efforts to improve cue-exposure treatment efficacy. What we found was a wealth of information that has direct implications for the focus and design of human cue-exposure treatment. In short, the very way in which extinction is conceptualized has changed. Older views of extinction as unlearning have been replaced by the idea that extinction represents new learning. Moreover, modern animal research reveals several threats to the strength of extinction training that must be guarded against if behavior is to be altered. Interestingly, human cue-exposure treatment, as it stands, does not protect addicts against these threats. We concluded that it is no wonder human cue-exposure treatment fails so often, and suggested ways in which the animal work can directly inform efforts to improve human cue-exposure treatment. Although our paper appeared to have been well received by both Niaura and Drummond, we recognize that we may not have been entirely clear on our position with regard to new ways of conceptualizing extinction. Niaura interpreted our assertion that extinction may best be viewed as new learning, rather than unlearning, as suggesting that, somehow, appetitive responses to drug cues may never be extinguished. However, that is not our position, nor could it be in light of 80 years of extinction demonstrations. Extinction is a real phenomenon; it simply may not operate in the way we once presumed that it does. Also, it is imperative that this modern understanding of extinction processes be applied to the design and execution of cue-exposure treatments. Certainly, the ability of a cue to elicit some sort of drug-seeking response may never be completely abolished. Nevertheless, people and other animals can learn that, under certain conditions, behavior that once led to drug no longer has that outcome. Niaura also asserted that, in light of the presumed intractability of the association between drug cues and appetitive responses, addicts must be taught to cope with the motivational consequences of those cues. He suggested further that only through 'higher-order learning' could addicts come to control their highly automated responses to drug-cues. His position is that coping-response training combined with cue-exposure therapy is the only hope addicts might have to control their over-learned drug-seeking behavior. Interestingly, coping-response training seems akin to teaching humans or animals to respond to cues in ways that are incompatible with the original appetitive response. Extinction training may simply be a different way of achieving that goal. However, animals can learn to stop responding in the presence of cues that previously signaled reward. That is, animals can be extinguished, and they do not need coping-response training to do so. Moreover, although there may be merit in combining cue-exposure with coping-response training, we know of no compelling empirical evidence that such training is overwhelmingly successful as a treatment for addiction. There is much that can and should be done to improve basic cue-exposure therapy without confusing matters by combining it with another treatment approach of uncertain impact. Drummond, on the other hand, while finding our review intelligent and provocative, raised concerns that our meta-analysis of the cue-exposure treatment literature might be both premature and methodologically unwise. Presumably, in Drummond's opinion, our conclusions about the general inadequacy of cue-exposure therapy were unduly dismal. Ironically, Niaura claimed that it was not necessary to conduct a quantitative analysis to 'conclude that the methods used to date have failed, miserably'. Drummond is certainly correct when he reminds us that a meta-analysis might combine studies that are disparate and methodologically weak. However, the traditional narrative review is prone to all these shortcomings and more. We fail to see how a narrative reviewer can protect him or herself from biases inherent in the subjective appraisal of the cue-reactivity literature. Related to this concern was Drummond's suggestion that not all studies showed insignificant effects for cue-exposure treatment. Indeed, Drummond and others can point to a few significant effects in past cue-exposure treatment studies. However, those few effects cannot mitigate the fact that the majority of studies on the efficacy of cue-exposure do not find that this therapy is effective. The mixed pattern of effects across these studies is troublesome for two reasons. First, as we have no basis for determining why cue-exposure therapy sometimes works and more often does not, how can we make it better? Secondly, Drummond and others may be assigning too much weight to a few isolated significant effects. Several years ago, Meehl (1978) pointed out that it is absurdly easy to prove that the null hypothesis is false. As Meehl stated, '[p]utting it crudely, if you have enough cases and your measures are not totally unreliable, the null hypothesis will always be falsified, regardless of the truth of the substantive theory'(Meehl 1978, p. 822). From Meehl's perspective, given that we keep performing more cue-exposure treatment studies and those studies are conducted with more patients and increasingly sensitive measures, we will find that cue-exposure will occasionally produce significant effects relative to control conditions. However, even in those instances, will we be impressed with the clinical efficacy of the treatment? Unlike Drummond and Niaura, Hughes expressed reservations regarding the aims and conclusions of our manuscript. He interpreted our purpose to be that of comparing extinction in animals to human attempts at abstinence. That was not our point. Rather, our aim was to apply animal extinction work to ongoing efforts to promote human abstinence via cue-exposure treatment. That is, to improve a treatment based on principles of animal extinction by incorporating contemporary animal findings that suggest the conditions under which extinction works best. However, Hughes does make an important point regarding the applicability of animal work to the human drug use situation, one that we ourselves have addressed in a recent commentary (Tiffany & Conklin 2002). Animal models of human behavior will be maximally relevant to the extent that they duplicate the critical components and circumstances of the human experience. As Hughes suggests, there are indeed elements of drug seeking and use more relevant to the human experience that have not been modeled in animal laboratory paradigms. Although we recognize clearly the need for animal models that capture key components of the human addiction and relapse scenario, consideration of that work goes beyond the main thrust of our manuscript. As we see it, there are four ways to proceed with cue-exposure treatment: (1) give up on it; (2) combine treatment with other therapies in hopes that there will be some synergistic advantage to the combination; (3) continue to modify treatment programs somewhat unsystematically, hoping that one of the variations will somehow emerge as consistently successful; or (4) refine cue-exposure in light of modern research and theory on extinction. We do not advocate the first two options. Although this may seem to support Drummond's view that we are 'essentially cue-exposure enthusiasts', our position is that while cue-exposure appears to have considerable theoretical merit, the efficacy of this technique must be established empirically. Moreover, we should not proclaim cue-exposure treatment ineffective, or begin to supplement it with additional therapy techniques to ­bolster its efficacy, before we have subjected it to rigorous scientific tests. After reviewing current cue-exposure treatment studies, we conclude that those tests have yet to be conducted. We also do not support the third option. This trial and error method of attempting to isolate and incorporate the most effective treatment com­ponents from studies showing marginally significant effect sizes may, through serendipity, lead to some small improvements in treatment. However, this unsystematic ap­-proach is slow going and inefficient, given other options. Simply put, why bother trying to find a needle in a ­haystack when there is a needle store next door? We ­propose that the wealth of contemporary animal theories and findings is that needle store. While there are a myriad choices, not all of which will aid our efforts, we are at least at a promising starting point. Clearly then, we support the fourth option. Cue-­exposure should be refined through the application of contemporary research and theory on extinction. If cue-exposure therapy continues to fail in the context of a systematic, theoretically informed research program, then we would think it time to abandon the therapy or consider combining it with some other intervention such as coping-response training. Even if treatment fails, we will undoubtedly learn a great deal about the application of animal research on extinction within the exigencies of the human condition.