Addiction (1995) 90, 1089-1095

RESEARCH REPORT

Conditioned responses to drug-related stimuli:is context crucial?

JANE POWELL

Department of Psychology, Institute of Psychiatry, De Crespigny Park, Denmark Hill, LondonSE5 8AF, UK and Drug Dependence Clinic, Maudsley Hospital, Denmark Hill, LondonSE5, UK

AbstractResearch into the elicitation of conditioned responses (CRs) by drug-related stimuli in drug and alcohol addictshas produced apparently conflicting results in relation both to the form of the CR (drug-like or drug-opposite)and to its importance in influencing relapse rates after detoxification. Most studies looking at cue-elicitedresponses directly have taken place with recently detoxified addicts in a laboratory setting. The present studyexamined cue-elicited reactions in eight currently dependent opiate users (receiving prescribed physeptone)during their drug preparation rituals. Seven were seen in their own homes and one in hospital. Subjects ratedtheir mood and physical state before getting out the drug and paraphernalia, during preparation to inject, andafter injection. By contrast with the withdrawal-like reactions, which are the most common finding inlaboratory exposure studies where drugs are not subsequently accessible for use, these subjects showed asignificant reduction in subjectively rated withdrawal symptoms during the pre-injection drug preparationritual. The results are discussed in relation to the impact of the context of exposure in influencing subjectiveexpectancies, with perception of drug availability seen as a critical part of the conditioned stimulus complex.

Introduction to elicit withdrawal-like symptoms as con-Current interest in cue exposure as a treatment ditioned responses (CRs). If, after detoxification,for reducing relapse rates in alcoholics (e.g. an addict comes across cues which have in thisDrummond & Glautier, 1994) and drug abusers way become CSs for withdrawal, they will ex-(e.g. Dawe et al., 1993) is based on a consider- perience the onset of conditioned symptoms andable volume of evidence that classically con- will then be at risk of lapsing to drug use in orderditioned responses develop during periods of to alleviate their discomfort,dependent use, and can be extinguished via sys- This model has generated a number of testabletematic unreinforced exposure to the con- predictions, and experimental studies haveditioned stimuli (CSs). It was originally argued confirmed that withdrawal sjmiptoms can beby Abraham Wikler (1952) that, through the conditioned both in animals (e.g. Wikler & Pes-repeated experience of withdrawal symptoms in cor, 1967) and in humans (O'Brien et al, 1975)jpredictable contexts, those contextual cues come that detoxified alcoholics and opiate addicts re-

port a subjective withdrawal-like reaction whenexposed to putative CSs (e.g. Teasdale, 1973);

Correspondence to: Jane Powell, Department of Psychology, , , . i .GoldsmiAs College, University of London, Lewisham Way, ^n^ ^^^ systematic Cue exposure results m aNew Cross, London SE14 6NW, UK. diminution of cue-elicited reactions in the labo-0965-2140/95/081089-07 © 1995, Society for the Study of Addiction to Alcohol

and Other Drugs

1090 Jane Powell

ratory setting (e.g. Powell, Bradley & Gray,1993). There are, however, a number of criticalassumptions which remain unsubstantiated orcontroversial. In particular, studies designed toascertain whether detoxified addicts experiencesuch reactions in naturalistic settings wheredrugs are actually available have found little evi-dence for this (Meyer & Mirin, 1979; McAuliffe,1982). Secondly, when relapse episodes areanalysed, very few appear to be clearly precededby conditioned reactions (Marlatt & Gordon,1980), albeit that temptations and urges to useare a salient part of the experience (Heather,Stallard & Tebbutt, 1991). Thirdly, relapse ratesin opiate addicts have been found to be unre-lated to the intensity of subjective reactions ex-perienced during laboratory-based cue exposure(Powell et al., 1993). Finally, the effectiveness ofcue exposure treatment itself in opiate addictionis debatable: controlled trials have so far re-ported little, if any, benefit (Childress, McLellan& O'Brien, 1986b; Dawe et al., 1993), althoughthere have been slightly more promising resultswith alcoholics (Drummond & Glautier, 1994).

There is also theoretical controversy concern-ing the nature of the CRs themselves. The with-drawal-like reactions observed for most subjectsin most laboratory studies can perhaps moreaptly be interpreted as conditioned opponentprocesses, i.e. homeostatic responses which areautomatically triggered by, and oppose, drug-in-duced effects (Siegel, 1975, 1983). Separately,and rather inconveniently, a small number ofsubjects in a small number of studies have showncompletely the opposite reaction to drug-relatedcues—an apparent drug-positive effect. O'Brien(1975) asked detoxified addicts to re-enact theinjection procedure using a saline solution, andfound that a few reported mild opiate-like effectssuch as a 'rush', and increased skin temperature.Meyer & Mirin (1979) allowed detoxified ad-dicts access to heroin after pre-treating half withnaltrexone (an opiate antagonist) and half with aplacebo, using a double-blind design. Althoughthe dose of naltrexone was adequate to blockcompletely the effects of the subsequently in-jected heroin, there was a tendency for thesesubjects to show weak but objectively measuredopiate-like effects such as pupillary constrictionand a reduction in respiratory rate. A similarfinding was reported by Temes et al. (1980),these opiate-like reactions diminishing and in

fact being replaced by withdrawal-like reactionsafter several such trials.

Childress et al. (1986a) interviewed metha-done-maintained patients on a weekly basis andfound that 76% reported at least one occurrenceof a 'high'-like feeling unrelated to actual druguse, while 32% of currently dependent subjectsreported a 'high'-like sensation during labora-tory-based cue exposure. By contrast, the sameresearchers (Childress et al., 1986b) failed to findany significant 'high'-like response in a group ofdetoxified addicts. This overall lack of an effectmay, however, have masked individual differ-ences in responsivity (cf. Powell, Bradley &Gray, 1992).

While there have been attempts from a bio-logical perspective to account for the diversity ofresponses which appear to be conditioned indifferent studies (e.g. Eikelboom & Stewart,1982), there has been no satisfactory explanationof the individual variations in response that areseen within single studies with human subjects.There is a similar failure to investigate or ac-count for the discrepancy between the frequentreports of cue-elicited withdrawal in laboratorystudies and the relative infrequency with whichsuch experiences are reported by addicts in thereal-life setting.

The present study was therefore designed toclarify the nature of any subjective reactions todrug-related stimuli experienced by opiate ad-dicts during a period when they are active drugusers. It is an absolutely central thesis of con-ditioning models of addiction that the CRs ob-served in laboratory studies must have beendeveloped via repeated association between un-conditioned responses and contextual stimuliduring the period of regular drug use. It there-fore follows that those same reactions should bemanifest in current addicts during their routinedaily exposure to drug-related cues. In particu-lar, when the effect of one dose of opiates iswearing off (i.e. when they are no longer intoxi-cated), then exposure to drug-related parapher-nalia should be potent in eliciting any responseswhich have become conditioned to it.

MethodAddicts being maintained on prescribed physep-tone (an injectable opiate) were visited in theirhomes at a time of day when they would nor-mally inject. They rated their craving and their

Conditioning in addiction: context effects 1091

subjective state on the Affect and Physical Symp-toms checklists (see below) as soon as the re-searcher arrived, i.e. before preparing their drug(pre-prep), immediately after getting their equip-ment out and filling the syringe with physeptone(post-prep), and then again shortly after they hadinjected (post-fix).

Subjects had previously been asked not to getout their equipment until the researcher arrived,and she arranged to arrive approximately 15minutes prior to their usual time for injecting.Following completion of the pre-prep ratings sheasked them to prepare their equipment, and todescribe it and their drug-using ritual as they didso. This prolonged the period of simple exposureto the stimuli and also ensured that the subjectwas fiiUy focusing on them, thus optimizing theprobability of any CRs being elicited.

The pre-prep data served as a baseline mea-sure. Although the difference between pre-prepand post-prep was confounded with the simplepassing of time, which would therefore allowwithdrawal symptoms occurring naturally toworsen, the time scale was so short (mean 12minutes, range 2-24) that it seemed unlikely tohave a discernible effect. The post-fix ratingswere taken to ascertain the speed of onset ofdrug effects. If they were relatively slow (as theyare with oral methadone), any conditioned ef-fects could theoretically be of equally slow onsetand thus might not have become apparent by thetime of the post-prep assessment. Mean intervalbetween the post-prep and post-fix assessmentswas 13.4 minutes (SD 6.3).

SubjectsOf 15 addicts prescribed physeptone by the out-patient clinic of the Drug Dependence Unit atthe Bethlem/Maudsley Hospital, nine agreed toparticipate in this study. All but one agreed to beobserved in their own homes; the final subjectbrought his drugs to the hospital instead, wherehe injected in the presence of a doctor and theresearcher.

One of the subjects revealed, after havingcompleted the pre-prep checklists, that he wasplanning to inject a mixture of physeptone andamphetamines and that this was his usual prac-tice. Since regular use of amphetamines, a psy-chostimulant whose effects are in general thereverse of those produced by opiates, may wellhave resulted in the development of anticipatory

conditioned reactions which would be con-founded with those associated with opiate use,this subject was dropped from the analysis.

The eight remaining subjects comprised fivemen and three women, with a mean age of 33.1years (range 29-34). They had been prescribedphyseptone for between 3 and 26 months (mean9.8 months), and doses ranged between 10 and30 mg (mean 21 mg). All but one were receivingoral methadone in addition to their physeptone.

Affect and physical symptoms checklistsThese checklists are described in detail by Pow-ell et al. (1992). Briefly, the Affect Checklist wasderived from the Profile of Mood States (Mc-Nair, Lorr & Droppleman, 1971) and comprised10 mood-related adjectives, two each indexinganger, depression, anxiety, fatigue and vigour.The Physical Symptoms Checklist includedseven symptoms characteristic of opiate with-drawal (coldness, runny nose, muscular tension,aches and pains, goosefiesh, yawning, restless-ness) and three drug-positive symptoms (pleas-ant feeling in stomach, itchy nose, 'high').

Items on both checklists were rated on a 9-point scale ranging fi'om 0 (nil) to 8 (extreme),yielding maximum scores of 16 for each of themood states, 56 for withdrawal-like symptomsand 24 for drug-positive symptoms.

CravingSubjects were asked to rate their degree of crav-ing for opiates ("strength of attraction to the ideaof drug use") on a scale ranging from 0 (nil) to8 (extreme).

ResultsIn view of the small sample size, non-parametricstatistics have been used. For each variable, thedirection and significance of change from pre-prep to post-fix (i.e. the unconditioned drugeffect) has first been established using theWilcoxon matched pairs signed-ranks test. One-tailed significance tests have been used here,since for each variable there is a clear expectationthat the change should be in the direction offeeling better.

Where a significant change has beenconfirmed, a second Wilcoxon's test has thenbeen used to evaluate the difference between

1092 Jane Powell

Table 1. Ratings of craving, physical symptoms, and affective state at the three assessment points

CravingWithdrawal-like

symptomsDrug-positive

effectsFatigueVigourDysphoria

(T + H + D)TensionHostilityDepression

Mean

2.8

18.9

3.44.01.9

11.36.32.72.3

Pre-prep

Median

1.5

10.5

0.03.50.5

7.56.51.00.0

Range

0-8

5-41

0-111-120-7

3-372-160-100-13

Mean

2.9

15.6

5.93.83.1

9.25.22.61.4

Post-prep

Median

1.5

8.5

5.52.01.5

7.56.00.50.0

Range

0-8

3-38

1-141-130-8

0-260-110-100-6

Post-fix

Mean

0.5

3.2

7.11.04.9

2.71.10.70.9

Median

0.0

0.0

6.50.05.0

3.01.00.50.0

Range

0-4

0-17

1-140-60-10

0-60-30-30-3

scores at pre-prep and post-prep, the period overwhich any changes are assumed primarily torefiect classically conditioned reactions. In thesecases, two-tailed significance tests have beenused because, as discussed earlier, conditionedreactions could work in the same or in the op-posite direction to the direct drug effect.

Mean and median scores and ranges areshown in Table 1.

Craving. Levels of craving were very lowthroughout the experiment, the mean scores atthe highest point (pre-prep and post-prep) beingjust below 3 out of a possible maximum of 8.While there was no change fi'om pre-prep topost-prep, however, craving all but vanished fol-lowing drug injection, and the decline was statis-tically significant (Z= - 1.8, p<0.05).

Physical symptoms. Considering first drug-posi-tive symptoms, these were as predicted, virtuallyabsent at the pre-prep assessment (medianscore = 0), while mild to moderate effects wereobserved after injection. This increase wasconfirmed to be significant (Z= 2.11, p< 0.02).However, while the median scores also increasedfrom pre-prep to post-prep, this change was notsignificant (Z<1.0).

Withdrawal symptoms declined significantlyfi'om pre-prep to post-fix (Z= — 2.4, p < 0.001),as predicted. Two-tailed test of the change fi-ompre-prep to post-prep found this reduction alsoto be significant (Z= -2 .2 , p<0.05). Interest-ingly, when the seven individual withdrawalsymptoms were examined separately, mean

scores conformed to this same pattern (i.e. pre-prep > post-prep > post-fix) for all of them. Ofthe eight addicts in the study, six in fact showedthis decline in discomfort from pre-prep to post-prep, with only two showing a slight increase.

Affective state. Vigour showed a significant in-crease from pre-prep to post-fix (Z=2.0,p < 0.05), indicating that it could be considereda drug-positive effect. There was similarly anincrease from pre-prep to post-prep, although,using a two-tailed test, it fell just short of statisti-cal significance (Z= 1.7, p = 0.10).

Fatigue, not surprisingly, showed the inversepattern, decreasing significantly from pre-prep topost-fix (Z= - 2.5, p < 0.01). The change frompre-prep to post-prep, although in the same di-rection, was not significant (Z< 1.0).

Scores for the three remaining mood states(hostility, depression, tension), where medianscores individually all followed the same patternof reducing from pre-prep to post-fix, were com-bined to yield a single 'dysphoria' index. Whilethe change from pre-prep to post-fix was statisti-cally significant (Z= -2.4 , p<0.01), thechange from pre-prep to post-prep was not(Z= -0.9).

It was notable, however, that although ratingsof tension were initially quite high, with no sub-ject giving a zero rating at pre-prep, the medianscore for depression was in fact zero and that forhostility was only 1. The analysis was thereforerepeated for tension alone, and the pattern wasfound to replicate that for dysphoria with asignificant change from pre-prep to post-fix

Conditioning in addiction: context effects 1093

(Z= -2 .2 , p<0.02) but not from pre-prep topost-prep (Z<1.0).

DiscussionThese data reveal a striking failure of subjects toshow any hint of the subjective withdrawal-likereaction which is so apparent in laboratory ex-posure studies with detoxified addicts. Indeed,the reverse pattern is apparent for most of theresponses assessed, with a general improvement insubjects' well-being as they prepare their equip-ment. This improvement was statisticallysignificant, using conservative two-tailed tests,for withdrawal-like symptoms and was shown bysix of the eight subjects. There was a trend forfeelings of vigour to show improvement over thesame interval.

That withdrawal symptoms were apparent atlow to moderate intensity prior to drug prep-aration meant that there was clearly room forthem to change in either direction during stimu-lus exposure. The fact that they reduced appearsto conflict with the conditioned withdrawalmodel, which must predict either stability or anexacerbation of subjective discomfort. The sameargument applies to the conditioned opponentprocess model. Indeed, the only conditioningmodel that is straightforwardly consistent withthese data is that which posits conditioned drug-positive efifects.

A key question, however, is how the presentresults can be reconciled with the well replicatedlaboratory-based observations that detoxifled ad-dicts typically respond to drug-related stimuliwith withdrawal-like reactions (e.g. Childress etal., 1986b; Powell et al., 1992. One possibleapproach is to elaborate existing theories to takeaccount of context.

In the laboratory setting it is usually explicitthat drugs are not available and therefore thatany urge to use will not be satisfied. By contrast,the addicts in the present study were currentusers not attempting to quit, and the drug stimulito which they were exposed quite unambiguouslyheralded imminent drug use and achievement ofdesirable efifects.

It is self-evident that drug availability wouldhave been an important dimension of the ad-dict's learning history, with exposure to a rangeof salient cues (e.g. an empty syringe) sometimesbeing associated with positive outcomes (drugingestion and pleasant effects) and sometimes

with aversive outcomes (unavailability of drugand exacerbation of symptoms). Given thisvariability and the complexity of humanassociative learning, it is plausible that [DRUGSTIMULUS + PERCEIVED A V A I L A B I U T Y ] COuld b e -

come a CS for conditioned drug-positive efifects,while [DRUG STIMULUS + PERCEIVED UNAVAILABIIITY] would become a CS for conditioned with-drawal. This elaboration of existing theory couldexplain many of the discrepant findings reviewedabove. Thus, when heroin is injected experimen-tally under naltrexone blockade (Meyer & Mirin,1979; Temes et al., 1980), or when opiates areactually available, as in the present study, sub-jects are effectively exposed to [DRUG STIMU-LUS + PERCEIVED AVAILABILITY]. This contrastswith the bulk of laboratory-based experiments inwhich purely symbolic cues are employed, effec-tively resulting in exposure to [DRUG STIMU-LUS + PERCEIVED UNAVAILABIUTY] . It ispostulated here that this difference in perceivedavailability is crucial in accounting for the differ-ent forms of responses observed between the twotypes of study.

Indirect support for this model can be ad-duced from studies which have manipulated ex-pectancies directly. For instance, whenalcoholics have been led to believe that they haveconsumed alcohol when in fact they have beengiven a non-alcoholic drink, they have beenfound to show physiological, behavioural andsubjective responses which are similar to thoseproduced directly by alcohol (e.g. Laberg &Loberg, 1989; Laberg, 1990). In a similar studywith non-dependent drinkers, Glautier, Drum-mond & Remington (1992) found that physio-logical reactions differed depending on whethersubjects drank or simply held and smelled thedrink. Newlin (1985, 1989) has similarly re-ported physiological responses to placebo bever-ages under conditions where subjects believeerroneously that they contain alcohol.

While the findings from the present study ap-pear to challenge current formulations of con-ditioning theory, their significance is clearlylimited by constraints inherent in the methodol-ogy used and by the small sample size. Difficultas it is to recruit current opiate addicts willing topermit this intrusion into their normal injectingroutine, a replication with larger numbers isneeded. On the methodological front it wouldobviously have been desirable to assess theseaddicts again in the context of drug non-avail-

1094 Jane Powell

ability, thereby testing the associated predictionthat the opposite reaction—an increase in sub-jective withdrawal-like symptoms—^would beelicited. This was not attempted because of thepractical problem that these addicts, who had noimmediate intention of quitting, were unwillingto abstain voluntarily for a meaningful period intheir own homes solely for the purposes of suchan experiment. However, a laboratory-basedcontrolled experiment with either opiate addictsor people dependent on other substances mightwell be feasible and would serve as a valuablecomplement—or challenge—to the present data.

AcknowledgementsThis research was conducted as part of the au-thor's doctoral dissertation which was funded bythe Wellcome Trust; I would like to thank theTrust, and also my supervisors, Dr BrendanBradley and Professor Jeffrey Gray, for theiradvice and guidance in planning the study.Thanks are also owed to Dr Strang, director ofthe Bethlem/Maudsley Drug Dependence Unit;to the staff of the outpatient clinic for their helpin recruiting subjects; to Lynne Barratt for assist-ing with data collection; to Dr Alan Pickering forstatistical advice; and to Dr Sharon Dawe forconstructive comments on the manuscript.

ReferencesCHILDRESS, A. R., MCLELLAN, A. T. & O'BWEN, C . P.

(1986a) Nature and incidence of conditioned re-sponses in a methadone population: A comparisonof laboratory, clinic, and naturalistic settings, in:Problems of Drug Dependence, 1985, NIDA ResearchMonograph No. 67, pp. 377-372 (Washington, DC,US Government Printing Office).

CHILDRESS, A. R., MCLELLAN, A. T. & O'BRIEN, C . P.(1986b) Abstinent opiate users exhibit conditionedcraving, conditioned withdrawal, and reductions inboth through extinction, British Journal of Addiction,81, 655-660.

DAWE, S., POWELL, J., RICHARDS, D . , GOSSOP, M . ,MARKS, I., STRANG, J. & GRAY, J. A. (1993) Doespost-withdrawal cue exposure improve outcome inopiate addiction? A controlled trial. Addiction, 88,1233-1245.

DRUMMOND, C . & GLAUTIER, S. (1994) A controlledtrial of cue exposure treatment in alcohol depen-dence. Journal of Consulting and Clinical Psychology,62, 809-817.

EiKELBOOM, R. & STEWART, J. (1982) Conditioning ofdrug-induced physiological responses. PsychologicalReview, 89, 507-528.

GLAUTIER, S., DRUMMOND, D . C . & REMINGTON, B .(1992) Different drink cues elicit different physio-logical responses in non-dependent drinkers, Psy-chopharmacology, 106, 550-554.

HEATHER, N., STALLARD, A. & TEBBLJTT, J. (1991) Im-portance of substance cues in relapse among heroineusers: comparison of two methods of investigation.Addictive Behaviours, 16, 41—49.

LABERG, J. C . (1990) What is presented, and whatprevented, in cue exposure and response preventionwith alcoholic subjects? Addictive Behaviours, 15,367-386.

LABERG, J. C. & LOBBRG, T . (1989) Expectancy andtolerance: a study of acute alcohol intoxication usingthe balanced placebo design. Journal of Studies onAlcohol, 50, 448-455.

McAuLlFFE, W. E. (1982) A test of Wikler's theory ofrelapse: the frequency of relapse due to conditionedwithdrawal sickness. International Journal of the Ad-dictions, 17, 19-33.

MCNAIR, D . M . , LORR, M . & DROPPLEMAN, L . F .(1971) Profile of Mood States (San Diego, CA, Edu-cational and Industrial Service).

MARLATT, G . A. & GORDON, J. R. (1980) Determinantsof relapse: implications for the maintenance of be-haviour change, in: DAVIDSON, P. O. & DAVIDSON, S.M. (Eds) Behavioural Medicine: Changing HealthLifestyles, pp. 410-452 (New York, Brunner/Mazel).

MEYER, R. E . & MIRIN, S. M . (1979). The HeroinStimulus: Implications for a Study of Addiction (NewYork, Plenum Press).

NEWLIN, D . B . (1985) The antagonistic placebo re-sponse to alcohol cues. Alcoholism: Clinical and Ex-perimental Research, 9, 411-416.

NEWLIN, D . B . (1989) Placebo responding in the samedirection as alcohol in women. Alcoholism: Clinicaland Experimental Research, 13, 36-39.

O'BRIEN, C . P. (1975) Experimental analysis of con-ditioning factors in human narcotic addiction. Phar-macological Reviews, 27, 533-543.

O'BRIEN, C . P., O'BRIEN, T . J., MINTZ, J. & BRADY, J. P.(1975) Conditioning of narcotic abstinence symp-toms in human subjects. Drug and Alcohol Depen-dence, 1, 115-123.

POWELL, J. H., BRADLEY, B . & GRAY, J. A. (1992)Classical conditioning and cognitive determinants ofsubjective craving for opiates: an investigation oftheir relative contributions, British Journal of Addic-tion, 87, 1133-1144.

POWELL, J. H., BRADLEY, B . & GRAY, J. A. (1993) Sub-jective craving for opiates: evaluation of a cue ex-posure protocol for use with detoxified opiateaddicts, British Journal of Clinical Psychology, 32,39-53.

POWELL, J. H., DAWE, S., RICHARDS, D . , GOSSOP, M . ,MARKS, I., STRANG, J. & GRAY, J. (1993) Can opiateaddicts tell us about their relapse risk? Subjectivepredictors of clinical prognosis. Addictive Behaviours,18, 473^90.

SlEGEL, S. (1975) Evidence from rats that morphinetolerance is a learned response. Journal of Compara-tive and Physiological Psychology, 5, 498-506.

SiEGEL, S. (1983) Classical conditioning, drug toler-ance, and drug dependence, in: SMART, R. G. ,

Conditioning in addiction: context effects 1095

GLASER, F , B. , ISRAEL, Y , , KALANT, H . , POPHAM,R. E. & SCHMIDT, W . (Eds) Research Advances inAlcoholand Drug Problems, Vol. 7, pp. 207-246 QNTewYork, Plenum Press).

TEASDALE, J. D. (1973) Conditioned abstinence in nar-cotic addicts. International Journal of the Addictions,8, 273-292.

TERNES, J. W . , O'BRIEN, C . P., GREENSTEIN, R . &MCLELLAN, A. T, (1980) Unreinforced self-injec-tions: effects on rituals and outcome in heroin ad-dicts, in: HARRIS, L , S . (Ed.) Problems of Drug

Dependence, 1979, NIDA Research Monograph No. 27,pp. 282-288 (Washington, DC, US GovernmentPrinting Office),

WiKLER, A, (1952) Psychodynamic study of a patientduring experimental self-regulated readdiction tomorphine. Psychiatric Quarterly, 26, 270-293.

WlKLER, A. & PESCOR, F , T . (1967) Classical con-ditioning of a morphine abstinence phenomenon,reinforcement of opioid-drinking behaviour and"relapse" in morphine-addicted rats, Psychopharma-cologia, 10, 255-284,