Another study investigating the effects of aversive conditioning on alcohol consumption was conducted by Caddy and Lovibond, at the University of New South Wales. Specifically, their study involved the placement of sixty alcoholics into one of the following four conditions:
1). aversion plus self-regulation - which included blood alcohol concentration feedback training, discriminated aversive conditioning, training in self-regulation, education, and psychotherapy
2). self-regulation - which was the same as the "aversion plus self-regulation" treatment condition, in that it also included blood alcohol concentration feedback training, discriminated aversive conditioning, training in self-regulation, education, and psychotherapy except that, in this latter condition, no shock was used.
3). aversion - which was also the same as the "aversion plus self-regulation" treatment condition, except that, in the aversion condition, no emphasis was given to the self-regulatory processes.

       For purposes of Caddy and Lovibond's study, the term discrimination training referred to patients being "taught to discriminate their own blood alcohol concentrations within the limits of 0-0.08% (expressed as g/100 ml)." (Caddy and Lovibond, 1976). This was achieved by providing each patient with a scale specifying the visible behavioral symptoms which might be expected to accompany specific blood alcohol concentration levels. Each patient was then instructed to ingest a variety of beverages containing alcohol, and, following this, to introspectively report on his own current blood alcohol concentration, using criteria such as his own subjective experience, current behavior, and the amount of alcohol consumed, to come up with an estimate of his own current blood alcohol concentration.
       The accuracy of each estimate was then verified using a breath analysis procedure conducted twenty minutes after the patient consumed each drink. The results were then reported back to the patient, allowing them  to verify the accuracy of their own previous prediction.
       Essentially, the idea was to get patients to associate their subjective impressions of their current state with objective measures of both how much alcohol they had consumed and descriptions of how consumption of such an amount of alcohol might be expected to impact an individual, with the end goal of such training being that of empowering individuals to make better decisions about their alcohol consumption by giving them the tools to estimate how consuming a particular amount of alcohol will impact their emotions and behavior.
      Interestingly, according to the reports of Caddy (1976), patients were able to come up with remarkably accurate predictions (estimation errors of less than 0.01%) within the very short time of less than three training sessions! Furthermore, this remarkable accuracy of estimation reportedly persisted both throughout the duration of the training and during a subsequent follow-up period.
       In Caddy and Lovibond's (1976) study, education was actually a component of the above-described discrimination-training procedure, and referred to the provision to patients of detailed informational materials describing the medical and social effects of alcohol consumption, as well as an explanation of what exactly was meant by the term "blood-alcohol concentration" and the emotional and behavioral effects which might be expected as a function of increases in its point value.
        A significant aspect of the treatment procedure in Caddy and Lovibond's (1976) study was that of discriminated aversive conditioning. This aspect of the treatment was initiated following the end of the second discrimination training session, when the patient's blood alcohol concentration was equal to 0.065% (as measured by a breath analysis test). At this time, shock electrodes were placed near the patients larynx. Following their attachment, the patient was offered the option of continuing to drink. If and only if the patient did indeed make the decision to continue, a "highly aversive" (Caddy and Lovibond, 1976, p. 224) shock of "up to 8.2mA x 6 sec" (Caddy and Lovibond, 1976, p. 224), was administered, the idea being that conditioned aversion of a type inspired by this operant conditioning procedure of sorts would be inspired, as the patient learned an association between continuing to drink beyond a pre-prescribed maximum limit and an aversive stimulus (such as a shock). To make the learning effects less subject to extinction, a variable ratio (see previous post) schedule of reinforcement was used (although in this case, the unconditional stimulus was that of positive punishment, as provided through the administration of shock, rather than any type of positive reinforcement). Use of such a schedule also added to the study's external validity; in a real life situation, an individual would not necessarily be able to define the specific amount of alcohol, above a legally-prescribed limit, that they will have to have consumed before the onset of various negative consequences (i.e., getting arrested for drunk driving is also "enforced" on a variable ratio schedule; while everyone is well aware that doing so is both illegal and immoral, it is impossible for an individual who, for whatever reason, chooses to engage in this behavior to predict the specific instance of the occurrence of the behavior following which they will be arrested. Likewise, it would be exceedingly difficult or even impossible for a recovering alcoholic to predict the specific instance of alcohol consumption above a previously specified limit which will lead them to relapse into their previous pattern of uncontrollable and excessive alcohol consumption- but they can nonetheless be sure that continued alcohol consumption above such a limit will indeed eventually lead  to relapse). Thus, inclusion of this specific type of reinforcement schedule allows the study to more closely simulate real-life conditions, allowing individuals to obtain within the lab skills which will remain useful to them even once they leave it.
       An criticism of this type of conditioning procedure, however, can be based on the fact that evidence has shown that the "extent to which a CS is relevant to or belongs with a US" (Domjan 2009 p. 107) can have a significant impact on the rate at which conditioned responding is acquired. A good example illustrating this is a study by Garcia and Koelling (1966) which compared the rates of learning of conditioned aversion to either a taste conditioned stimulus or an audiovisual conditioned stimulus among a group of laboratory rats involved in a lick suppression experiment. This experiment consisted of a group of mildly water-deprived rats drinking from a drinking tube, the water in which had been flavored with either a salty or a sweet taste. As the rats licked the tube, (the fact that they were mildly water-deprived particularly motivated them to do so), they activated an audiovisual stimulus, which in this case consisted of a click and a flash of light, which they encountered at the same time as they did the flavor of the water they were drinking (either salty or sweet). These taste and audiovisual conditioned stimuli were then followed by an unconditioned stimulus, which in this case consisted of either a brief cutaneous shock (this shock was administered to their paws through the grid floor), or of the rats being made nauseous. Following this aversion conditioning procedure, the investigators tested the rats to see what type of aversions had been acquired to each of the two conditioned stimuli (the taste of the water and the audiovisual stimuli) separately. This was done by presenting each of the conditioned stimuli separately and measuring the rat's response to them. Specifically, the rats' response to the conditioned stimulus of taste was measured by presenting them with water that was flavored in the same way as before (i.e., either salty or sweet), but the licking of which no longer resulted in the activation of the audiovisual stimulus (which had been previously been paired with taste). Similarly, during tests of the audiovisual cues, the water was now unflavored, but licking it did activate the previously-seen audiovisual cue. The degree of aversion to a particular conditioned stimulus (either taste or the audiovisual stimulus) was measured through the degree of conditioned lick suppression (the degree to which the amount which the rats were drinking decreased as a function of the presentation of one or the other of the conditioned stimuli). Crucially, the results of the study indicated both that "It seems that given reinforcers are not equally effective for all classes of discriminable stimuli. The cues, which the animal selects from the welter of stimuli in the learning situation, appear to be related to the consequences of the subsequent reinforcer" (Garcia and Koelling, 1966, p. 124). In other words, the results of Garcia and Koelling's experiment indicated that, in order to be effective  (and to compete effectively with the multitude of other stimuli present in an experimental situation), a Conditioned Stimulus has to be meaningfully related to the Unconditioned Stimulus that is employed. For example, as Garcia and Koelling also stated in their discussion, natural selection likely predisposed animals to be particularly responsive to cues (such as conditioned stimuli) relating to taste and smell being paired with internal discomfort (such as the experimentally-induced nausea used in this experiment). Likewise, natural selection may also have predisposed animals to be particularly sensitive to cutaneous shock being paired with audiovisual stimuli. As Domjan (2009) states, this makes sense from an evolutionary standpoint, because "In their natural environment, rats are likely to get sick after eating a poisonous food. In contrast, they are likely to encounter peripheral pain after being chased and bitten by a predator that they can hear and see" (Domjan, 2009, p. 107). Given all of this, it seems somewhat strange that Caddy and Lovibond's (1976) study employed a pairing of a gustatory stimulus (the taste and smell of alcohol) with that of cutaneous shock specifically- when Garcia and Koelling's experiment demonstrates that the effectiveness of taste as a conditioned stimulus is highest when it is paired with the unconditioned stimulus of illness, and that the effectiveness of audiovisual cues as stimuli is greatest when they are paired with peripheral pain (such as that induced by electrodes being placed on the skin, as in Caddy and Lovibond's experiment). 
       Another important criticism which can be leveled at Caddy and Lovibond's study is that of its categorization of "moderate drinking" as an acceptable outcome of treatment.  Specifically, while Caddy and Lovibond's study claimed that one of the purposes of the treatment administered to each participating patient was to get them to "... recognize the degree to which his excessive drinking was under stimulus control" (Caddy and Lovibond, 1976, p. 224), and, subsequently, to get each patient to "... change his own environment in order to reduce drinking," (Caddy and Lovibond, 1976, p. 224), the behaviors it encourages in practice seem to detract from these outcomes. For instance, Caddy and Lovibond's study cites, as an example of appropriate treatment, a case in which "... a patient was taught ways of avoiding the pressure of the tradition in Australia which requires each individual in turn to buy drinks for the whole group. One method involved the subject in buying the drinks early, drinking slowly, and dropping out of most subsequent rounds," (Caddy and Lovibond, p. 224). Common sense seems to indicate that a recovering alcoholic, just fresh out of treatment, should not be in situations where stimuli which might cause him to incite him to resume his previous drinking behavior in the first place; if, as Caddy and Lovibond's study asserts, drinking is "under stimulus control," (Caddy and Lovibond, 1976, p. 224), why allow a patient to expose himself to stimuli which might encourage a return to the previous drinking behavior in the first place, instead of condoning such dangerous exposures? All of this is particularly surprising when it is noted that Caddy and Lovibond's study itself mentions that "The maintenance of behavioral changes induced in the laboratory or clinic probably depends largely upon the extent to which environmental contingencies support the new behavior," (Caddy and Lovibond, 1976, p.225). While Caddy and Lovibond's study also defined the treatment of a patient as "completely successful" if "(a) he were drinking in a controlled fashion and was exceeding a blood alcohol concentration of 0.07% less than once a month (b) he and his family collateral were both extremely satisfied with his alcohol related behavior and (c) the therapist and follow-up staff were satisfied that the patient was drinking in an essentially normal and moderate fashion" (Caddy and Lovibond, 1976, p. 227)-it and thus, the criteria upon which the investigators based their judgment of treatment success was at least internally consistent- that does not necessarily make it valid.
       As Domjan (2009) states, convergent evidence already suggests that "the administration of a drug constitutes a conditioning trial in which cues related to drug administration are paired with the pharmacological effects of the drug," (Domjan, 2009, p. 115). This in itself leads to some interesting questions regarding how these conditioned cues then come into play in drug addictions and other drug-related problems.
      A study by Ehrman, Robbins, Childress, and O'Brien (1992) sheds some light on this issue. In that study, a group of men, all of whom had a history of abusing cocaine (through such means as free-basing as well as smoking), but who at the same time had no history of heroin use were compared with a control group of men, none of whom had used cocaine or heroin. The participants were exposed to three different types of test conditions, specifically, those of being exposed to cocaine-related cues (such as listening to an audio tape of people talking about their cocaine usage, watching a video of cocaine-related activities, and role-playing the motions of using cocaine), being exposed to similar types of cues, but as related to heroin, rather than cocaine, and being exposed to unrelated, control stimuli. Responses to the three types of stimuli were measured and recorded using both physiological and self-report measures. Interestingly, both the physiological and self-report measures provided data that indicating that the cocaine-related stimuli elicited conditioned responses- including increases in heart rate and symptoms of cocaine withdrawal- but, interestingly, these effects were to be found among only the participants who had previously been users of cocaine. Neither the participants who had previously been habitual users of heroin, nor the participants who had no previous exposure to those drugs experienced those same changes. The fact that these changes were both pronounced and limited to only members of the group who had previously been exposed to cocaine would indicate that the increased heart rate and other symptoms which members of the group were experiencing were indeed coming as a consequence of the formation of conditioned associations formed over the course of their drug use. Furthermore, the fact that the participants who had previous experience with cocaine experienced (as the study mentioned) strong cravings for cocaine would indicate that individuals who have previously struggled with a particular drug should stay away from both that drug and cues related to it (to prevent the onset of strong drug-related cravings and withdrawal symptoms)- in contrast to how, in Caddy and Lovibond's study, a former addict could still be considered "successful" in overcoming his addiction, despite occasionally still engaging in drinking alcohol!
      Furthermore, as Domjan (2009) goes on to elaborate on this issue, "For drug addicts, the beginnings of a buzz or high are typically followed by substantial additional drug intake and a more intense high. Therefore, the early weak drug effect can serve as a CS signaling additional drug intake and can elicit drug cravings and other drug conditioned reactions. In this case, the CS is an internal or introceptive cue. The additional craving elicited by a small dose of the drug makes it difficult for addicts to use drugs in moderation. This is why abstinence is their best hope for controlling cravings," (Domjan, 2009, p. 116).
       Thus, it would seem that in oder to achieve the stated goal of Caddy and Lovibond's therapeutic program of getting recovering alcoholics to realize that their addiction really is stimulus-dependent, and that, as such, it can largely be mitigated through the avoidance of particular stimuli, it is indeed best, as Domjan mentions, to encourage total abstinence from both drug related cues and both large and small doses of the drugs themselves, in order to facilitate the recovery of former alcoholics.
       When this is combined with the conditioning model of drug tolerance described by Domjan (2009), in which, as a habitual user of a particular drug continues taking it on multiple successive occasions, the initial response or "high" that he obtains (this "high" is brought on by the stimulus-dependent "a process" of the opponent process model discussed in a previous post), is, over time, compensated for to an ever-larger extent by the compensatory "b-process" of the same opponent process theory of motivation model (the "b-process" comes online as a response of the body to being thrown violently out of homeostatic equilibrium by the effects of the "a process;" over time, cues related to the forthcoming administration of a particular drug become conditioned, and, thus, the "b process" is able to start earlier in response to the onset of those cues and before the body has been thrown too greatly out of equilibrium by the "a-process"-and the b-process also remains active for longer periods of time once it has been activated, often outlasting the "a process")- the user has to take ever-larger doses of the drug in order to obtain the same "effect" he had originally gotten (where "originally" would refer to such a time as before the "b process" became more readily activated by certain conditioned cue related to forthcoming drug administration, and before it would come on earlier and remain active for longer)- a phenomenon referred to as the buildup of tolerance (Domjan, 2009, p. 117).. An interesting feature of this phenomenon is that, because the drug user has built up so much tolerance in relation to particular drugs (a tolerance which is largely dependent on the consistent administration of the drugs in the same types of environments and under the same types of circumstances, with these cues serving to activate the compensatory "b process" which serves to help restore homeostasis), a habitual user is often able to take a much larger dose of a particular drug than someone who is naive to the drug. And in fact, habitual users of particular drugs often do take such much-larger doses, and indeed, they are protected from any particularly dangerous effects of such an increased dose because they do so in the same types of environments and under similar circumstances.
       Taking the results of all of these studies together, however, it should become evident that recovering alcoholics and former drug-users of all types should stay away from any cues that might remind them of their former drug-use patterns! From the Ehram et al (1972) study, it should become evident that exposure to such cues will only result in withdrawal symptoms- which might lead the previous user might give into, allowing themselves "just a tiny bit" (and, in fact, the criteria for success in Caddy and Lovibond's study would still label them as a "treatment success" if they did so!) However, according to Domjan's (2009) elaboration on this, such a slight dosage of the drug might serve as a CS which would stimulate the expectation that additional drug intake would soon be taking place. If it did, that would be bad enough in itself (as it might lead to relapse)- but, when the results of the conditioning model of drug tolerance and the fact that such a recovering user might be taking an especially high dose in an unfamiliar context and perhaps under different circumstances than those under which they had consumed the drug in the past (since, at this point, the recovering alcoholic will have just completed treatment)-such an event might actually result in tragedy!
Thus, contrary to the specifications denoted in the study by Caddy and Lovibond (1992), convergent evidence would indeed suggest that total abstinence, rather than any other means of treatment, really is "the way to go" for recovering alcoholics and other drug users.
                                                           
                                                                           References
1). Caddy, G. R., & Lovibond, S. H. (1976). Self-regulation and discriminated aversive conditioning in the modification of alcoholic's drinking behavior.Behavior Therapy, 7(2), 223-230. doi: 10.1016/S0005-7894(76)80279-1

2). Domjan, M. (2009). Learning and behavior. (6 ed., pp. 107, 115). Belmont, CA: Wadsworth, Cengage Learning.


3). Ehrman, R. N., Robbins, S. J., Childress, A. R., & O'brien, C. P. (1992). Conditioned responses to cocaine-related stimuli in cocaine abuse patients.Psychopharmacology, 107(4), 523-529. doi: 10.1007/BF02245266


4). Garcia, J., & Keolling, R. A. (1966). Relation of cue to consequence in avoidance learning. Psychonomic .Sci., 4, 123-124. Retrieved from http://www.houptlab.org/Papers/classicCTA.html