Conditioned learning in alcohol dependence - Wiley Online Library

British Journal of Addiction (1990) 85, 725-743

COMMENTARY

Conditioned learning in alcohol dependence: implications for cue exposure treatment D. COLIN DRUMMOND, TROY COOPER & STEVEN P. GLAUTIER Addiction Research Unit, National Addition Centre, Institute of Psychiatry, 101 Denmark Hill, London SE5 8AF, United Kingdom

Summary A review of the literature pertinent to cue exposure treatment in alcohol dependence is presented. Psychological models of relapse, based on conditioning and social learning theories, are critically evaluated. In particular, attention is drawn to the potential implications for cue exposure research and treatment of an interaction between Pavlovian and operant conditioning, problems with the application of the concepts of arousal and craving and the importance of a systems model to understand physiological responses. It is concluded that no study has so far demonstrated a link between conditioned responses to alcohol-related cues and relapse, an assumption on which cue exposure treatment is based. Further, the evidence for the effectiveness of cue exposure as a treatment is lacking. Promising research directions are identified.

Introduction Wikler (1965) proposed a model of relapse of drug use in abstinent opiate addicts which has inspired an era of research into the effects of environmental stimuli (or cues) on drug taking behaviour. He suggested that environmental cues associated with drug withdrawal come to elicit conditioned withdrawal-like responses which precipitate further drug use in the abstinent addict. He also suggested that these conditioned responses (CRs) remain unaltered by conventional addiction treatment. A later study by Ludwig, Wikler & Stark (1974) which explored the effects of exposure to alcohol related cues (ARCs) in alcoholics, went further by concluding: "Obviously any therapeutic approach, whether it be insight, behaviourally or pharmacologically orientated, that does not recognize the powerful, evocative effects of interoceptive and exteroceptive stimuli... and that neglects to provide techniques for modifying the strength of these effects will likely be destined for failure" (p. 547)

While several authors have questioned the importance of CRs to environmental cues in the relapse process (McAuliffe, 1982; Chaney, Roszell & Cummings, 1982; Marlatt, 1985a) the numerous studies which have explored the existence, significance and modification of these responses in both animals and humans attest to the impact which Wikler's theory has had. The technique of cue exposure, originally described in relation to the treatment of cat phobia (Freeman & Kendrick, 1960), has gained considerable popularity in the treatment of both phobic and obsessive-compulsive neurosis (Marks, Hodgson & Rachman, 1975). More recently cue exposure has been applied to the treatment of alcohol and drug dependence (Blakey & Baker, 1980; Rankin, Hodgson & Stockwell, 1983; Laberg & EUertsen, 1987; McLellan et al., 1986) and owes much to this and Wikler's early pioneering work. In comparison to the neuroses, however, we argue that no study has so far demonstrated that cue exposure has any

725

726

D. Colin Drummond, Troy Cooper & Steven P. Glautier

influence on clinical outcome in either alcohol or drug dependence. This view stands in sharp contrast to the optimistic views concerning the effectiveness of cue exposure expressed by some authors. Niaura et al. (1988) have provided a useful review of some of the literature pertinent to theoretical issues in cue exposure research and in doing so, have stimulated considerable debate. This paper aims to extend the debate into several important areas which have not so far been considered. In order to orientate the reader there is some overlap with Niaura et al. (1988) in terms of the description of existing theoretical models, but important distinctions will be made in our interpretation of the evidence. In particular we draw attention to the possible implications for cue exposure research of an interaction between Pavlovian and operant conditioning, problems with the application of the concepts of arousal and craving and the importance of applying a systems models to understand physiological responsivity in cue exposure. We argue that these issues have important implications for cue exposure treatment and suggest new directions in which research could usefully be directed. While reference will be made to comparable research in the opiate dependence field, this review is largely restricted to the alcohol literature. We feel that the diversity of the literature in the two fields combined, would confuse rather than clarify the issues involved. Having restricted the scope of the paper in this way however, many of the theoretical issues raised here may have direct relevance to cue exposure in opiate dependence. The paper is divided into two main sections. In the first section we will compare the competing psychological models which have been invoked to explain the process of relapse, exploring their possible relationship to cue exposure as a treatment method. In doing so we intend to alert the reader to sources of conceptual confusion and identify important and so far unanswered theoretical questions which serve to restrict further progress in cue exposure research. The second section is focused on the experimental evidence for the existence of conditioned respones (CRs) to alcohol-related cues (ARCs) in dependent and non-dependent drinkers and provides a critique of studies utilizing cue exposure as a treatment technique. It is suggested that while subjective and psychophysiological responses to ARCs have been identified, their exact nature and interrelationship is uncertain. Further, the relationship between CRs to

ARCs and relapse has yet to be conclusively established. We intend to show that while cue exposure is a promising approach to the treatment of drug and alcohol dependence it's effectiveness has not so far been demonstrated.

Competing models of relapse Conditioned responses in relapse The importance of conditioned responses to environmental stimuli in the process of relapse was asserted by Wikler (1965). This was based on his observation that heroin addicts tended to return to drug use even after months of abstinence when exposed to an environment in which drug use had previously taken place. This appeared to be due to the re-emergence of withdrawal symptoms which could not be attributed to elimination of the drug from the body. He suggested that formerly neutral environmental stimuli become associated with the unconditioned stimulus (UCS) of a declining drug plasma level which leads to the unconditioned response of withdrawal symptoms (UCR) (see Fig. 1). These environmental stimuli (or cues) become conditioned stimuli (CSs) which, when presented in the absence of the UCS, invoke a conditioned withdrawal response (CR). Declining plasma drug level (UCS)

. Withdrawal Symptoms (UCR)

Environmental > Conditioned Stimuli (cues) Withdrawal (CS) (CR) Figure 1. Model of conditioned withdrawal.

According to this model, the connection between a conditioned withdrawal response and relapse is that drug use is motivated by the relief or avoidance of withdrawal symptoms which are aversive in nature. In effect, aversive CRs to ARCs act as discriminative stimuli (SDs) for operant drinking behaviour. While it should be noted that Wikler later acknowledged that these CRs may be, in some cases, drug-like and that drug taking may also be motivated by the pursuit of positive hedonic effects of the drug, this early theory remains popular with contemporary researchers in the cue exposure field (Pomerleau, 1981; Melchior & Tabakoff, 1984; McLellan et al., 1986; Laberg & EUertsen, 1987; Dolinsky et al, 1987; Bradley & Moorey, 1988). Siegel has proposed an alternative explanation of the observed phenomenon of CRs to drug cues (for

Cue exposure treatment in alcohol dependence a review see Siegel, 1988). He argues that environmental stimuli which precede drug administration, rather than follow drug withdrawal, come to elicit drug compensatory (or antagonistic) CRs which oppose the unconditioned effects of the drug and thus account for the development of tolerance. Subsequent animal research has demonstrated conditioned antagonistic responses to the analgesic (Siegel, 1975; Krank et al., 1981), thermic (Siegel, 1978), sedating (Hinson & Siegel, 1983), gastrointestinal (Raffa et al, 1982) and lethal (Siegel et al, 1979) effects of morphine. Similarly, conditioned antagonistic respones have been demonstrated in animals to the hypothermic (Le et al, 1979; Mansfield & Cunningham, 1980; Crowell et al, 1981) hypnotic (Melchior & Tabakoff, 1981) and lethal (Melchior & Tabakoff, 1982) effects of ethanol. It would appear that the distinction between models of Siegel and Wikler lies in the temporal relationship between the cues and the unconditioned effects of the drug. In reality this distinction is extremely difficult to make, in that drug cues are often present both during withdrawal and for some time after the onset of drug agonist effects when an individual takes a drug whilst experiencing withdrawal. What is perhaps of greater importance is that both models predict CRs which are opposite in direction to the unconditioned drug effects. Several studies have identified conditioned agonistic responses to drug cues (for a review see Eikelboom & Stewart, 1982). Eikelboom & Stewart (1982) suggest that the main factor which determines the direction of the CR to drug cues is the site of drug action. Drugs which act on the efferent limb of the central nervous system tend to produce antagonistic responses, whereas drugs acting on the afferent limb lead to conditioned agonistic responses to drug cues. Any particular drug may have different sites of action. Thus, drug cues may produce both agonistic and antagonistic responses in the same individual. Conditioned agonistic responses to drug cues may be important in relapse because of their similarity with unconditioned drug effects which serve to maintain drug taking behaviour by positive reinforcement (Stewart, deWit & Eikelboom, (1984). Solomon & Corbitt's (1974) opponent process theory predicts that a drug engenders an O; or agonistic process which leads to a ^ or antagonistic process which serves to maintain homeostasis. It provides a means of explaining the coexistence of conditioned agonistic and antagonistic responses in

727

the same individual to the same environmental cues at different times. Unlike Siegel's model, environmental cues can come to elicit either a- (CSa) or b(CSb) processes. Solomon (1977) later speculated that the direction of the CR would depend mainly on the temporal relationship between the cue and the a- or 6-process. Donovan & Chaney (1985) have elaborated this theory to account for differences in response between dependent and non-dependent drinkers. Essential to this is Opponent Process Theory's predictions regarding the timing and strength of the Zi-process. Solomon & Corbitt (1974) suggested that with repeated administrations of the drug, the bprocess will beconie greater in magnitude and more rapid in onset to a point where only a minimal agonistic effect will be observed, followed rapidly by a much larger and longer lasting antagonistic effect. Donovan & Chaney (1985) propose that a CSa will lead to a predominantly agonistic effect in a relatively non-dependent drinker whereas in the severely dependent drinker the same CSa will lead to a predominantly antagonistic process. Another theory which predicts relapse following both agonistic and antagonistic responses to drug cues, is that of Baker, Morse & Sherman (1987). Drawing on the work of Wise & Bozarth (1982) they postulate two distinct, mutually inhibitory neural networks which are responsible for the motivation of drug taking behaviour and hence, relapse. Drug cues can activate both an appetitive and a withdrawal relief network leading in either case to relapse. In common with Donovan & Chaney, they predict that in the more severely dependent, drug cues will predominantly stimulate the withdrawal relief network whereas in the less dependent, the positive network will be more important. A further prediction of this theory is that the effects of cues will be affected by the position of the drug taker in the cycle of intoxication and withdrawal. In abstinent or withdrawing addicts drug cues will activate the withdrawal relief network. So far we have emphasized the possibility of the existence of both agonistic and antagonistic responses to environmental cues. In contrast to more parsimonious models (e.g. Niaura et al, 1988), it has been suggested that both types of response may theoretically occur within the same individual either simultaneously or at different times or in response to different cues. The importance of this theoretical prediction is that, as we shall see later, the direction of the CR may be important in determining operant

728


interpretations of the effects of cue exposure. Also it may be particularly difficult for an alcoholic in treatment to say that he is craving (by definition, thinking of relapse) if he has publicly committed himself to abstinence. Clearly, subjective responses cannot be ignored. Further research is needed to elucidate the physiological correlates of subjective states and in particuCraving and relapse The concept of 'craving' has been important in the lar the effects of setting on attributions. Until this development of cue exposure research. It has been has been done we argue that subjective responses to argued that the subjective experience of conditioned '• ARCs should not be assumed to be indicative of any craving in response to drug cues is the important particular underlying physiological state or behavdeterminant of relapse (Ludwig & Wikler, 1974). ioural intention. It is possible in technical accounts We have considered craving separately from condi- to describe subjective responses to ARCs in terms of tioned responses because it represents an ambiguous positive or negative affect (as suggested by Baker et concept which serves to impede progress in the cue al., 1987), or indeed by using a variety of adjectives, exposure field. Ludwig, Wikler & Stark succinctly without recourse to the ambiguity of the concept of craving. described the status of craving in 1974, thus: responses. Further, the extinction of CRs to ARCs in cue exposure treatment may have unpredictable effects depending on the appetitive or aversive nature of the CR and its relationship to the operant response.

Although the related phenomena of 'craving' and 'loss of control' have been invoked as explanations of relapse, there is little current agreement . . . about their nature or even existance . . . [It has been] argued that the construct of craving represents a superfluous, logical tautology since it is often defined by subsequent drinking behaviour. As a remedy to this conceptual confusion Ludwig & Wikler (1974) put forward the theory that craving represented the 'cognitive symbolic correlate of conditioned subclinical withdrawal' and that craving in turn leads to alcohol-seeking behaviour and relapse. This definition is also tautologous. This problem is not helped by expanding the definition of craving to include appetitive or drug-like subjective states. Baker, Morse & Sherman (1987) have used the term 'urges' to describe the subjective states associated with responses to drug cues. They suggested that 'positive-effect urges' are associated with appetitive cues whereas 'negative-effect urges' are associated with aversive cues. They relate these urges to the previously described underlying and separate neural networks. While this theory is more sophisticated than that of Ludwig & Wikler's view of craving, it is also subject to the problem of attributional bias. Attributional bias in an individual's linguistic labelling of an internal physiological state has been observed to exert a powerful effect in other experimental settings (Schachter & Singer, 1962; Nisbett & Wilson, 1977). The subjective experience described as a 'craving' or an 'urge' may represent a variety of different internal states between individuals and give rise to erronenous

Extinction of CRs to environmental cues Several animal experiments have demonstrated that CRs to drug cues can be extinguished by repeated presentation of the CS without the unconditioned stimulus of drug administration. Siegel et al. (1979) demonstrated loss of tolerance to the lethal effect of morphine and concluded that antagonistic responses to morphine had been extinguished. Similar results have been observed in respect of tolerance to the analgesic effect of morphine (Siegal, Sherman & Mitchell, 1980) and the hypothermic effect of ethanol in rats (Mansfield & Cunningham, 1980). These results suggest that, at least in animals, CRs can be extinguished by unreinforced exposure to drug cues. Rankin et al. (1983) have proposed an alternative model of the effects of repeated exposure to alcohol cues in alcoholics. They draw a comparison between alcoholism and phobic neurosis suggesting that drinking alcohol in response to drug cues is similar to phobic avoidance of a feared stimulus. While such a comparison may be theoretically unwarranted they proposed that unreinforced exposure to alcohol cues would lead to extinction of a drinking response, citing in support of this hypothesis the work of Baum (1969). Baum, observed that rats who were trained to escape from a shock by jumping onto a ledge initially continued to try to escape when the ledge was removed, but over repeated trials the avoidance response extinguished. Reiss (1980) has examined the implications of expectancy theory of conditioning in the extinction of phobic avoidance responses. Of particular poten-

Cue exposure treatment in alcohol dependence tial relevance to cue exposure in drug and alcohol dependence are the lessons drawn from RescorlaWagner theory (1972) of Pavlovian conditioning. During extinction of CRs, any new stimuli (e.g. therapists) presented in compound with the target stimuli (i.e. the ARCs), will gradually acquire inhibitory properties (become safety signals in the case of phobic subjects) as the excitatory properties of the target stimulus declines. No further conditioning will take place once the inhibitory properties of the safety signal counterbalance the excitatory properties of the feared stimulus. This will leave some of the excitatory properties intact, 'protected' by the existence of safety signals. If we wish to extinguish CRs to ARCs the Rescorla-Wagner theory suggests that maximum extinction will be achieved when there are no safety signals present. Thus therapist and other novel stimuli should be minimal or gradually phased-out in the therapeutic setting. If, however, we assumed that in theory it is possible to extinguish CRs to ARCs and that this would decrease the probability of an operant drinking response and hence relapse, what can we expect to happen if an alcoholic then takes alcohol. Conditioning theory would predict that the CRs to ARCs would be reinstated in a much shorter time than was required to acquire CRs previously. While extinction of CRs to ARCs may in theory be expected to reduce the likelihood of relapse in alcoholics, assuming that such CRs are determinants of relapse and that extinction leads to reduced rather than increased operant drinking responses, vulnerability to reinstatement of CRs following an initial priming dose remains a problem. Further, several experiments have shown that animals may experience spontaneous reinstatement following extinction which would further limit the potential effects of cue exposure. Finally, there is the problem of generalization. In practise it would be impossible in a laboratory setting to extinguish CRs to every ARC which a drinker had been exposed to during an entire drinking career: the animal experiments so far described have had the advantage of establishing CRs in a controlled setting and therefore the experiments have the knowledge of which specific drug cues to target in extinction. The most that could be hoped for is that extinction of CRs to the most salient stimuli (if they can be identified) might generalize to other stimuli the natural drinking environment. Alternatively, extinction could be attempted in the natural environment, a suggestion

729

which was made by Blakey & Baker (1980) to which we will return later.

The relationship between classical and operant conditioning in relapse Research in the cue exposure field has so far assumed a simple relationship between CRs to environmental stimuli and relapse, predicting a decreased likelihood of relapse following extinction of CRs. We now present evidence derived from other areas of animal learning research which suggests that there exists a more complex relationship between classical and operant conditioning than has so far been considered by cue exposure researchers and which has important implications for cue exposure treatment. Classical and operant conditioning can be argued to be different types of learning, the distinction being more than just a procedural one. The focus of classical conditioning research is the learning which takes place as a result of an organism's experience events which occur independently of its behaviour. The focus of operant conditioning research is the learning which takes place as a result of an organism's experience of events which take place as a consequence of its behaviour. These two types of learning are clearly not always separate: many situations occur in which both types of learning may operate. The question is, what is the nature of their relationship? We need to consider what the results of this interaction could be for the way in which drug or alcohol use is learnt and for the way in which we might attempt to intervene in those learnt behaviours.

A. Central emotional state theory. One theory of the interaction between the two types of learning was presented by Rescorla & Solomon (1967). They suggested that central emotional states generated by Pavlovian or operant reinforcers were important determinants of the interaction between the conditioning processes. The theory was elaborated because of the difficulties experienced by experimenters demontrating the function of conditioned responses as, for example, discriminative stimuli. Williams (1965) found that lever pressing for food preceded salivation. If the salivatory response had served a discriminatory function it should have occurred before the lever pressing. The value of reinforcement in either the Pavlovian or operant paradigm can be characterized as

730


aversive or appetitive. Organisms can learn about what predicts the occurrence or absence of either type of event/reinforcer. It has been proposed that if the value of the UCS in the Pavlovian paradigm or the reinforcer in the operant paradigm are the same, then they become mutually facilitative. If they are opposite in value, for example if one were appetitive and the other aversive, then the operant and classically conditioned responses become mutually inhibitory. An example of this has been described by Rescorla & Lolordo (1965) who found that dogs trained to avoid a shock by pressing a bar would increase their reponding (operant) if they heard a tone that had previously been paired with shock (classical). Since shock is aversive, any stimuli associated with it arouse an aversive central emotional state. So if dogs are bar pressing motivated by the central emotional state of fear, a tone signalling shock will increase their central emotional state, thereby increasing their rate of operant responding. Conversely, if the dogs were bar pressing to avoid shock and heard a tone which had previously been associated with food, a positive central emotional state, bar pressing decreases because the positive central emotional state inhibits the negative central emotional state. Many of the predictions made by Rescorla & Solomon have been tested experimentally. Needless to say there have been failures to confirm the predictions as well as successes. For example, Lolordo (1971) found a signal for food produced increased responding for food, a success, whereas Azrin & Hake (1969) found the opposite, a failure. Nevertheless, the implications of these findings for extinction of CRs to ARCs in cue exposure are interesting. In the typical cue exposure treatment setting, ARCs are repeatedly presented without ingestion of alcohol. Let us suppose that conditioned antagonistic responses (CAntRs) occur, and that as these CRs are opposite in direction to the usual drug effects and are aversive in nature. These CRs should be extinguished as the UCS, alcohol, signalled by the CSs (i.e. the ARCs) does not occur. We could suppose that the central emotional state associated with the reinforcement, drinking, is negative since the CAntRs are aversive, and that drinking is an avoidant response. But also suppose that the act of drinking as an operant response to ARCs is associated through learning history from the beginning of drinking with a positive central emotional state (drinking, not to relieve withdrawal, but for a positive, hedonic drug effect). If the CAntRs to ARCs are extinguished, the concommitant negative

central emotional state is reduced. But according to two process theory this negative central emotional state may have been opposing the positive central emotional state motivating operant drinking. So the positive central emotional state now producing the operant response will be disinhibited, making the strength/probability of operant responding more likely. Drinking will not necessarily decrease at all even though CRs have been extinguished.

B. Expectancy theory. Another theorization of the interaction between different kinds of learning can be found in expectancy theory, as outlined by BoUes (1972). The general implication of expectancy theory is that by a form of practical inference an organism will decide upon appropriate behaviours based on its representation of action to consequence in the operant paradigm. This theory does not differ from conditioning theory in terms of its predictions in many areas. One differential prediction, however, concerns the appetitive case, which may be of particular interest in looking at conditioned learning in drinking behaviour. Whereas central emotional state theory predicts that an appetitive instrumental paradigm will produce increased responding when a CS signals appetitive reinforcement, because the two central emotional states are additive, expectancy theory predicts a decrease. Conversely, a decrease in positively reinforced responding when the CS is aversive is predicted by two-factor theory because the two central emotional states are opposite in value, whilst expectancy theory predicts the opposite. The reasoning for this in expectancy theory is that since an appetitive CS indicates that reinforcement is going to occur, the organism does not see the need to make an operant response for reinforcement that it believes is going to occur anyway and will decrease responding. Conversely, where the CS predicts an absence of a desirable event the responding for positive reinforcement will increase because the organism calculates that more work is required to produce the reinforcer for which the CS predicts non-occurrence. But this may not be the end of the story either. Mackintosh (1983) has argued that theories concerning classical and operant interactions must also take account of peripheral response competition in the two types of associative learning. For example, Lolordo, McMillan & Riley (1974) in an experiment with pigeons, found that the interactions between Pavlovian and operant schedules could be influenced by arranging the experiment so that

Cue exposure treatment in alcohol dependence peripheral CRs competed with, or facilitated the operant response (for a review see Schwartz & Gamzu, 1974). Clearly, the relationship between classical and operant conditioning is complex. We do not have a theory which adequately explains and predicts the outcomes of interactions between classical and operant conditioning. In addition we do not have a clear idea of the kinds of operant and Pavlovian contingencies involved in drug taking. Therefore we should not assume that extinguishing CRs to drug cues will have the desired effect on the variable of central interest, drug taking.

Cognitive models of relapse The concept of expectancy has been applied to explain the process of relapse. BoUes (1972) suggested that expectancy has both informational and incentive components which will influence future behaviour. A drinker will have a belief about the effects of alcohol (e.g. relaxation) and attach a value to the desirability of such an effect (Marlatt, 1985a). Thus a positive outcome expectancy describes a desirable and rewarding outcome and a negative outcome expectancy describes an undesirable or aversive outcome. Marlatt suggests that in a given situation a drinker will take alcohol if a positive outcome expectancy is attached to drinking. A further elaboration of this model is that craving is synonymous with positive outcome expectancies which can be elicited by drug cues through classical conditioning (Marlatt, 1985a, p. 140). Marlatt, however, suggests that conditioned learning rarely precipitates relapse in comparison to, amongst other factors, positive and negative mood states or social pressure to drink (Marlatt, 1978; Chaney, O'Leary & Marlatt, 1978). Attributional processes, however, influence a subject's view of causal mechanisms in relapse (Schachter & Singer, 1962). Thus, in a situation where social pressure is exerted on an individual to drink it is possible that alcohol related cues are also present (e.g. the sight and smell of alcohol) but that relapse is attributed to social pressure rather than CRs (Heather & Stallard, 1989). Indeed social pressure in itself could act as a discriminative stimulus in an operant drinking response. Mood states may operate in a similar way. Alternatively, a negative mood st^te could bring about relapse through its similarity with a conditioned antagonistic response by a process of generalization. Recent work has suggested that mood states

Ti\

can elicit conditioned withdrawal (Childress et al, 1987; Litt et al, 1988). Bandura's (1977, 1981, 1982) theory of selfefficacy may provide an important additional dimension to our understanding of the effects of cue exposure on relapse. He suggests that "In the face of difficulties people who entertain serious doubts about their capacities slacken their efforts or give up altogether, whereas those who have a strong sense of efficacy exert greater effort to master the challenges" (Bandura, 1981, p. 201). Self-efficacy is believed to interact with outcome expectancy in such a way that an alcoholic faced with the possibility of consuming alcohol will drink in the combined presence of high positive outcome expectancies and low perceived self-efficacy. The effect of cue exposure may be to increase positive outcome expectancies as we have suggested and relapse would occur if this is combined with low self-efficacy. Bandura also predicted that increased aversive arousal, in particular that over which one had no control, would lead to decreased self-efficacy (Bandura, 1981). Thus arousal induced by CRs to ARCs may have a direct effect on self-efficacy as well as outcome expectancy. This view is supported by a recent experimental study of cue exposure in alcoholics (Cooney et al., 1987). It is also possible that the experience of a high level of arousal in the cue exposure setting which is not followed by consumption of alcohol may lead to increased self-efficacy. In effect, an individual may form an internal attribution of mastery over his environment and urges to drink alcohol. The implication for cue exposure as a treatment is that subjects should be encouraged to regard their participation as involving the active implementation of effective coping responses. The phasing out of safety signals, such as the presence of a therapist, during cue exposure treatment may also increase the likelihood of the formation of such an internal attribution. That self-efficacy may be an important factor in relapse is suggested by the work of Condiotte & Lichtenstein (1981) who found that ex-smokers who relapsed following treatment tended to have low self-efficacy compared to non-relapsers. Di Clemente (1981) reported a similar finding. There is strong empirical evidence of the power of selfefficacy judgements in predicting drinking behaviour (Rist & Watzl, 1983; Annis & Davis, in press). It has also been suggested that skill training to increase self-efficacy may be useful in preventing

732


alcoholic relapse (Chaney, O'Leary & Marlatt, 1978; Annis & Davis, in press). However, that selfefficacy judgements may be self-fulfilling prophesies and act as a method of dissonance reduction, rather than representing a causal mechanism, remains to be determined. Nevertheless, we concur with the views of Niaura et al. (1988) and Marlatt (1988) that the question as to whether cue exposure treatment may prevent relapse not only by extinction of conditioned responses but also by change in outcome expectancy and self-efficacy deserves further consideration. Self-efficacy theory is not incompatible with conditioning theories of relapse, but instead offers a different level of description and explanation.

Conclusions regarding the competing theoretical models of relapse We have described the development of conditioning theories of relapse since Wikler's original formulation (1965). There seems to be little doubt that CRs to drug cues develop following repeated administration of drugs in animals and that it is possible to extinguish these responses during unreinforced cue exposure. The terms 'conditioned agonistic' and 'antagonistic response' (CAgR, ContR) afford greater conceptual clarity than 'conditioned withdrawal'. If, as has been suggested, both types of response can occur within the same individual either simultaneously or at different times or in response to different cues, and that the interaction between Pavlovian and operant contingencies is more complex than has so far been assumed, the effects of extinction during cue exposure may be highly unpredictable. We recommend a more detailed investigation of the effect of the directionality of CRs on operant drinking responses and the effects of extinction. Assuming that extinction of CRs leads to reduced drinking behaviour, however, the effects of cue exposure as a treatment will theoretically be limited by the problem of reinstatement. The best one might therefore expect from cue exposure treatment is an increased latency to relapse rather than a reduction in the extent of a relapse. In addition, the problems of safety signals and generalization leads one to the conclusion that cue exposure in the natural drinking environment, with minimal therapist presence, might be more likely to influence drinking behaviour than exposure conducted in a laboratoryjetting; a^view which has been previously expressed, but not so far adequately investigated.

Because of the ambiguity of the term 'craving' and the effect of attributions we have suggested that subjective responses to ARCs should not be assumed to reflect any particular underlying physiological response or behavioural intention. Subjective states can be described in technical studies in terms of a variety of adjectives without recourse to the concept of craving. This should go a long way towards resolving conceptual confusion in the field. Social learning theorists such as Marlatt (1985a) have de-emphasized the importance of CRs to ARCs in explaining relapse, by applying an overly narrow definition of conditioned stimuli and reinforcers (Heather & Stallard, 1989). This has led to a dichotomy in relapse research between conditioning and social learning theories, with a resultant lack of cross-fertilization of ideas. The possibility that the effects of cue exposure treatment may be mediated either instead or in addition by alterations in outcome expectancies or self-efficacy deserves further consideration. It is possible that cue exposure shares a common effect with a more 'office based' skills training or relapse prevention approach, and that rather than simply extinguishing CRs, the alcoholic undergoing cue exposure treatment may acquire coping strategies to avoid future drinking. We would argue that self-efficacy and conditioning theories of relapse are complementary rather than contradictory, differing in terms of their level of description and explanation.

Laboratory studies or conditioned responses to alcohol related cues in human alcohol dependence So far we have reviewed the theoretical developments in the study of CRs to environmental cues since Wikler's (1965) original formulation. From the animal literature, there is considerable evidence in support of the existence of both conditioned agonistic (CAgR) and antagonistic (CAntR) responses which may be modified by extinction. In reviewing laboratory studies which have examined CRs to ARCs in human alcohol dependence there are three main questions which need to be addressed. What is the relationship between CRs to ARCs and alcohol dependence? What is the nature of these responses? Is there any evidence to implicate such responses in relapse? Although some of these issues have previously been explored by Niaura et al. (1988), we offer here an alternative interpretation of the evidence. Comparison of studies in this area is complicated

Cue exposure treatment in alcohol dependence by the different aims of investigators and the variety of methodologies employed. In broad terms this work can be divided into five categories: (1) Studies of cued responses in alcohol dependent and social drinkers. (2) Studies comparing responsivity to alcoholrelated and neutral stimuli. (3) Studies of the influence of expectancy on CRs. (4) Studies of the development of CRs to ARCs in social drinkers. (5) Priming dose studies. While several investigators have employed a combination of these methodologies, studies of cued responses have a broadly similar design. The subject is presented with alcohol related stimuli which often include the sight and smell of alcohol. This has been described as cue exposure. With the exception of priming dose studies, alcohol is not consumed. Physiological and subjective measurements are taken before and after cue presentation. The difference between these measurements is deemed to be a response.

CRs to ARCs in alcohol-dependent and social drinkers An important problem in the interpretation of these studies is that few investigators have specified the degree of alcohol dependence of their subjects. Newlin (1985a) has found antagonistic responses to alcohol cues in social drinkers given an nonalcoholic drink which they were led to believe contained alcohol. He has also found enhanced antagonistic responses to placebo in sons of alcoholics suggesting the possibility of a genetic predisposition to cued responses (Newlin, 1985b). Several studies have observed differences between severely dependent and moderately or nondependent subjects (Stockwell et al., 1982; Rankin et al, 1982; Kaplan et al, 1983; Laberg, 1986; Corty et al, 1988). Although these findings are consistent with conditioning theory predictions, it is possible that conditioned responses to alcohol cues may develop after only a few exposures. It is perhaps not surprising therefore that several authors have failed to find consistent differences between alcohol dependent and social drinkers. Bearing in mind that the limitations inherent in the term 'alcoholic', it will be used in future sections as a form of shorthand for a subject who has presented to a clinical service with alcohol-related problems and an

733

unspecified degree of alcohol dependence since greater descriptive precision is not possible from the majority of studies under review. A further problem in determining the specificity of conditioned responses to alcohol cues in alcoholics is indicated by a number of studies comparing the psychophysiology of alcoholics and controls. Coopersmith (1964) for example found that alcoholics were more responsive than controls in terms of skin conductance. This finding was replicated by Chandler et al. (1975) during a demanding neuropsychological task. They also found greater heart rate variability in alcoholics in spite of finding no difference on these measures during a resting period. In contrast to this Chotlos and Goldstein (1967) reported that alcoholics' heart rate increased and skin conductance decreased during silent periods of the experiment. These comments notwithstanding, several studies have found differences between alcoholics and controls. Increased subjective desire for alcohol, skin conductance level and heart rate have been observed to differentiate alcoholics from controls following CE (Kaplan et al, 1983; Kaplan et al, 1985; Laberg, 1986). A greater number of studies have found no differences on these measures (Pomerleau et al, 1983; Dolinsky et al, 1987; Monti et al, 1987; Corty et al, 1988). The predominant explanation for these changes in HR and SCL in response to ARCs is that of an increase in physiological arousal (Kaplan et al, 1985; Laberg, 1986). This view of arousal (Malmo, 1959) has been the subject of considerable criticism. For example Lacey (1959) has described the concept of autonomic response specificy suggesting that HR and SCL can increase or decrease depending on the nature of the stimulus. Lacey later concluded (1967) that physiological responses should be viewed in the context of interactive feedback systems which bear a complex relationship to emotion. Similarly, in reviewing studies to tonic HR responses to stimuli Elliot (1974) observed "There are in these data, in short, grounds for scepticism of any view of the motivational significance of HR" (p. 536). Such observed responses are also inconsistent with a model of CAntRs to ARCs (described above) since alcohol has been consistently shown to have the unconditioned effects of tachycaria (Docter & Perkins, 1960; Docter et al, 1966; Holmberg & Martens, 1955; Mendelson & LaDou, 1984) and increased skin conductance (Carpenter, 1957;

734


Greenberg & Carpenter; 1957; Levenson et al., 1980). Two studies have found a peripheral hypothermic response to CE (Laberg, 1986; Dolinsky, 1987) and two have found no measurable change (Kaplan et al, 1983; Corty et al., 1988), although the latter study found that temperature was affected by the order of presentation of neutral and alcoholic stimuli in alcoholics. Again this would appear antithetic to a CAntR model where conditioned hyperthermia is predicted (Crowell et al., 1981; Le et al., 1979; Mansfield & Cunningham, 1980). That is, unless one considers the interactive nature of physiological response systems (Lacey, 1967). Alcohol not only has the unconditioned effect of decreasing temperature, it also produces vasodilation (Docter & Bernal, 1964; Ritchie, 1965; Roth & Sheard, 1947). Thus an antagonistic vaso constrictive response, which has been observed in social drinkers in response to CE (Newlin, 1985a), could account for this peripheral hypothermic response. Salivation in response to CE has been measured directly by measuring weight gain in dental rolls or indirectly by measuring digastricus muscle electromyographic activity, and has been reported to differentiate alcoholics from controls (Pomerleau et al., 1983). Subsequent studies have failed to replicate this finding (Kaplan et al., 1985; Monti et ah, 1987; Corty et al, 1988), although two of these studies found that salivation in alcoholics was influenced by the order of stimulus presentation terms of both an increase (Monti et al, 1987) and a decrease (Corty et al, 1988) in salivation when alcohol presentation followed a neutral stimulus. Pomerleau et al. (1983) have explained their observations in terms of both appetitive arousal and conditioned responding to alcohol stimuli. Beazell & Ivy (1940) noted a fall in salivation following alcohol consumption and Kissin et al. (1959) observed a decrease in salivia sodium, interpreting the change as due to parasympatholytic effects of alcohol. Pomerleau et al's findings may indicate the existence of a salivatory CAntR to ARCs. Until these findings have been replicated however, the specificity of such a response remains uncertain. Other responses to ARCs which have been reported include raised serum insulin and glucose and decreased cortisol (Dolinsky et al, 1987) the significance of which is at present unclear. One early study found pupillary dilation in response to the smell of a favourite alcoholic drink in alcoholics (Kennedy, 1971). In support of the social learning model of relapse, cue exposure has been found to

increase positive and decreased negative outcome expectancies of alcohol's effect (Cooney etal, 1984; Laberg, 1986; Cooney et al, 1987; Stockwell et al, 1982). In this context Cooney et al (1987) also noted that alcoholics reported decreased self-efficacy, and desire for alcohol was found to be inversely correlated with confidence in being able to resist alcohol.

Alcohol-related versus neutral stimuli That cued responses are specific to alcohol and not merely the experimental setting has been demonstrated by several studies comparing the effects of alcohol-related and neutral stimuli in alcoholics (Pomerleau et al, 1983; Kaplan et al, 1983; Eriksen & Gotestam, 1984; Kaplan et al, 1985; Laberg, 1986; Cooney et al, 1987; Monti et al, 1987). As suggested earlier these experiments are subject to methodological difficulties which serve to affect the possibility of observing a difference between conditions. The order of presentation of the stimuli, expectancy of receiving alcohol and the possibility of a carry over effect from one stimulus to another may all serve to reduce the likelihood of finding an observed difference between responses to alcoholic and neutral stimuli. It adds greater weight to these positive findings therefore that differences between alcoholic and neutral stimuli have been observed in terms of increased desire for alcohol, SCL, HR, salivation and decreased peripheral skin temperature.

Expectancy and CRs to ARCs The third group of studies are those which involve manipulation of instructional set or expectancy through a full or partial realization of the balanced placebo design (Marlatt & Rosenhow, 1980, 1981). Priming dose studies have also made use of this experimental manipulation to which reference will be made in a later section. The prediction of this experimental design is that responses to alcohol cues will be dependent on the belief concerning the alcohol content of a stimulus rather than its actual content. Several studies have demonstrated the importance of this effect in CE. Kaplan et al (1983) found that alcoholics who were told that a drink contained alcohol responded with increased desire for alcohol compared to controls. This finding has been replicated in several similar studies (Kaplan et al, 1984; 1985; Laberg, 1986) both in terms of subjective and physiological

Cue exposure treatment in alcohol dependence responses (see also priming dose studies). These results are in keeping with a conditioning model in that a positive expectancy of receiving alcohol can be seen as 'part of the CS (complex) associated with past alcohol consumption' (Shapiro & Nathan, 1986). Further, Newlin (1985a) has demonstrated in social drinkers that CAntRs only occur in those subjects who consume a placebo drink which they believe to be alcohol. Staiger & White (1988) on the other hand have de-emphasized the importance of expectancy in conditioned responses to alcohol cues by demonstrating that a conditioned heart rate response occurred irrespective of subject's expectancy. This contradicts the proposal made by Newlin (1986) and Shapiro & Nathan (1986) that expectancy acts as a determinant of the conditioned compensatory response. The expectancy hypothesis should not be rejected on the basis of this one study. On balance, expectancy should be seen as an important determinant of a CR.

Experimental induction of CRs to ARCs in social drinkers Several investigators have sought to establish CRs to novel stimuli associated with alcohol consumption in social drinkers. This approach is theoretically preferable to post hoc explanations of responses to alcohol cues in established alcoholics and avoids the complicating factor of differences in pre-existent physiological responsivity between alcoholics and non-alcoholics referred to above. Dafters & Anderson (1982) observed the development of tolerance to the tachycardia effect of ethanol in normal social drinkers following repeated alcohol exposure in a novel and distinctive environment. While they did not attempt to demonstrate an antagonistic response to placebo, they concluded that a conditioned heart rate response to environmental stimuli served to counteract the unconditioned alcohol effects. Later, Newlin (1984) gave social drinkers alcohol in a novel environment on four occasions. Subjects then given a placebo drink believed to be alcohol showed a decrease in pulse transit time and finger temperature in comparison to controls. In a similar study, involving 10 exposures to alcohol in a distinctive and novel environment, Shapiro & Nathan (1986) found evidence of tolerance to alcohol on a cognitive vigilance task in same treated subjects compared to controls, but found no effect of placebo on heart rate, finger pulse volume or peripheral temperature.

735

More recently Staiger & White (1988) have reported antagonistic heart rate and temperature responses to alcohol cues following conditioning in two distinctive environments, each for four sessions. They also found that cues associated with the room tended to produce a CAntR whereas cues associated with the drink tended to produce a CAgR, demonstrating both drug-like and drug-opposite responses within the same response systems.

Priming dose studies Priming dose studies differ from studies of cue exposure so far described in that alcohol is actually consumed. Subjects are generally then presented with further alcohol and instructed to resist the desire to drink or are given operant tasks to perform. While these studies provide an interesting insight into operant responses in alcoholics, analysis of psychophysiological responses to presentation of alcohol cues after consumption is complicated by the unconditioned effects of alcohol. This problem is particularly salient if a CAntR to ARCs is predicted: the UCR effectively counteracting any CR effects. Some studies have examined responses both befoi-e and after consumption of a priming dose (Kaplan et al., 1983; 1984; Laberg, 1986). Priming dose studies were conducted early in the history of cue exposure and provided an impetus for research into cue exposure as a treatment method. An increase in desire for alcohol in alcoholics receiving priming doses of alcohol has been consistently observed (Engle & Williams, 1972; Ludwig et al., 1974; Hodgson et al., 1979; Rankin et al, 1979; Stockwell et al, 1982; Kaplan et al, 1983; Kaplan et al, 1984; Laberg, 1986; Laberg & Ellertsen, 1987), particularly when the subjects believed that they had consumed alcohol (Ludwig et al, 1974, Stockwell et al, 1982; Kaplan et al, 1983; Kaplan et al, 1985; Laberg, 1986). Although interpretation of physiological responses to cue exposure following a priming dose of alcohol is complicated as described above, results seem to be generally consistent with those of nonpriming CE studies with increases in SCL (Ludwig et al, 1974; Kaplan et al, 1984; Laberg, 1986; Laberg & Ellertsen, 1987) and HR (Ludwig et al, 1974; Hodgson et al, 1979; Stockwell et al, 1982; Laberg & Ellertsen, 1987). Other observed physiological changes following priming dose CE include increased tremor amplitude (Ludwig et al, 1974) and acceleration (Rankin et al, 1979; Stockwell et al, 1982), decreased temperature (Rankin et al.

736


1979; Laberg, 1986), increased alpha wave EEG activity and respiratory rate (Ludwig et al, 1974). Several investigators employing a priming dose paradigm have been interested in the effects of an initial dose of alcohol on subsequent drinking behaviour in the experimental setting. In Engle & Williams' (1972) study, for example, subjects were instructed that alcohol was obtainable in another part of the hospital and were given detailed instructions on how to obtain it. Only one subject, however, asked for further alcohol, interestingly a subject who had been told that he had been given alcohol but in fact had received placebo. In an operant task on the other hand, Ludwig et al. (1974) found that subjects who were told that they had received alcohol worked more for a further drink as did those subjects who actually received a large priming dose of alcohol compared to controls. Similarly, Hodgson et al. (1979) found that alcoholics given a large priming dose showed increased speed of drinking a further drink in comparison to controls. The same group later replicated these findings (Rankin et al., 1979; Stockwell et al, 1982). Marlatt, Demming & Reid (1973) have also demonstrated that both alcoholics and social drinkers drank significantly more beverage when they believed they were drinking vodka. Conversely, subjects told they were drinking tonic consumed significantly less beverage even when the drink contained vodka. Other studies have not found changes in operant responses following a priming dose of alcohol in a variety of experimental designs (Kaplan et al, 1983; Laberg, 1986). Exposure to alcohol cues in the absence of consumption of a priming dose has not been found to infiuence alcoholics' drinking behaviour in a laboratory setting (Griffiths et al, 1976; Pomerleau et al, 1983; Monti et al, 1987).

Conclusions regarding the evidence for conditioned responses to ARCs in humans There are clearly many methodological problems involved in determining the nature of CRs to ARCs, not least the problem of response specificity. That social drinkers might exhibit CRs to ARCs underlines the importance of defining subjects drinking history and degree of dependence in cue exposure studies. To return to our original questions regarding laboratory studies of CRs to ARCs, first, the evidence suggests that while CRs may develop early in an individual's drinking career increased respond-

ing is associated with higher degrees of alcohol dependence. This could find application in the clinical assessment of alcohol dependence. Further, that CRs are specific to ARCs and not merely an artifact of the experimental setting has been demonstrated in several studies comparing alcohol-related and neutral stimuli. That differential responding to alcohol-related and neutral stimuli has been observed in spite of several factors which reduce the likelihood of this occurrence, adds greater weight to the specificity of ARCs. In terms of the nature of CRs to ARCs, simplistic concepts such as arousal, appetitive responding and craving are inadequate in explaining the complexity of the observed responses. Heart rate, skin conductance and temperature responses are consistent with a CAgR, whereas finger pulse volume and salivatory responses are consistent with a CAntR. Furthermore, both types of physiological response have been observed to occur within individual response systems to different ARCs. Individual physiological measures, however cannot be viewed in isolation of each other. Thus there should be greater concentration on the multivariate patterning of CRs to ARCs. Recent advances in other areas of psychophysiological research, in particular a systems approach (Schwartz, 1982) has much to offer in this respect. For example, it is unclear at present whether an observed hypothermic response is a primary (centrally mediated) CAgR to ARCs or a secondary (peripherally mediated) response to a primary vasoconstrictive CAntR. The monitoring of a greater number of physiological parameters as well as the use of multivariate statistical techniques will contribute greatly to our understanding of these interactions. Another point of interest in relation to the nature of CRs to ARCs is the finding that priming and nonpriming dose studies yield similar physiological results in spite of the probable co-occurrence of tJCRs in the former. Certainly, skin conductance, heart rate and temperature responses may be isodirectional (CAgR) to a UCR, leading to similar findings in the two paradigms. Alternatively, this may be a refiection of the existence of cognitive dissonance in alcoholic compared to social drinkers in both experimental paradigms. We know that following exposure to ARCs, both alcoholics and controls report increased desire for alcohol and have a positive outcome expectancy of the effects of alcohol. In comparison to controls, however, alcoholics may experience cognitive dissonance in that they have also committed themselves to an absti-

Cue exposure treatment in alcohol dependence nence goal (Lipscomb & Nathan, 1980). Festinger (1957) has described cognitive dissonance as an unpleasant subjective experience. As a method of dissonance reduction, an alcoholic may form an external attribution that he is 'craving', which may be seen as a phenomenon over which one has no control. Dissonance has been observed in other experimental paradigms to lead to sympathetic activation with the effects of vasoconstriction (Gerard, 1967), increased skin conductance level (Quanty & Becker, 1974) and tachycardia (Cacioppo & Petty, 1982). Vasoconstriction could as we have suggested account for peripheral hypothermia. While this formulation of the effect of ARCs in alcoholics is speculative, the role of cognitive dissonance in CRs to ARCs deserves further inquiry. Experiments investigating the effects of manipulating expectancy suggest that expectancy of receiving alcohol is an important determinant of CRs to ARCs. Future research should ensure that this factor is taken into account in experimental design. Finally, the evidence linking CRs to ARCs and relapse is currently lacking. The fact that no published study has found a link between CRs to ARCs and relapse has important ramifications for the potential of cue exposure as a treatment approach. A comprehensive prospective study of CRs to ARCs and relapse is much needed. The induction and extinction of CRs in social drinkers has particular appeal as an experimental paradigm in that conditioning history in a novel environment can be more carefully controlled than with clinical populations of alcoholics presenting for treatment. A better understanding of the relationship between CRs and operant drinking responses in this setting could provide a model for relapse in alcoholics.

Cue exposure as treatment That exposure to alcohol related cues, combined with prevention of a drinking response, might infiuence drinking behaviour was suggested several years before any systematic research into the nature of responses to cues, or their effect on operant responses had been conducted. Nevertheless these early optimistic studies, mostly single case designs (e.g. Pickens et al., 1973; Hodgson & Rankin, 1976, 1982), gave considerable impetus to research into the application of CE as a potential treatment for addictive disorders. Using an individualized treatment approach combined with coping skills training Blakey & Baker

737

(1980) exposed six alcoholic subjects to their favourite drink in a variety of settings. This included some subjects who were exposed to alcohol cues in a laboratory setting and some who received exposure in a more typical drinking environment. While the exposure design was not standardized and was combined with other treatments the authors found that 4 out of 6 subjects were abstinent at between 2 and 12 months follow-up. They concluded that the treatment approach held some promise and warranted fuller investigation. The first controlled trial of cue exposure was conducted by Rankin, Hodgson & Stockwell (1983). Based on a theoretical model of treatment for compulsive behaviours (Rachman & Hodgson, 1980) five severely dependent alcoholics, who were hospitalized and had been abstinent for 7-10 days, were given a priming dose of alcohol and were then asked to resist drinking further available alcohol (in vivo CE). Exposure involved holding and smelling a glass of their favourite drink for 3 min on 18 occasions spread over 8 days. A control group of five alcoholics were asked to imagine being exposed to alcohol during the same period (imaginal CE). Compared to the control group the in vivo CE group showed a decrease in desire for alcohol and difficulty in resisting a drink across sessions. It was also found that the time taken to consume a dose of alcohol during a subsequent behaviour test had significantly increased in the in vivo group compared to controls. The authors regarded speed of drinking as a behavioural measure of craving and an analogue of drinking in the natural environment. A follow-up study would be helpful in establishing whether these changes observed in the laboratory extended to drinking behaviour in the natural environment. Further, the physiological measures of HR and tremor, while increased during CE, did not show a decrement across sessions and did not differentiate in vivo CE subjects from controls (Rankin, 1981). The authors prediction of changes in physiological response to CE following treatment was not confirmed by the results although may have been affected by measurement error or the confounding effect of the subjects unconditioned response to the priming dose of alcohol. A more recent study by Laberg & EUertsen (1987) employed a broadly similar design, but included additional control groups. A total of 16 severely dependent alcoholics were assigned to four groups each containing 4 subjects. One group was equivalent to Rankin et al.'s (1983) in vivo CE

738


group in that they received a priming dose of alcohol and were asked to resist further available alcohol ( P + / C + ) . A second group received an alcoholic priming dose, but were exposed to a soft drink ( P + / C — ) . The other two groups received no priming dose followed by exposure to either a soft drink CE ( P - / C - ) or alcohol ( P - / C + ). While the groups were too small to form any firm conclusions, the groups receiving a priming dose of alcohol showed greater initial increases in desire for alcohol, difficulty resisting alcoliol, skin conductance level, non-specific skin conductance responses and heart rate during cue exposure. These changes tended to decrease both within and across sessions regardless of group although the greatest decreases occurred in subjects who had been given an alcoholic priming dose (P + ). The authors concluded that CE is 'an effective treatment method for severely dependent alcoholics'. Closer analysis of the study however reveals that in addition to the absence of follow-up data to support this view, and the fact that responses to alcohol cues were not compared with that of neutral stimuli in a within subject cross over design, it is not clear whether the subjects simply showed habituation to the experimental setting over time. Further, as in the previous study, it is unclear to what extent these laboratory findings might be applicable to the natural setting. Other investigators have drawn attention to the limitations of the demand characteristics of a controlled laboratory setting on alcoholic's drinking behaviour (Lawson et al., 1976). Cue exposure has also received attention as a potential treatment for heroin and cocaine addiction (Childress et al., 1986a; Legarda, 1989; Bradley & Moorey, 1988). In a variety of studies Childress and co-workers (Childress et al., 1984, 1986a, 1986b, 1986c) have consistently demonstrated subjective and physiological responses to drug related stimuli compared to neutral stimuli in both abstinent and methadone maintained heroin addicts. Increases in desire for heroin, skin conductance level and respiratory rate and decreases in skin temperature have been reported. While repeated unreinforced exposure to a variety of drug related stimuli has resulted in significant decreases in subjective desire for drugs, attempts to extinguish physiological responses have been unsuccessful (Childress etal, 1986d). Legarda (1989) recently reported broadly similar results. It is possible that this failure to extinguish physiological responsiveness has contributed to the lack of a

significant influence of the treatment on outcome. McLellan et al. (1986) found that 35 sessions of exposure to drug-related stimuli combined with cognitive behavioural therapy conferred no advantage over cognitive behavioural therapy alone in a group of methadone maintained heroin addicts at 6month follow-up. Both treatments conferred a marginal advantage over drug counselling. Further study is required to examine the influence of duration of exposure and stimulus salience on extinction of physiological responses in heroin addicts. Some commentators have suggested that these factors may have a critical effect on extinction (Legarda, 1989; Bradley & Moorey, 1988). Further, no follow-up study comparable to that of McLellan et al. (1986) has so far been conducted in the alcohol dependence field.

Conclusions In summarizing over 20 years of research and a large number of studies, it is necessary to come to the disappointing conclusion that no published human study has so far demonstrated that CRs to environmental cues represent a causal factor in relapse in either drug or alcohol dependence. Further, no treatment study has demonstrated that cue exposure has any advantage over more conventional and less intensive interventions in terms of relapse. We must therefore be cautious not to allow fashions to dictate treatment trends until an adecjuate demonstration has been made that cue exposure is effective. It would be contrary to the evidence however, to suggest that cue exposure has not progressed since Wikler's innovative theory was proposed. Several findings provide an essential base of knowledge from which future research may proceed. Several studies employing a variety of research designs have found CRs to both drug and alcohol related cues which develop early in a drug using career, the magnitude of which is related to degree of dependence. CRs may be both agonistic and antagonistic within a single response system depending on the nature of the environmental cue and are mediated by expectancy. That respones may be bi-directional makes the case for arguing in favour of only an appetitive mechanism in relapse (Niaura et al., 1988) inapplicable. Further, until physiological measures of responsiveness to ARCs are viewed as part of an interactive system of responses, erroneous interpretations of the directionality of responses can be made. While extinction of CRs to ARCs remains to be

Cue exposure treatment in alcohol dependence convincingly demonstrated in humans, animal studies suggest that in theory at least this should be possible providing adequate information is known regarding conditioning history and stimulus salience. Inevitably, the effects of laboratory extinction procedures will be limited by the problem of generalization to the natural environment. Nevertheless, conditioning theory and the cue exposure method provide a useful paradigm to study relapse. Concepts of arousal, craving and appetitive responding are oversimplistic, ill-defined and inadequate in explaining the observed phenomena in cue exposure. In particular, craving is an ambiguous label which is subject to a variety of interpretations and the influence of attribution. We argue that its use impedes progress in this field. It remains to be determined whether the subjective desire or urge to take a drug is mediated by CAgRs and/or CAntRs. Moreover, it cannot be assumed, as has been previously suggested, that physiological responses to ARCs are covariates of craving. We suspect, however that the use of the term 'craving' will prove resistant to extinction. Throughout the course of this paper we have pointed to a number of potential research questions and directions in which the study of cue exposure should proceed. In summarizing our recommendations we draw attention to the application of concepts which have been developed from both conditioning and cognitive theories of relapse. The application of expectancy theory to the study of conditioned learning in alcohol dependence has led to advances in our understanding of CRs to ARCs. Future research should extend this cross-fertilization of ideas between these two theoretical models. As stated at the beginning of this review, many of the theoretical issues raised have direct relevance to both drug and alcohol dependence.

1. The study of CRs to ARCs in the laboratory remains an open and potentially valuable area of research. While previous work has revealed subjective and psychophysiological responses to environmental cues, little is known about the multivariate patterning of responses and the factors which may influence cue salience. The application of a systems approach to understanding physiological responses to environmental cues should reduce the possibility of making erroneous, interpretations about the directionality of responses. Inevitably such research in the future will require greater technical sophistication than has previously been applied. As we have

739

suggested, concepts such as self-efficacy and cognitive dissonance also deserve further study in terms of their influence on CRs to ARCs. 2. Subjective responses to environmental cues cannot be assumed to be indicative of any particular underlying physiological state or behavioural intention. As we have argued, the term 'craving' should be replaced by more precise descriptions of subjective responses to environmental cues in technical studies. The study of the effects of context and attributions in the cue exposure setting is needed to expand our understanding of physiological and subjective responses beyond that of a simple reflexive model.

3. The relationship between classical and operant conditioning in human drug and alcohol dependence requires to be subjected to a more detailed analysis. The effect of directionality of CRs to drug and alcohol-related cues on operant responses is at present unclear. Study of this interaction in the laboratory may lead to an experimental model of relapse which will allow more accurate predictions to be made about the potential effects of cue exposure treatment than could be inferred from a clinical trial. The induction and extinction of CRs to environmental cues in social drinkers provides a potentially useful experimental paradigm.

4. Cue exposure without consumption of a priming dose provides a valid experimental paradigm for the investigation of CRs to ARCs. Giving alcohol to alcohol dependent individuals in cue exposure studies is controversial (Laberg & Ellertsen, 1987) and complicates the interpretation of physiological responses due to unconditioned alcohol effects. We conclude that it is possible to demonstrate CRs to ARCs without consumption of a priming dose of alcohol and that this represents a valid and useful paradigm for the conduct of cue exposure research.

5. Greater emphasis should be placed upon the study of the basic mechanisms important in relapse. It is perhaps an unfortunate consequence of the urgent need to find an effective treatment for drug and alcohol dependence that so many resources have been directed towards treatment outcome studies in spite of a lack of clear evidence to implicate CRs to environmental cues in relapse. Further, there is

740


evidence to suggest that extinction of CRs to ARCs may not necessarily result in a reduced likelihood of relapse. It could be argued that repeated negative findings from treatment outcome studies, on the way to discovery of the important factors mediating relapse, may have the unwanted effect of turning attention away from the whole field of research endeavour in cue exposure, including promising basic research. Treatment outcome studies should not be abandoned, but instead allied to basic nonclinical research, a point which has relevance to the rest of the addiction research field.

single case reports. Behavioural Psychotherapy, 16, pp. 45-57. CACIOPPO, J. T . & PETTY, R. E. (1982) A biosocial model of attitude change, in: J. T. CACIOPPO & R. E. PETTY

(Eds.) Perspective in Cardiovascular Psychophysiology (New York, Guildford). CARPENTER, J. (1957) The effects of alcoholic beverages on skin conductance: an exploratory study. Quarterly Journal of Studies on Alcohol, 18, pp. 1-18. CHANDLER, B . C , PARSONS, O . A. & VEGA, A. (1975)

Autonomic functioning in alcoholics: a study of heart rate and skin conductance. Journal of Studies on Alcohol, 36, 5, pp. 566-577. CHANEY, E . F . , O'LEARY, M . R. & MARLATT, G . A. (1978)

Skill training with alcoholics. Journal of Consulting and Clinical Psychology, 46, pp. 1092-1104. CHANEY, E . F . , ROSZELL, D . K . & CUMMINGS, C . (1982)

Acknowledgements We gratefully acknowledge the helpful comments and suggestions of Professor Griffith Edwards and Dr Brendan Bradley on an earlier draft of this paper. Many thanks are also due to Mrs Evelyn Langford for her expert secretarial assistance. This work was supported by the Medical Research Council.

References ANNIS H . M . & DAVIS, C . S. (in press) Relapse prevention, in: R. K. HESTER & W. R. MILLER (Eds.) Handbook of

Alcoholism Treatment Approaches (New York, Pergamon Press). AzRiN, N. H. & HAKE, D . F . (1969) Positive conditioned suppression: conditioned suppression using positive reinforcers as the unconditioned stimuli. Journal of the Experimental Analysis of Behaviour, 12, pp. 167-173. BAKER, T . B., MORSE, E . & SHERMAN, J. E. (1987) The

motivation to use drugs: a psychobiological analysis of urges, in: P. E. NATHAN et al. (Eds.) Alcohol and Addictive Behavior, Nebraska Symposium on Motivation, 1986 (Lincoln, University of Nebraska Press). BANDURA, A . (1977) Self-efficacy: toward a unifying theory of behavioural change. Psychological Review, 84, pp. 191-215. BANDURA, A. (1981) Self-referent thought: a developmental analysis of self-efficacy, in: J. H. FLAVELL & L. Ross (Eds.) Social cognitive development: frontiers and possible futures (Cambridge, Cambridge University Press). BANDURA, A. (1982) Self-efficacy mechanism in human agency, American Psychologist, 37, pp. 122-147. BAUM, M . (1969) Extinction of an avoidance response following response prevention: some parametric investigations. Canadian Journal of Psychology, 23, pp. 1-10. BEAZELL, J. M . & IVY, A. C. (1940) The influence of alcohol on the digestive tract: a review. Quarterly Journal of Studies on Alcohol, 1, 45. BLAKEY, R. & BAKER, R. (1980) An exposure approach to alcohol ahuss. Behaviour Research and Therapy, 18, pp. 319-325. BOLLES, R. C. (1972) Reinforcement, expectancy and learning. Psychological Review, 74, pp. 344-409. BRADLEY, B . P. & MOOREY, S. (1988) Extinction of

craving during exposure to drug-related cues: three

Relapse in opiate addicts: a behavioural analysis, .(4(fdictive Behaviours, 7, pp. 291-297. CHILDRESS, A. R . , MCLELLAN, A. T. & O'BRIEN, C . P.

(1984) Measurement and extinction of conditioned withdrawal like responses in opiate-dependent patients, in: Problems of drug dependence, NIDA Research Monograph, 49, (Washington, DC, DHHS). CHILDRESS, A. R., MCLELLAN, A. T. & O'BRIEN, C . P.

(1986a) Abstinent opiate abusers exhibit conditioned craving, conditioned withdrawal and reductions in both through extinction, British Journal of Addiction, 81, 5, pp. 655-660. CHILDRESS, A. R., MCLELLAN, A. t". & O'BRIEN, C . P.

(1986b) Conditioned responses in a methadone population: a comparison of laboratory, clinic and natural setting. Journal of Substance Abuse Treatment, 3, pp. 173-179. CHILDRESS, A. R., MCLELLAN, A. T. & O'BRIEN, C . P.

(1986c) Nature and incidence of conditioned responses in a methadone population: a comparison of laboratory clinic and naturalistic setting, in: Problems of Drug Dependence 1985, NIDA research Monograph, 67, pp. 366-372 (Washington, DC, DHHS). CHILDRESS, A. R., MCLELLAN, A. T., EHRAM, R . &

O'BRIEN, C . P. (1986d) Classically conditioned responses in opioid and cocaine dependence: a role in relapse? Presented at NIDA technical review meeting: Learning Factors in Substance Abuse, 3-4 November 1986. CHILDRESS, A. R., MCLELLAN, A. T., NOTALE, M . &

O'BRIEN, C . P. (1987) Mood states can elicit conditioned withdrawal and craving in opiate abuse patients, in: Problems of Drug Dependence 1986, NIDA Research Monograph, 76, pp. 137-144 (Washington, DC, DHHS). CHOTLOS, J. W . & GOLDSTEIN, G . (1967) Psychophysiolo-

gical responses to meaningful sounds. Journal of Nervous and Mental Disease, 145, pp. 315-325. CONDIOTTE, M. M. & LiCHTENSTEIN, E. (1981) Selfefficacy and relapse in smoking cessation programs. Journal of Consulting and Clinical Psychology, 49, pp. 648-658. CooNBY, N. L., BAKER, L . H . , POMERLEAU, O . F . &

JOSEPHY, B. (1984) Salivation to drinking cues in alcohol abuse: towards the validation of a physiological measure of craving, ^ddictroe Behaviours, 9, pp. 91-94. COONEY, N. L., GiLLESPIE, R. A., BAKER, L. H. & KAPLAN,

Cue exposure treatment in alcohol dependence

741

R. F. (1987) Cognitive changes after alcohol cue GREENBERG, L . A. & CARPENTER, J. A. (1957) The effect exposure. Journal of Consulting and Clinical Psychof alcoholic beverages on skin conductance and emoology, 55(2), pp. 150-155. tional tension. I. Wine, whiskey and alcohol. Quarterly Journal of Studies on Alcohol, 18, pp. 190-204. CooPERSMiTH, S. (1964) Adaptive reactions of alcoholics and nonalcoholics. Quarterly Journal of Studies on GRIFFITHS, R . , BIGELOW, G . & LIEBSON, I. (1976) Alcohol, 25, pp. 262-278. Facilitation of human tobacco self-administration by ethanol: a behavioural analysis. Journal of the ExperiCORTY, E., O'BWEN, C. P. & MANN, S. (1988) Reactivity mental Analysis of Behaviour, 25, pp. 279-292. to alcohol stimuli in alcoholics: is there a role for temptation? Drug and Alcohol Dependence, 21, pp. HEATHER, N . & STALLARD, A. (1989) Does the Marlatt 29-36. model underestimate the importance of conditioned CROWELL, C . R . , HINSON, R . B . & SIEGEL, S. (1981) The craving in the relapse process? in: M. Gossop (Ed.) role of conditional drug responses in tolerance to the Relapse and Addictive Behaviour (London, Routledge). hypothermic effects of ethanol, Psychopharmacology, HODGSON, R. J. & RANKIN, H . J. (1976) Modification of 73, pp. 51-54. excessive drinking by cue exposure. Behaviour Research DAFTERS, R. & ANDERSON, G . (1982) Conditioned tolerand Therapy, 14, pp. 305-307. ance to the tachycardia effect of ethanol in humans, HODGSON, R. J. & RANKIN, H . J. (1982) Cue exposure and Psychopharmacology, 78, p. 365. relapse prevention, in: P. NATHAN & W. HAY (Eds.) Case studies in the Behavioural Modification of AlcoholDi CLEMENTE, C . C . (1981) Self-efficacy and smoking ism (New York, Plenum Press). cessation maintenance. Cognitive Therapy and Research, 5, pp. 175-187. HODGSON, R . , RANKIN, H . & STOCKWELL, T . (1979) DocTER, R. F. & BERNAL, M . E . (1964) Immediate and Alcohol dependence and the priming effect. Behaviour prolonged psychophysioiogical effects of sustained alcoResearch and Therapy, 17, pp. 379-387. hol intake in alcoholics. Quarterly Journal of Studies on HOLMBERG, G. & MARTENS, S. (1955) ElectoencephaloAlcohol, 25, p. 438. graphic changes in man correlated with blood alcohol concentration and some other conditions following DocTER, R. F. & PERKINS, R. B . (1960) The effects of standardized ingestion of alcohol. Quarterly Journal of ethyl alcohol on automomic and muscular responses in Studies on Alcohol, 16, p. 411. humans. Quarterly Journal of Studies on Alcohol, 22, p. 374. HiKSON, R. E. & SIEGEL, S. (1983) Anticipatory hyperexDocTER, R., NAITOH, P . & SMITH, J. C. (1966) Electroencitabiiity and tolerance to the narcotizing effect of cephaiographic changes and vigilance behaviour during morphine in the rat. Behavioural Neuroscience, 97, pp. experimentally induced intoxication with alcoholic sub759-767. jects. Psychosomatic Medicine, 28, pp. 605-615. KAPLAN, R. F . , MEYER, R . E . & STOEBEL, C . F . (1983) DoLiNSKY, Z. S., MORSE, D . E . , KAPLAN, R. F . , MEYER, Alcohol dependence and responsivity to an ethanol R. E. et al. (1987) Neuroendocrine, psychophysiological stimulus as predictors of alcohol consumption, British and subjective reactivity to an alcohol placebo in male Journal of Addiction, 78, pp. 259-267. alcoholic patients. Alcoholism: Clinical and ExperimenKAPLAN, R . F . , MEYER, R . E . & VIRGILIO, L . M . (1984) tal Research, 11(3), pp. 296-300. Physiological reactivity to alcohol cues and the awareness of an alcohol effect in a double blind placebo DONOVAN, D . M . & CHANEY, E . F . (1985) Alcoholic design, British Journal of Addiction, 79, pp. 439-442. relapse prevention and intervention: models and methods, in: G. A. MARLATT & J. R. GORDON (Eds.)

Relapse Prevention: Maintenance Strategies in the Treatment of Addictive Behaviors (New York, Guildford). EiKELBOOM, R. & STEWART, J. (1982) Conditioning of drug induced physiological responses. Psychological Review, 89, pp. 507-528. ELLIOT, R. (1974) The motivational significance of heart

KAPLAN, R . F . , COONEY, N . L . , BAKER, L . H . , GILLESPIE, A., MEYER, R . E . & POMERLEAU, O . F . (1985)

Reactivity to alcohol-related cues: physiological and subjective responses in alcoholics and non-problem drinkers. Journal of Studies on Alcohol, 46, 4, pp. 267-272. KENNEDY, D . A. (1971) Pupilometrics as an aid in the rate, in: P. A. OBRIST, A. H. BLACK, J. BRENER & L. V. assessment of motivation, impact of treatment and DICARA (Eds.) Cardiovascular psychophysiology: curprognosis of chronic alcoliolics. Unpublished doctoral rent issues in response mechanisms, biofeedback and dissertation. University of Utah. methodology (Chicago, Aldine). KiSSIN, B., SCHENKER, V. & SCHENKER, A. C. (1959) The ENGLE, K . B . & WILLIAMS, T . K . (1972) Effect of an acute effects of ethyl alcohol and chloropromazine on ounce of vodka on alcoholic's desire for alcohol. certain physiological functions in alcoholics. Quarterly Quarterly Journal of Studies on Alcohol, 33, pp. Journal of Studies on Alcohol, 20, p. 480. 1099-1105. KRANK, M . D . , HINSON, R . E . & SIEGEL, S. (1981) ERIKSEN, L . & GOTESTAM, K. G. (1984) Conditioned Conditional hyperalgesia is elicited by environmental abstinence in alcoholics: a controlled experiment. Intersignals of morphine. Behavioural and Neural Biology, national Journal of the Addictions, 19(3), pp. 287-294. 32, pp. 148-157. FESTINGER, L . (1957) A Theory of Cognitive Dissonance LABERG, J. C. (1986) Alcohol and expectancy: subjec(Stanford, Stanford tJniversity Press). tive, psychophysiological and behavioural responses to FREEMAN, H . L . & KENDRICK, D . C . (1960) A case of cat alcohol stimuli in severely, moderately and non-depenphobia, British Medical Journal, ii, pp. 497-502. dent drinkers, British Journal of Addiction, 81, pp. GERARD, H . B. (1967) Choice difficulty, dissonance, and 797-808. the decision sequence. Journal of Personality, 35, pp. LABERG, J. C. & ELLERTSEN, B. (1987) Psychophysiologi91-108. cal indicators of craving in alcoholics: effects of cue

742


exposure, British Journal of Addiction, 82, pp. 1341-1348. LACEY, J. I. (1959) Psychophysiological approaches to the evaluation of psychotherapeutic processes and outcome, in: E. RUBENSTEIN & M. B. PARLOFF (Eds.) Research in

Psychotherapy (Wasington, DC, American Psychological Association). LACEY, J. I. (1967) Somatic response patterning and stress: some revisions of activation theory, in: M. H. APPLEY & -R. TRUMBULL (Eds.) Psychological Stress, pp. 14-42 (New York, Appleton-Century-Crofts). LAWSON, D . M . , WILSON, G . T . , BRIDDELL, D . W . & IVES,

C. C. (1976) Assessment and modification of alcoholic's drinking beiiaviour in controlled laboratory settings: a cautionary note. Addictive Behaviours, 1, pp. 299-303. LE,

A. D . , POULOS, C . X. & CAPPELL, H . (1979)

Conditioned tolerance to the hypothermic effect of ethyl alcohol, Science, 206, pp. 1109-1110. LEGARDA, J. (1989) Subjective and psychophysiological effects of exposure to drug-related cues in detoxified heroin addicts, ch. 5, PhD. Thesis, University of London. LEVENSON, R . W . , SHER, K . J., CROSSMAN, L . M . ,

NEWMAN, J. & NEWLIN, D . B. (1980) Alcohol and stress response dampening: pharmacological effects, expectancy and tension reduction. Journal of Abnormal Psychology, 89, pp. 528-538. LiPSCOMB, T. R. & NATHAN, P. E. (1980) Effect of family history of alcoholism, drinking pattern, and tolerance on blood alcohol level discrimination. Archives of General Psychiatry, 37, pp. 571-576. LiTT, M. D., CoONEY, N. L., KADDEN, R. M . & GAUPP, L .

(1989) Reactivity to alcohol cues and induced mood in alcoholics (in preparation). LOLORDO, V. M. (1971) Facilitation of food reinforced responding by a signal for response independent food. Journal of the Experimental Analysis of Behaviour, 15, pp. 49-55. LOLORDO, V. M., MACMILLAN, J. A. & RiLEY, A. L. (1974) The effects upon food reinforced key pecking and treadle pressing of auditory and visual signals for response independent food. Learning and Motivation, 5, pp. 24-41. LuDWIG, A. M. & WlKLER, A. (1974) Craving and relapse to drinking. Quarterly Journal of Studies on Alcohol, 35, pp. 108-130. LuDwiG, A. M., WiKLER, A. & STARK, L . H . (1974) The first drink: psychobiological aspects of craving. Archives of General Psychiatry, 30, pp. 539-547. MACKINTOSH, N . J. (1983) Conditioning and Associative Learning (Oxford, Clarendon Press). MALMO, R. B . (1959) Activation: a neuropsychological dimension. Psychological Review, 66, pp. 367-386.

Alcoholism: Nezv Directions in Behavioural Research and Treatment (New York, Plenum). MARLATT, G . A. (1985a) Situational determinants of relapse and skill-training interventions, in: G. A. MARLATT & J. R. GORDON (Eds.) Relapse Prevention: Maintenance Strategies in the Treatment of Addictive Behaviours (New York, NY, Guildford). MARLATT, G . A. (1985b) Cognitive processes in alcohol use: expectancy and the balanced placebo design, in: N. K. MELLO (Ed.) Advances in Substance Abuse, 1 (Greenwich, Conn., JAI). MARLATT, G . A. (1988) Cue exposure and relapse prevention in the treatment of addictive behaviours. Discussion paper presented at the Cue Exposure in Addictive Behaviours Symposium, Behaviour Therapy World Congress, Edinburgh, September, 1988. MARLATT, G . A. & ROSENHOW, D . R. (1980) Cognitive

processes in alcohol use: expectancy and the balanced placebo design, in: N. K. MELLO (Ed.) Advances in Substance Abuse, 1 (Greenwich, Conn., JAI). MARLATT, G . A. & ROSENHOW, D . R. (1981) The think-

drink effect. Psychology Today, 15, pp. 60-93. MARLATT, G . A., DEMMING, B . & RIED, J. B. (1973) Loss of

control drinking in alcoholics: an experimental analogue. Journal of Abnormal Psychology, 81, pp. 223-241. McAuLlFFE, W. E. (1982) A test of Wikler's theory of relapse due to conditioned withdrawal sickness. International Journal of the Addictions, 17, pp. 19-23. MCLELLAN, A. T . , CHILDRESS, A. R., EHRMAN, R . W . &

O'BRIEN, C . P. (1986) Extinguishing conditioned responses during treatment for opiate dependence: turning laboratory findings into clinical procedures. Journal of Substance Abuse Treatment, 3, pp. 33-40. MELCHIOR, C . L . & TABAKOFF, B . (1981) Conditioned

tolerance to the hypothermic and hypnotic effects of ethanol in mice. Federation Proceedings of the Federation of American Social Experimental Biology. MELCHIOR, C . L . & TABAKOFF, B . (1982) Environment

dependent tolerance to the lethal effects of ethanol. Alcohol: Clinical & Experimental Research, 6, p. 306. MELCHIOR, C. L . & TABAKOFF, B. (1984) A conditioning

model of alcohol tolerance, in: M. GALANTER (Ed.) Recent Developments in Alcoholism, 2, pp. 5-16. MENDELSON, J. H . & LADOU, J. (1964) Experimentally

induced chronic intoxication and withdrawal in alcoholics. Part 1. Background and experimental design. Quarterly Journal of Studies on Alcohol, Suppl. No. 2, p. 1. MONTI, P. M., BINKOFF, J. A., ABRAMS, D . B., ZWICK,

W. R. et al. (1987) Reactivity of alcoholics and nonalcoholics to drinking cues. Journal of Abnormal Psychology, 96, pp. 122-126. NEWLIN, D . B . (1984) The conditioned compensatory MANSFIELD, J. G., CUNNINGHAM, C. L. (1980) Conditionresponse to alcohol placebo in humans, Psychophysioling and extinction of tolerance to the hypothermic effect ogy, 21, p. 590. of ethanol in lats. Journal of Comparative Physiological NEWLIN, D . B. (1985a) The antagonistic placebo response Psychology, 94, pp. 962-969. to alcohol cues. Alcoholism: Clinical and Experimental MARKS, L M . , HODGSON, R. & RACHMAN, S. (1975) Research, 9(5), pp. 411-416. Treatment of chronic obsessive-compulsive neurosis by in vivo cue exposure: a two year follow-up and issues in NEWLIN, D . B . (1985b) Offspring of alcoholics have enhanced antagonistic placebo response. Journal of treatment, British Journal of Psychiatry, 127, pp. Studies on Alcohol, 46(6), pp. 490-494. 349-364. NEWLIN, D.B. (1986) Conditioned compensatory response MARLATT, G . A . (1978) Craving for alcohol, loss of to alcohol placebo in humans, Psychopharmacology, 88, control and relapse: a cognitive-behavioural analysis, in: pp. 247-251. P. E. NATHAN, G . A. MARLATT & T. LABERG (Eds.)

Cue exposure treatment in alcohol dependence

743

ROTH, G . M . & SHEARD, C. (1947) Effect of smoking on the vasodilation produced by the oral administration of 95 per cent ethyl alcohol or a substantial meal, American Heart Journal, 33, p. 654. SCHACHTER, S. & SiNGER, J. E. (1962) Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69, pp. 379-394. SCHWARTZ, B. & GAMZU, E. (1977) Pavlovian control of operant behaviour: an analysis of autoshaping and its implications for operant conditioning, in: W. K. HONIG & J. F. R. STADDON (Eds) Handbook of Operant Behavior (New Jersey, Prentice Hall). SCHWARTZ, G . E . (1982) Cardiovascular psychophysiology: a systems perspective, in: J. CACIOPPO & R. E. PETTY (Eds) Perspectives in Cardiovascular PsychophyPOMERLEAU, O. F., FERTIG, J., BAKER, L . & COONEY, N. siology (New York, Guildford). (1983) Reactivity to alcohol cues in alcoholics and nonalcoholics: implications for a stimulus control analysis of SHAPIRO, A. P., NATHAN, P. E. (1986) Human tolerance to drinking. Addictive Behaviours, 8, pp. 1-10. alcohol: the role of Pavlovian conditioning processes, Psychopharmacology, 88, pp. 90-95. QUANTY, M. B. & BECKER, L . A. (1974) Physiological indices of dissonance arousal and reduction in a stressful SlEGEL, S. (1975) Evidence from rats that morphine situation (Columbia, University of Missouri). tolerance is a learned response. Journal of Comparative Physiological Psychology, 89, pp. 498-506. RACHMAN, S. & HODGSON, R. (1980) Obsessions and Compulsions (New Jersey, Prentice Hall, EnglewoodSiEGEL, S. (1978) Tolerance to the hyperthermic effect of Cliffs). morphine in the rat is a learned response. Journal of RAFFA, R . B., PORRBCA, F . , COWAN, A. & TALLARIDA, R. J. Comparative Physiological Psychology, 92, pp. (1982) Evidence for the role of conditioning in the 1137-1149. development of tolerance to morphine-induced inhibiSlEGEL, S. (1988) Drug anticipation and drug tolerance, tion of gastrointestinal motility in rats. Federation in: M. LADER (Ed.) Psychopharmacology of Addiction Proceedings, 41, pp. 1317. (New York, Oxford University Press). SlEGEL, S., HiNSON, R. E. & KRANK, M . D . (1979) RANKIN, H . (1981) The behavioural assessment and Modulation of tolerance to the lethal effect of morphine treatment of alcoholism, PhD thesis. University of by extinction. Behavioural and Neural Biology, 25, pp. London, pp. 236-307. 257-262. RANKIN, H . , HODGSON, R. & STOCKWELL, T . (1979) The concept of craving and its measurement. Behaviour SlEGEL, S., SHERMAN, J. E. & MITCHELL, D . (1980) Research and Therapy, 17, pp. 389-396. Extinction of morphine analgesic tolerance. Learning and Motivation, 11, pp. 289-301. RANKIN, H . , HODGSON, R . & STOCKWELL, T . (1983) Cue exposure and response prevention with alcoholics: a SOLOMON, R. L . (1977) An opponent-process theory of controlled trial. Behaviour Research and Therapy, acquired motivation: IV. The affective dynamics of 21(4), pp. 435-446. addiction, in: J. MASER & M. E. P. SELIGMAN (Eds.) Psychopathology: Experimental Models (San Francisco, RANKIN, H . , STOCKWELL, T . & HODGSON, R. (1982) Cues Freeman). for drinking and degrees of alcohol dependence, British Journal of Addiction, 77, 287-296. SOLOMON, R. L . & CORBITT, J. D. (1974) An opponentprocess theory of motivation: I, temporal dynamics of REISS, S. (1980) Pavlovian conditioning and human fear: an affect. Psychological Review, 81, pp. 119-145. expectancy model. Behaviour Therapy, 11, pp. 380-396. RESCORLA, R. A. & LoLoRDO, V. M. (1965) Inhibition of STAIGER, P . k . & WHITE, J. M. (1988) Conditioned alcohol-like and alcohol-opposite responses in humans, avoidance behaviour. Journal of Comparative PhysioPsychopharmacology, 95, pp. 87-91. logical Psychology, 59, pp. 406-412.

NiAURA, R. S., ROSENHOW, D. J., BiNKOFF, J. A., MONTI, P. M. et al. (1988) Relevance of cue reactivity to understanding alcohol and smoking relapse, Journal of Abnormal Psychology, 97, pp. 133-152. NisBETT, R. E. & WILSON, T . D . (1977) Telling more than we know: verbal reports on mental processes. Psychological Review, 84, pp. 231-259. PiCKENS, R., BiGELOW, F. & GRIFFITHS, R. (1973) An experimental approach to treating chronic alcoholism: a case study and one year follow-up. Behaviour Research and Therapy, 11, pp. 321-325. PoMERLEAU, O. F. (1981) Underlying mechanisms in substance abuse: examples from research on smoking. Addictive Behaviours, 6, pp. 187-196.

RESCORLA, R. A. & SOLOMAN, R. L . (1967) Two-process

STEWART, J., DE Wrr, H. & EIKELBOOM, R. (1984) Role of

learning theory: relationships between Pavlovian conditioning and instrumental learning. Psychological Review, 74, pp. 151-182.

unconditioned and conditioned drug effects in the self administration of opiates and stimulants. Psychological Review, 91, pp. 251-268.

RESCORLA, R. A. & WAGNER, A. R. A. (1972) A theory of

STOCKWELL, T . R . , HODGSON, R. J., RANKIN, H . J. &

TAYLOR, C. (1982) Alcohol dependence, beliefs and the Pavlovian conditioning: variations in the effectiveness priming effect. Behaviour Research and Therapy, 20, of reinforcement and non-reinforcement, in: A. H. pp. 513-522. BLACK & W. F. PROKASY (Eds.) Classical Conditioning II (New York, Appleton-Century-Crofts). WlKLER, A. (1965) Conditioning factors in opiate addiction and relapse, in: D. I. WIHER & G. KASSEBAUM RiST, F. & WATZL, H . (1983) Self assessment of relapse (Eds.) Narcotics (New York, McGraw-Hill). risk and assertiveness in relation to treatment outcome of female alcoholics. Addictive Behaviours, 8, pp. WILLIAMS, D . R . (1965) Classical conditioning and 121-127. incentive motivation, in: W. F. PROKASY (Ed.) Classical Conditioning (New York, Appleton Century Crofts). RITCHIE, J. M. (1965) The aliphatic alcohols, in: L. S. B. GOODMAN & A. GILMAN (Eds.) The Pharmacological

WISE, R. A. & BOZARTH, M . A. (1987) A psychomotor

Basis of Therapeutics, pp. 143-158 (New York, MacMillan).

stimulant theory of addiction. Psychological Review, 4, pp. 469-492.

Conditioned learning in alcohol dependence - Wiley Online Library

Conditioned learning in alcohol dependence - Wiley Online Library

Suggest Documents

Alcohol Abuse/Dependence in Motor Vehicle ... - Wiley Online Library

Alcohol - Wiley Online Library

Familial Risk for Alcohol Dependence and Brain ... - Wiley Online Library

Windsock memory COnditioned RAM - Wiley Online Library

Tasting Alcohol - Wiley Online Library

alcohol oxidase - Wiley Online Library

Windsock memory COnditioned RAM - Wiley Online Library

Acceleration dependence and taskspecific ... - Wiley Online Library

Decayrate dependence of anhysteretic ... - Wiley Online Library

Conspecific negative density dependence ... - Wiley Online Library

Dependence of depurination of ... - Wiley Online Library

Doseresponse relationship between alcohol ... - Wiley Online Library

Monitoring prenatal alcohol exposure - Wiley Online Library

Energy dependence and angular dependence ... - Wiley Online Library

Alcohol or tobacco research versus alcohol and ... - Wiley Online Library

Learning progressions in context - Wiley Online Library

Individual differences in learning ... - Wiley Online Library

Flipped Learning in - Wiley Online Library

Elevational variation in density dependence in a ... - Wiley Online Library

Alcohol consumption in homicide victims in the ... - Wiley Online Library

Interleukin 1 beta enhances conditioned fear ... - Wiley Online Library

Mesenchymal stem cells conditioned with ... - Wiley Online Library

The Conditioned Media of Adipose Stromal ... - Wiley Online Library

Quantifying predator dependence in the ... - Wiley Online Library