INTRODUCTION

 

Considerable progress in the understanding of behavioral factors in drug dependence was made after methods were described for inducing animals to administer drugs to themselves. This development led to a new emphasis on the direct study of drug-taking behavior under controlled conditions, and it very effectively complemented the information obtained with the traditional pharmacological and psychiatric approaches, as well as challenging some common assumptions.

 

Researchers studying drug self-administration in animals have frequently interpreted their findings within an operant conditioning framework where the behavior is said to be maintained by positive reinforcing effects of the drugs. Many people have found it difficult to accept that some of the same self-administered drugs are extremely effective agents for producing the phenomenon called conditioned taste aversion (CTA). Drugs such as amphetamine, morphine, and chlordiazepoxide, which are known to serve reinforcing stimulus functions in self-administration experiments are also able to serve as the unconditioned, presumably aversive, stimulus in CTA studies. These superficially "paradoxical" effects need not be considered surprising in view of other evidence that the same nondrug stimulus can serve either reinforcing or aversive functions according to the circumstances (Morse and Kelleher, 1977), and in view of the multiple actions exerted by nearly all drugs.

 

This article reviews some research into CTA which was aimed at clarifying its relevance to drug dependence and to our understanding of how drugs could apparently serve multiple stimulus functions depending on the circumstances surrounding their administration. Although the problem is still unresolved, the work carried out to date exemplifies an analytical approach which has been used in many areas of behavioral pharmacology where drugs have varying and complex effects depending on behavioral and pharmacological variables. The work to be considered has suggested new directions for studies both of self-administration and of CTA, and has also challenged the usual interpretation of CTA in terms of drug-produced illness or toxicity. Identifying the specific variables involved when the same drug can either maintain or suppress responding may eventually make a substantial contribution to knowledge of the conditions under which dependence is most likely to develop. The possible implications of CTA for self-administration research will therefore be discussed. Stoleman and D'Mello (1981) have provided a more detailed account of this work.

 

CONDITIONED TASTE AVERSIONS PRODUCED BY NALOXONE

 

While many behavioral pharmacologists were busily applying the ideas and techniques of operant conditioning to the analysis of drug dependence, a different group of workers was engaged in the intensive study of CTA. In these latter experiments, changes in the consumption of distinctive foods or liquids were taken as evidence for the aversive properties of noxious events such as radiation sickness or emetic drugs (most often lithium or apomorphine salts). Provocative reviews describing much of the earlier work on CTA were written by Garcia and Ervin (1968) and by Revusky and Garcia (1970).

 

An example of CTA techniques in the behavioral analysis of morphine dependence can be found in the work of Pilcher and Stolerman (1976 a,b). Both the natural and the antagonist-precipitated morphine abstinence syndromes had been much studied, at least in part because prevention or termination of abstinent states may be involved in the reinforcement of self-administration behavior. The experiments summarized here were carried out to examine the possible value of CTA as a technique for assessing any aversive property of precipitated abstinence in rats repeatedly treated with morphine. The clear evidence that the narcotic antagonist naloxone could actually control behavior by acting as an aversive event in morphine-dependent rhesus monkeys was a major stimulus for this work (Goldberg et al, 1971).

 

Rats were allowed access to fluid for 1 hour only each day (normally 13.00-14.00 hr). After a period of adaptation to this regimen, one of two distinctively-flavored solutions was presented for lhr on test days. Immediately after the flavored solutions were removed, the rats were injected intraperitoneally with either naloxone or saline solutions (flavor-injection "pairing"). For half of the rats, one flavor was repeatedly paired with naloxone, whereas the other flavor was repeatedly paired with saline. The flavor-injection pairings were reversed in the remaining rats, thus ensuring that effects due to the unconditioned palatabilities of the flavors were balanced out in the averaged results.

 

In rats maintained chronically on morphine (5 mg/kg ip) given twice daily at 09.30 and 17.30 hours, naloxone (10 mg/kg) induced very strong CTA. There were no substantial differences between the initial intakes of the naloxone- and the saline-paired flavors. On trial 2, there was a decrease in the intake of those flavors which had been paired with naloxone (10 mg/kg) on trial 1, as compared with the intake of saline-paired flavors by the same rats. Further, very marked decreases in intake were seen in subsequent trials.

 

It was also possible to induce clear CTA in rats chronically treated with nalorphine, a narcotic partial agonist, whereas much weaker CTA was induced with naloxone in rats which had received neither morphine nor nalophine. The greatly enhanced effectiveness of naloxone in the rats chronically treated with morphine or nalorphine suggested that precipitated abstinence reactions were the major events inducing the CTA in these rats. It seemed, therefore, that the CTA technique provided a means for assessing the morphine and nalorphine abstinence reactions with easily quantifiable and objective behavioral measures. Furthermore, the sensitivity of the method was such that evidence for an abstinence reaction could be found in rats maintained on doses of morphine as low as 1 mg/kg twice daily, ie, at doses around the minimum frequently used to assess acute effects of morphine in nontolerant rats (Pilcher and Stolerman, 1976 a,b). Such observations suggest that great care should be taken before assuming that the self-administration of opioids in low doses is not influenced by withdrawal reactions.

 

The experiments on naloxone-induced CTA may also significantly affect the interpretation of earlier studies on the oral self-administration of morphine (Stolerman and Kumar, 1970). It now seems possible that the behavior called "preferences" for morphine solutions could equally well be called "aversions" to tap water, the alternative source of fluid in those studies. The rats may have learned not that bitter-tasting substances were followed by the pharmacological effects of morphine; instead, they may have acquired a CTA to tap water since on occasions when they drank it instead of the morphine solution, they would subsequently have been exposed to an increasingly intense abstinence reaction.

 

Presently available information does not enable a distinction to be made between the original explanation of morphine preference in terms of primary reinforcement and the new account based on CTA. The plausibility of the latter is strengthened by indications that the natural as well as the antagonist precipitated abstinence reactions may be able to support CTA (Parker and Radow, 1974). Although CTA to familiar fluids such as tap water is slow to develop, it appears eventually (Elkins, 1974; Nakajima, 1974); nobody who has worked on oral morphine intake in rats needs to be reminded that this behavior also takes a long time to develop. In the special circumstances of the morphine-preference experiments, a powerful CTA may have had a considerable but unintended influence on the amounts of drug that were self-administered. A clearer example of the need to take account of possible CTAs when conducting self-administration research can hardly be expected.

 

CONDITIONED TASTE AVERSIONS PRODUCED BY AMPHETAMINE

 

The successful development of a CTA has frequently been taken as evidence that the drug or other event acting as the unconditioned stimulus can serve an aversive function. The preceding series of experiments with naloxone was initiated with such an assumption, but Goudie (1979) has warned that merely because a drug (ie, naloxone) can be found that supports both operant avoidance/escape behavior and CTA, it does not follow that it is serving a similar stimulus function in the two cases. Indeed, as long ago as 1971, Cappell and his co-workers had shown that CTA could be induced with several psychoactive drugs which had not previously been thought to have aversive properties. These drugs were amphetamine, morphine, alcohol, and chlordiazepoxide. Supporting observations with these and other drugs many of which produced dependence in man appeared from many more laboratories and Cappell and Le Blanc (1975; 1977) provided excellent reviews.

 

These observations of CTA produced by agents which, far from being grossly toxic or emetic, were actually capable of supporting operant self-administration behavior by serving as positive reinforcers stimulated much of the work into a so-called "paradox;" how can the same agent both serve as a positive reinforcer and yet appear to have aversive properties? Prior to' reviewing one series of attempts to answer this question, it is worth noting that other, as yet unresolved, basic questions in CTA research might have some bearing on the matter. On the one hand, CTA is most frequently categorized as a form of classical conditioning, presumably because presentations of the unconditioned stimulus are related in time to (consummatory) behavior, but are not programmed as consequences of behavior; regardless of how much behavior is elicited (or emitted?), the rat receives the same dose of drug. On the other hand, the drinking response can be treated as either an operant or a respondent and, more to the point, the amount of contact that the animal makes with the conditioned stimulus (flavor) is very much contingent on its behavior. Eliminating this contingency by directly infusing flavored solutions into the mouth weakened but did not abolish the development of CTA (Domjan and Wilson, 1972). These and related findings have been discussed more fully by Spiker (1977).

 

Common preconceptions about the determinants of behavior have frequently delayed acceptance of research findings; however, there are ample precedents establishing that the same event can serve either positively reinforcing or aversive functions. This seems to be the case both with common exteroceptive stimuli such as electric shock and even with narcotic antagonist drugs (Morse and Kelleher, 1977; Goldberg et al, 1971). There should be little more difficulty in accepting that, for example, amphetamine can serve superficially contrasting siimulus functions than in accepting that it can either increase or decrease the rate at which behavior is emitted according to dose, schedule-associated factors or the previous history of the organism.

 

Goldberg and Spealman (1982) have reported that in squirrel monkeys, infusions of nicotine can either reinforce or punish behavior depending on the circumstances surrounding their presentation. An explanation that drugs have different effects in different circumstances, while very likely correct, is inadequate with regard to CTA and self-administration because it has no predictive power unless the critical factor or factors can be identified. Thus, the robustness and reproducibility of both the reinforcing and the CTA effects of amphetamine provoked the experiments now described.

 

When the ability of amphetamine to produce CTA was first reported, it seemed possible that the doses used may have been too large to support self-administration behavior (Cappell and Le Blanc, 1971). Although later experiments from the same laboratory used a range of doses and still produced only CTA without any sign of a conditioned taste "preference," the conditions used were such that the appearance of any other result was very unlikely (Cappell and Le Blanc, 1973). The baseline comprised the consumption of highly palatable saccharin solutions which were presented as the only source of fluid to rats severely deprived of water. It was thought possible that different results might have been obtained if the baseline of fluid intake was decreased. D'Mello et al (1977) reinvestigated the role of dose-level in conjunction with manipulations of deprivation level, palatability of solutions, times of drug injections and presence of spatial as well as flavor stimuli.

 

The procedure used in these studies involved two-flavor discriminations similar to those described above in connection with the experiments on naloxone-induced CTA. The period of access to flavored solutions was reduced to 15 min in order to shorten the delay between contact with the flavor and onset of drug effect. The lowest dose of amphetamine which produced statistically significant CTA was 0.1 mg/kg, whereas when the dose of amphetamine was increased to 3.2 mg/kg, intake of the amphetamine-paired flavor was almost completely suppressed after only a single flavor-amphetamine pairing (Booth et al, 1977). These dose-response relations were broadly consistent with those found in other laboratories (Cappell and Le Blanc, 1973; Nathan and Vogel, 1975).

 

Neither increasing nor decreasing the interval between presentation of the flavor and injection of amphetamine (1 mg/kg), nor decreasing the baseline amounts of fluid intake altered the final degree of CTA developed after four flavor-amphetamine pairings (D'Mello et al, 1977). Decreases in dose to as low as 0.01 mg/kg, even when combined with reductions in baseline drinking levels, failed to enhance the intake of amphetamine-paired flavors; either there was a weak CTA, or if the dose was too small, there was no effect at all. An amphetamine-produced taste "preference" cannot be obtained simply by manipulating parameters of the situation where it produces CTA.

 

Exteroceptive stimuli paired with reinforcing drug injections can also maintain behavior (Goldberg et al, 1975). Thus, it may be supposed that during drug self-administration experiments, animals may be required to orientate themselves to a particular position in their immediate environment, such as that where the response device may be found. D'Mello et al (1977) examined the role of spatial cues by ensuring that in some rats, the flavored solutions followed by drug injections were always placed on the same side of the cage, whereas the saline-paired flavored solution was always placed on the other side of the cage. Thus, the possibility of increased drinking in places where drinking had previously been followed by amphetamine injections was pitted against the acquisition of the amphetamine CTA. However, CTA developed to about the same extent regardless of whether the rats were presented with flavor cues only or with flavor and possible competing spatial cues. Similar findings have been reported by Martin and Ellinwood (1974).

 

The negative result of the preceding experiment was used to argue against a role for stimulus modality as a factor determining whether amphetamine maintains or suppresses behavior, but it was not conclusive. Variables which may have contributed to the negative finding include differences in the saliency of the flavor and spatial cues, and the initial spatial preferences of the animals prior to conditioning, which may have been too strong to be shifted. Reicher and Holman (1977) presented evidence in favor of stimulus modality as a determining factor, Rats were injected with amphetamine (1.4 mg/kg) and were then confined in one side of a shuttle box in the presence of a flavored solution and of distinctive visual and tactual cues. Subsequently, the rats showed on extremely weak preference for the side of the box associated with amphetamine over the side associated with saline, but would not drink the amphetamine-paired solution.

 

In most self-administration experiments, amphetamine has been taken intravenously, whereas in CTA studies the intraperitoneal route has probably been the most widely used. However, both phenomena have been demonstrated with several routes and there is presently no evidence that simply injecting a drug intraperitoneally instead of intravenously is sufficient to reverse its effects in these experiments. Specifically, Wise et al (1976) and Coussens (1974) have reported on CTA produced by amphetamine administered intravenously. Another possibly relevant factor relates to the frequency of dosing; self-administration typically occurs over many sessions and could perhaps be attributed to chronic effects of drugs, whereas in many cases CTA occurs after an acute administration. This argument is weakened by observations of reinforcing effects of amphetamine within single sessions of self-administration in a very high proportion of the rats tested (eg, Davis and Smith, 1973; 1975), although the weight that can be attached to evidence from these experiments is limited by the narrow range of environmental conditions, the lack of dose-response information, and the general desirability of independent replications. It has not yet proved possible to produce enhanced intakes of flavored solutions after chronic administration of drugs such as amphetamine, although there is a large literature detailing attenuations of CTA after previous exposure to drugs (eg, Cappell and Le Blanc, 1975; Goudie et al, 1976).

 

It has sometimes been suggested that an animal's control of the rate at which stimuli are received may be important in determining their reinforcing value. Behavior was maintained by termination of the same parameters and temporal pattern of stimuli that themselves maintained intracranial self-stimulation behavior (Steiner et al, 1969), although the negative reinforcing effects seemed less powerful than the more frequently studied positive reinforcement. This principle may also be relevant to drug self-administration during which an animal typically has considerable control over the frequency of drug intake, which is not the case during CTA experiments where a single dose is given by the experimenter at a predetermined time (Vogel and Nathan, 1975).

 

The study of Stolerman et al (1971) provides an instance where rats had control over the amounts and temporal patterning of amphetamine intake, and yet the drug seemed to have aversive effects. The procedures in this study were similar to those used successfully for developing the preferential intake of morphine solutions. Initially, the rats consumed approximately equal amounts of dilute amphetamine solutions and water. Over successive choice tests, the intake of the drug solution was suppressed in a concentration-related manner. This was not simply due to a hypodipsic effect of the drug since the consumption of the plain water which was also available remained high. It was suggested that the solutions of amphetamine were rejected because the drug had an aversive pharmacological effect. If the drug solution was unpalatable, it should have been rejected on the first choice trial, which was not the case, and masking the taste should have increased intake. In fact, adding saccharin enhanced the rate at which the aversion developed, presumably because this made the drug solutions more easily discriminable from water (Hill and Powell, 1976). Other experiments have amply confirmed the apparently aversive effects of amphetamine added to rats' food or water (Le Magnen, 1969; Carey, 1973a; Panksepp and Booth, 1973; Glick, 1973). Experiments of this type seem to demonstrate nothing more or less than CTA produced by orally self-administered amphetamine.

 

There are also a number of studies where stimuli paired with programmed administrations of drugs acquired positively reinforcing properties (Davis and Smith, 1973; Marcus et al, 1976; Numan et al, 1976). The weight that can be given to these comparisons is limited by the absence of a systematic study with amphetamine similar to that carried out with electric shock; visual stimuli paired with programmed shocks acquired aversive properties, but the same visual stimuli paired with the same number and temporal pattern of response-produced shocks did not become aversive (Orme-Johnson and Yarczower, 1974). Although none of the experiments reported to date seem to have compared directly the effects of programmed with self-administered amphetamine, they all militate against this factor as critical for the production of CTA.

 

Finally, attention was given to the role of response variables since the preceding considerations tended to exclude most of the other obvious factors. These experiments are described next and, while also failing to resolve "the paradox", they do show how flavor-drug pairings can have a very powerful influence on operant behavior and they may have considerable heuristic value.

 

INFLUENCE OF CONDITIONED TASTE AVERSIONS ON OPERANT BEHAVIOR

 

The type of response required from the rat is one of the many variables which may be relevant to the multiple stimulus properties of amphetamine; CTA involves drinking but self-administration typically requires bar-pressing. Seligman and Hager (1972) were among those who argued that different classes of response were conditioned most rapidly by different classes of consequences; it was suggested that flavor-drug pairings would exert less effect on an "arbitrary" operant such as bar-pressing than on a "naturalistic" consummatory response such as drinking. However, little evidence was cited to suggest that flavors were inherently unable to influence schedule-controlled behavior in ways resembling the more frequently studied auditory or visual stimuli. The experimental techniques used in CTA research generally involved only a limited range of standardized measurements. Usually, only aspects of the gross intake of food or water were assessed and much less was known about possible changes in other types of behavior after encounters with drug-paired flavors (Best et al, 1971).

 

The effects of flavor-amphetamine pairings were examined on bar-pressing for liquid reinforcers delivered on fixed ratio and fixed interval schedules. In the first of these experiments, rats were trained to press a bar for water delivered on a fixed ratio 40 schedule (FR 40). For certain sessions in the next stage of the experiment, flavored solutions were presented in the dipper cup instead of distilled water. At the end of each such session, the rats were injected with either amphetamine (1 mg/kg) or saline, and were then returned to their home cages. The two flavored solutions were presented alternately until each rat had been presented with each flavor on four occasions over a period of several weeks. On days between those on which flavored solutions were presented, the rats responded for distilled water; no injections were given (Stolerman and D'Mello, 1978a).

 

In some of the rats, lemon-flavored water was repeatedly paired with saline and chicken-flavored water with amphetamine. On the first occasion that each flavored solution was presented, responding did not differ greatly from that for the distilled water which was available previously. Responding for the lemon-flavored water remained reasonably stable throughout the experiment. After only a single pairing of chicken flavored water with amphetamine, responding was disrupted on the next occasion that the chicken flavor was presented. An even more marked disruption of responding was seen after further flavor-amphetamine pairings, culminating in total suppression after a single "reinforcer" in the fifth session where chicken flavored-water was presented.

 

The results just described could merely have been due to an unconditioned effect of chicken flavored water, although the trend across sessions made this unlikely. However, in some rats, the flavor-injection pairings were reversed; in these animals suppression of responding developed to lemon flavor, whereas responding for chicken flavor remained reasonably stable.

 

Similar experiments were carried out with rats which were initially stabilized on a fixed interval 1 min schedule of water reinforcement (FI 1). Responding for whichever flavor was paired with amphetamine (1 mg/kg) was disrupted after a single drug administration; responding for the control flavor remained stable. The disruptive effects became more marked after repeated flavor-amphetamine pairings (D'Mello and Stolerman, 1978).

 

After flavor-amphetamine pairings, both bar-pressing for and the mean amounts consumed of the flavored liquids were greatly reduced. The reductions in fluid intake was brought about partly by the presentation of fewer reinforcers and partly by a reduced consumption of the reinforcers which were obtained. The temporal pattern of responding was assessed by means of the index of curvature described by Fry et al (1960). Amphetamine-paired flavors decreased the value of the index, but this was mainly attributed to the very marked reductions in the total numbers of responses. Responding was irregular throughout the intervals, but at no time was its rate increased by amphetamine-paired flavors.

 

Amphetamine (1 mg/kg) was injected 5 min before sessions of responding for distilled water at a later stage of these experiments. This dose of the drug reduced the total numbers of responses emitted by the rats on the FR 40 schedule without influencing total responses or the amounts of water consumed under the FI 1 schedule. However, the temporal pattern of FI responding was disrupted since the drug increased the numbers of responses early in the intervals and decreased the numbers of responses late in the intervals (D'Mello and Stolerman, 1978). These well-known, schedule-associated effects of amphetamine confirmed the adequacy of the schedule manipulations.

 

It was concluded that flavor-amphetamine pairings could have a very powerful influence on operant behavior and that this was worthy of more intensive study, albeit that suppression of the consummatory response may have mediated the effect. Insofar as the results involved only one dose of amphetamine and two schedules of reinforcement, they were considered as preliminary, but it was difficult in the face of such data to attribute the contrasting reinforcing and aversive properties of amphetamine merely to the classes of responses required by the experimental procedures. The effects of the amphetamine-paired flavors were not schedule-dependent and were much greater than those of omitting the primary water reinforcement throughout the session. The conditioned effects of the flavors were therefore different from the effects of either the unconditioned stimulus (amphetamine) or of an extinction procedure. Glowa and Barrett (1983) have extended these findings by showing that post-session injections of amphetamine paired with visual stimuli can suppress responding in pigeons. There is a very striking contrast between these findings and those where single, post-session injections of drugs maintain responding of monkeys on second-order schedules (Goldberg et al, 1975; Katz, 1979). Analogous positive reinforcing effects of post-session injections in rats seem not to have been reported, but might provide a basis for further studies of variables affecting the appearance of CTA.

 

The experiments reviewed up to now have tended to exclude as critical factors drug dose, route of administration, acute versus chronic dosing, baseline amounts of behavior, stimulus modality and response variables. In some cases the evidence was strong, in others it was weak and derived from experiments originally carried out for other reasons. This was particularly the case because the problem seems largely one of acquisition of behavior, which has been of secondary concern in many studies of drug self-administration. Nevertheless, the research was fast approaching an impasse where every possible factor seemed to have been excluded and it was decided to adopt a more pharmacological approach in a further attempt to resolve the "paradox."

 

A PHARMACOLOGICAL APPROACH TO CONDITIONED TASTE AVERSION

Many hypotheses have been presented as to what actions of drugs produce CTA, but no mechanism has been generally agreed. In early work, radiation or drugs with obvious toxic effects were used, and despite the multiple effects which such treatments had, it was assumed that toxicity, nausea, gastrointestinal distress, or illness was responsible for the CTA. These ideas received little experimental testing until Coil et al (1978) reported that pretreatment with antiemetic agents reduced the suppression of CTA produced by lithium, but attempts to replicate these observations have not yet been successful (Goudie et al, 1982; Rabin and Hunt, 1983). In general, notions such as distress or illness proved difficult to assess and the potency of drugs in producing such effects was not found to correlate with their potency in CTA (reviewed by Braveman, 1977; Goudie, 1979). The CTA produced by nutrient substances or by very low doses of psychoactive drugs such as amphetamine was also difficult to explain as solely due to toxicity (Cappell and Le Blanc, 1975; Deutsch et al, 1976).

 

The aim of the experiments described here was to determine the relative potencies of (+)-amphetamine and some related compounds. It was known from earlier work that there were differences between the profiles of action of these substances, and it was hoped that effects such as anorexia, or actions on particular neurochemical systems, could be correlated with potency in producing CTA. Studies with congeners of amphetamine had been reported previously, but comparisons between different drugs can be misleading unless a range of doses of each substance is studied with a standardized procedure.

 

The results from this series of experiments showed that it was feasible to use a standardized, discriminative CTA procedure to examine differences in potency between different amphetamines (Booth et al, 1977). Even small groups of rats, counterbalanced for flavor-injection pairings and injection sequences, gave reliable results. Indeed, the method was sensitive enough to give results with (+)- and (—)-amphetamine which suggested that the relative effectiveness of drugs may be dependent on time-related factors; delayed injections differentiated between the isomers whereas low doses failed to do so. The study also showed that in CTA, p-chloromethamphetamine was significantly more potent than the parent compound, methamphetamine. These two substances had previously been found to be approximately equipotent as anorexigenic agents in rats, but p-chloromethamphetamine had only about one tenth the potency of methamphetamine in facilitating conditioned avoidance behavior (Cox and Maickel, 1972).

 

It was found that agents potent in producing CTA may be either powerful (amphetamine, methamphetamine) or weak (fenfluramine, pchloromethamphetamine) behavioral stimulants. Furthermore, cocaine, a powerful stimulant, was weak in CTA. Firstly, therefore, it appeared that the potency of amphetamine congeners in producing CTA was unrelated to their stimulant potency. Secondly, the CTA produced by the halogen-substituted derivatives p-chloromethamphetamine and fenfluramine was at least as great as that of the unsubstituted parent compounds; CTA potency could not be correlated with the differential activity of these drugs on catecholamine and serotonin mechanisms. Thirdly, the CTA effect of amphetamine was largely central since para-hydroxylation greatly reduced conditioning, but this conclusion has been challenged recently (Greenshaw and Buresova, 1982). In vitro experiments have indicated little difference between the effects of (+)-amphetamine and p-hydroxyamphetamine on aminergic mechanisms, but the more polar hydroxy-derivative penetrates poorly to the brain. This conclusion about the site of action has been supported by a number of other studies discussed by Lorden et al (1980). Finally, the most effective agents in producing CTA seemed to have in common a powerful anorexigenic effect and a long duration of action. The significance of these two factors was investigated in turn.

 

The conditioned anorexia hypothesis attempts to account for CIA as classical conditioning of the anorexigenic or hypodipsic effects of drugs to the flavor stimuli (Le Magnen, 1969; Carey and Goodall, 1974). Such conditioning would be expected to decrease the intake of flavored solutions, thus producing the effect typically seen in CTA experiments. The idea was attractive since it related CTA to a known behavioral effect and if correct, it would have helped to explain how many self-administered drugs appeared to have aversive properties in CTA procedures; if CTA was essentially a complex way for assessing anorexigenic activity, then its validity as a measure of aversiveness would have been undermined. However, events soon indicated that this rather radical reinterpretation of CTA was unnecessary.

 

Doses of X-radiation which induced CTA did not produce hypodipsia (Garcia et al, 1972; Carroll and Smith, 1974), whereas chlordiazepoxide, a drug which frequently increases food intake, also produced clear CTA (Cappell and Le Blanc, 1975). Doses of lithium, ammonium sulphate, arginine, and glucose which produced similar degrees of anorexia produced varying intensities of CTA (Martin and Storlien, 1976). All these findings argued against the conditioned anorexia hypothesis, but experiments with amphetamines seemed to support it. Carey and Goodall (1974) reported that the relative potencies of (+)- and (—)-amphetamine were similar in CTA and hypodipsia experiments. The studies by Booth et al (1977) also consistently suggested a possible correlation between potencies in these two procedures. Although conditioned anorexia seemed not to be viable as an account of CTA in general, it might have been tenable with regard to CTA produced by amphetamine, a noted and indeed prototypical anorexigenic drug. However, Booth et al (1977) had to rely on estimates of anorexigenic activity based on studies from other laboratories and comparisons with their own work on CTA were difficult for several reasons, each one of which might have influenced relative potencies. It was therefore necessary to compare the results of the CTA studies with assessments of hypodipsia carried out under similar conditions in the same laboratory (Stolerman and D'Mello, 1978b).

 

The results confirmed that the hypodipsic potency of (+)-amphetamine was at least three times that of (—)-amphetamine. Under very similar conditions, and in rats of the same sex, strain and weight, the two isomers were essentially equipotent in CTA (potency ratio = 1.2:1). Cox and Maickel (1972) found little difference between the anorexigenic potencies of meth-amphetamine and p-chloromethamphetamine, and the finding that the hypodipsic potencies of these drugs were similar (1.1:1) was therefore not surprising. Nevertheless, n-chloromethamphetamine was at least twice as potent as methamphetamine in CTA. In summary, drugs which are equipotent in CTA can differ in hypodipsic potency, and drugs with similar hypodipsic potencies can differ in CTA. Fenfluramine also had less than half the hypodipsic potency of (+)-amphetamine, but was at least equipotent in CTA. In a dose of 2 mg/kg, fenfluramine produced very strong CTA after only a single flavor-injection pairing, whereas this dose of the drug was totally lacking in hypodipsic activity. The various dissociations between hypodipsic and CTA potencies were incompatible with the conditioned anorexia hypothesis of amphetamine-induced CTA.

 

Other lines of research have also yielded evidence which appears inconsistent with the conditioned anorexia hypothesis of amphetamine CTA. Pretreatment with a-methyl-p-tyrosine can differentiate the responses to amphetamine and to amphetamine-paired flavors (Carey, 1973b). The effects of presenting flavored solutions previously paired with amphetamine (1 mg/kg) were uniformly depressant on FR 40 and Fl 1 responding, whereas the same dose of amphetamine produced the well-known, rate-dependent effects (see above). Thus, the conditioned response to the flavored solutions differed from the response to the drug both in the studies where the solutions were freely available for drinking and in the studies where they were presented on intermittent schedules. Since conditioned anorexia did not seem to account for CTA produced by amphetamine, attention was then focussed on the growing body of evidence that the duration of drug action was a critical factor.

 

The results from several laboratories consistently implied that the longer the duration of action of a drug, the more effective it was likely to be in producing CTA (Cappell and Le Blanc, 1975; 1977; Goudie et al, 1978; Booth et al, 1977). Testa and Ternes (1977) provided a theoretical basis for such ideas. Cocaine, a short-acting drug, was of low potency whereas the long-acting halogen-substituted amphetamines such as fenfluramine and pchloromethamphetamine were very potent in CTA. The conclusions that could be reached from this work were limited because, in most cases, the drugs between which comparisons were made differed in several respects in addition to duration of action. Furthermore, pretreatment with the metabolic inhibitor SKF-525A increased the effectiveness of amphetamine but not that of cocaine (Goudie and Thornton, 1977; Goudie, 1980). In one experiment, it was shown that a 4-hr inhalation of nitrous oxide produced greater CTA than a 30-min inhalation (Goudie and Dickens, 1978).

 

The recent development of long-acting analogues of cocaine and apomorphine provided an opportunity for reassessing the role of duration of drug action in CTA (D'Mello et al, 1981). It was known that some behavioral and neurochemical effects of the cocaine analogue Win 35,428 were qualitatively similar to those of cocaine, but it had a considerably longer duration of action and was more potent as a central stimulant (Spealman et al, 1977; Heikkila et al, 1979). Diisobutyrylapo-

morphine (DBA) had prolonged apomorphine-like behavioral effects but was similar in potency to apomorphine (Baldessarini et al, 1976; 1977). Comparisons of cocaine with Win 35,428 and of apomorphine with DBA seemed likely to provide more critical tests of the role of duration of action in CTA than, for example, the comparison of cocaine with amphetamine. For these new comparisons to be meaningful, it was necessary to know the relative

potencies of the drugs in a standard behavioral procedure in the relevant species. For this reason the compounds were studied on behavior maintained by a fixed ratio schedule as well as in CTA.

 

In sufficient doses, all the drugs produced CTA and in general, the greater the dose, the more marked was the CTA. Cocaine produced significant CTA only at 100 pmol/kg, the largest dose tested, whereas the lowest effective dose of Win 35,428 was 1.8 pmol/kg. Apomorphine and DBA produced CTA at 0.32 pmol/kg (the lowest dose tested) and at all higher doses. In the study of operant behavior, the key-press responding of rats was maintained at high steady rates by a fixed ratio 30 schedule of food reinforcement. All the drugs reduced response rates in a dose-related manner and except for cocaine, the largest doses used almost completely suppressed responding. Studies of time-effect relations with cocaine and Win 35,428 on the FR 30 schedule confirmed the approximately tripled duration of action of the analogue. Win 35,428 was also much more potent than cocaine, whereas apomorphine and DBA did not differ from each other in potency.

 

Two potency ratios were derived for cocaine and Win 35,428, one based on the studies of CTA (ratio of 35:1) and the other on the studies of FR 30 responding (ratio of 41:1). Clearly, these potency ratios agreed well with each other. There was also close agreement between potency ratios for apomorphine and DBA derived from the studies of CTA and FR 30 responding (ratios of 1.0:1 and 1.1:1 respectively).

 

Since the potency ratios for cocaine and apomorphine relative to their long-acting analogues Win 35,428 and DBA were essentially the same in the CTA and the FR 30 procedures, it was suggested that even marked increases in duration of action do not necessarily have a specific effect on potency in CTA. Goudie and Dickins (1978) have relied upon temporal factors to explain why many drugs with positively reinforcing effects can also produce CTA, but such accounts seem to need revision if not rejection since none of the experiments reported to date has established unequivocally that potency in CTA is correlated with duration of action.

 

SUMMARY AND CONCLUSIONS

 

The experiments discussed in the preceding sections of this article provided several examples of ways in which the conditioning of taste aversions, either intentionally or accidentally, might have influenced drug-taking behavior. Although considerable progress has been made in producing sustained high levels of oral intake of opioid drugs, the possibility that unintended CTA might control this behavior in some circumstances has not been much discussed; the demonstrations of clear CTA produced by the morphine abstinence reaction suggests that such possibilities deserve serious consideration. Indeed, CTA seemed to provide a particularly sensitive behavioral marker for precipitated withdrawal signs with opioids, suggesting that it might also have potential in similar applications with other classes of drugs where abstinence phenomena are much harder to identify by methods used until now. It has been also shown that the oral intake of opioids can be suppressed by CTA produced to their tastes by a variety of agents, including but not limited to naloxone (Fernandez and Ternes, 1975; Ternes, 1975; Mumford et al, 1978). The schedule-induced consumption of drug solutions can be suppressed by taste aversion conditioning, but there is some doubt as to whether the suppression is as great as with deprivation-induced fluid intake (Corfield-Sumner and Bond, 1978; Riley et al, 1979).

 

Most of these studies assumed that CTA was, as its name implied, a straightforward means for assessing aversive effects of drugs. The demonstrations of CTA produced by self-administered substances provides a challenge insofar as a full analysis of the effect might yield information about aversive properties of drugs which limit their intake, perhaps even in the conventional self-administration situations. Drug-taking behavior could then be regarded as regulated by a balance between positively reinforcing and aversive drug effects (Kumar and Stolerman, 1977) and indeed, such assumptions lay behind many of the studies of amphetamine-produced CTA which were discussed earlier.

 

In virtually the only study of its type, Gorman et al (1978) attempted to assess directly the relations between the CTA effect of morphine and oral self-administration of the same drug. Rats were first subjected to a conventional CTA procedure with injected morphine as the unconditioned stimulus, and were then offered sweetened morphine solutions as an alternative to plain water. Rats in which the CTA was very marked subsequently self-administered only small amounts of morphine, the implication being that the CTA effects of morphine are relevant to self-administration and can limit its extent. These studies need to be replicated and developed further before any general conclusions can be reached since Gorman et al (1978) reported only a transient, strain-dependent correlation between CTA and self-administration. Furthermore, only one of the experiments controlled for stimulus generalization between the flavor cue in the CTA phase and the sweetened water in which morphine was subsequently presented; unfortunately, in just this one experiment, CTA was barely detectable.

 

Long series of experiments carried out in various laboratories have failed to identify any specific behavioral or pharmacological variables which might determine whether reinforcing or aversive effects of drugs predominate in a given situation. Dose level, route of administration, frequency of dosing the subjects' control over the situation, stimulus and response variables all seem to be contraindicated by one or more experiments. Although some of the evidence on these matters is inconclusive, the overall trend suggests that the wrong questions may have been asked. Instead of searching for controlling variables in the ways described, perhaps it would be worthwhile to question the interpretations of the basic datum; is the development of CTA sufficient to establish that the event serving as the unconditioned stimulus really has aversive properties? Although there is no doubt that very powerful and reproducible behavioral changes can occur in CTA studies, could the naming of the phenomenon as "conditioned taste aversion" have directed attention away from how it is best interpreted? In a review, McKearney and Barrett (1978) commented,

 

Events that control behavior are necessarily presented or withdrawn according to a schedule, and the exact nature of this relation between behavior and its consequences is of inestimable importance in determining both ongoing behavior itself and the behavioral effects of drugs. Indeed, the processes of reinforcement and punishment cannot be conceptualized independently of considerations related to the scheduling of consequent events.

 

As Spiker (1977) has emphasized, much remains to be done before it can be claimed that the exact nature of the relations between behavior and conse-

quences in CTA experiments has been analyzed fully.

 

It can now be argued that CTA produced by self-administered drugs or by drug-induced withdrawal reactions may have a very significant influence on oral drug self-administration, as illustrated above for morphine and amphetamine. It is also reasonable to suppose that these influences are not limited solely to consummatory responding, but extend also to operant behavior maintained by presentations of drug solutions. Before extrapolations can be made to other laboratory situations or to drug dependence in humans, answers to some very basic questions about CTA are still needed. Precisely which aspects of the contingencies in CTA are necessary and sufficient for the effect to occur? What characteristics are shared by the drugs which are effective in producing CTA? Flow far can the implication that aversive drug effects limit oral intake be extended to other self-administration situations, possibly including the intravenous paradigms, or are the CTA effects specific to particular species, stimuli and responses? Indeed (Stolerman and D'Mello, 1981) have questioned whether it is reasonable to make even the common assumption that the drugs used in CTA are serving an aversive stimulus function. The relative neglect of this whole area by behavioral pharmacologists, despite its twenty-five year history and its prominence in the psychological literature, can be traced to the very different, ethologically-orientated, theoretical positions held by most of the pioneering researchers in CTA and to the associated polemics from both groups. Elimination of these restrictive influences and

further integration of the two approaches seems to be long overdue; it is obvious that CTA, regardless of whether it is best considered as classical conditioning, punishment, or conditioned suppression, can have very powerful effects, the full significance of which can only be determined by future research.

 

REFERENCES

Baldessarini RJ, Walton KG and Borgman RJ: Prolonged apomorphine-like behavioral effects of apomorphine esters. Neuropharm 15:471-478, 1976

Baldessarini RJ, Kula NS, Walton KG and Borgman RJ: Behavioral effects of apomorphine and diisobutyrylapomorphine in the mouse. Psychopharmacol 53:45-53, 1977

Best PJ, Best MR and Ahlers RH: Transfer of discriminated taste aversion to a leverpressing task. Psychonomic Science 25:281-282, 1971

Booth DA, D'Mello GD, Pilcher CWT and Stolerman IP: Comparative potencies of amphetamine, fenfluramine and related compounds in taste aversion experiments in rats. Br J Pharmacol 61:669-677, 1977

Braveman NS: What studies on pre-exposure to pharmacological agents tell us about the nature of the aversion-inducing agent? In: LM Barker, M Best and M Domjan (eds.), Learning Mechanisms in Food Selection. Waco, Texas: Baylor University Press, pp 511-530, 1977

Cappell H and Le Blanc AE: Conditioned aversion to saccharin by single administrations of mescaline and d-amphetamine. Psychopharmacologia 22:352-356, 1971

Cappell H and Le Blanc AE: Punishment of saccharin drinking by amphetamine in rats and its reversal by chlordiazepoxide. J Comparative Physiol Psycho! 85:97-104, 1973

Cappell H and Le Blanc AE: Conditioned aversion by psychoactive drugs: Does it have

significance for an understanding of drug dependence? Addictive Behaviors 1:55-64, 1975

Cappell H and Le Blanc AE: Gustatory avoidance conditioning by drugs of abuse. In: NW Milgram, L Krames and TM Alloway (eds.), Food Aversion Learning. New York: Plenum, pp 133-167, 1977

Carey RJ: Acquired aversion to amphetamine solutions. Pharmacology, Biochemistry and Behavior 1:227-229, 1973a

Carey RJ: Long-term aversion to a saccharin solution induced by repeated amphetamine

injections. Pharmacology, Biochemistry and Behavior 1:265-270, 1973b Carey RJ and Goodall EB: Amphetamine-induced taste aversion: A comparison of

d- versus 1-amphetamine. Pharmacology, Biochemistry and Behavior 2:325-330, 1974

Carroll ME and Smith JC: Time-course of radiation-induced taste aversion conditioning. Physiology and Behavior 13:809-812, 1974

Coil JD, Hankins WG, Jenden DJ and Garcia J: The attenuation of a specific cue-toconsequence association by antiemetic agents. Psychopharmacol 56:21-25, 1978

Corfield-Sumner PK and Bond NW: Taste aversion learning and schedule-indiced alcohol consumption in rats. Pharmacology, Biochemistry and Behavior 9:731-733, 1978

Coussens WR: Conditioned taste aversion: Route of drug administration. In: JM Singh and H Lal (eds.), Drug Addiction, Volume 3. Neurobiology and Influences on Behavior. New York: Stratton Intercontinental Medical Book Co., pp. 241-248, 1974

Cox RH and Maickel RP: Comparison of anorexigenic and behavioral potency of some phenylethylamines. J Pharmacol Exp Ther 181:1-9, 1972

Davis WM and Smith SG: Blocking effect of a-methyltyrosine on amphetamine based reinforcement. J Pharmacy and Pharmacology 25:174-177, 1973

Davis WM and Smith SG: Effect of haloperidol on (+)-amphetamine self-administration. J Pharmacy and Pharmacology 27:540-542, 1975

Deutsch JA, Molina F and Puerto A: Conditioned taste aversion caused by palatable nontoxic nutrients. Behav Biol 16:161-174, 1976

D'Mello GD and Stolerman IP: Suppression of fixed-interval responding by flavour-

amphetamine pairings in rats. Pharmacology, Biochemistry and Behavior 9: 395-398, 1978

D'Mello GD, Stolerman IP, Booth DA and Pilcher CWT: Factors influencing flavour aversions conditioned with amphetamine in rats. Pharmacology, Biochemistry and Behavior 7:185-190,1977

D'Mello GD, Goldberg DM, Goldberg SR and Stolerman IP: Conditioned taste aversion and operant behavior in rats: Effects of cocaine, apomorphine and some long-acting derivatives. J Pharmacol Exp Ther 219:60-68, 1981

Domjan M and Wilson NE: Contribution of ingestive behaviors to taste-aversion learning in the rat. J Comp Physiol Psycho! 80:403-412, 1972

Elkins RL: Conditioned flavor aversions to familiar tap water in rats: An adjustment with implications for aversion therapy treatment of alcoholism and obesity. J Abnormal Psychology 83:411-417, 1974

Fernandez B and Ternes JW: Conditioned aversion to morphine with lithium chloride in morphine-dependent rats. Bull Psychonomic Soc 5:331-332, 1975

Fry W, Kelleher RT and Cook L: A mathematical index of performance on fixed-interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior 3:193-199, 1960

Garcia J and Ervin FR: Gustatory-visceral and telereceptor-cutaneous conditioning-

adaptation in internal and external milieus. Communications in Behavioral Biology Part A, 1:389-415, 1968

Garcia J, McGowan BK and Green KF: Biological constraints on conditioning. In: AH Black and WE Prokasy (eds.), Classical Conditioning II: Current Research and Theory. New York: Appleton-Century-Crofts, pp 3-27, 1972

Glick SD: Impaired tolerance to the effects of oral amphetamine intake in rats with frontal cortex ablations. Psychopharmacologia 28:363-371, 1973

Glowa JR and Barrett JE: Response suppression by visual stimuli paired with post-session d-amphetamine injections in the pigeon. J Exp Anal Behav 39:165-173, 1983

Goldberg SR and Spealman RD: Maintenance and suppression of behavior by intravenous nicotine injections in squirrel monkeys. Fed Proc 41:216-220, 1982

Goldberg SR, Hoffmeister F, Schlichting U and Wuttke W: Aversive properties of nalorphine in morphine-dependent rhesus monkeys. J Pharmacol Exp Ther 179:268-276, 1971

Goldberg SR, Kelleher RT and Morse WH: Second-order schedules of drug injection. Fed Proc 34:1771-1776, 1975

Gorman JE, De Obaldia RN, Scott RC and Reid LD: Morphine injections in the taste aversion paradigm: Extent of aversions and readiness to consume sweetened morphine solutions. Physiological Psychology 6:101-109, 1978

Goudie AJ: Aversive stimulus properties of drugs. Neuropharmacol 18:971-979, 1979 Goudie AJ: Aversive stimulus properties of cocaine following inhibition of hepatic enzymes. IRCS Medical Science 8:58-59, 1980

Goudie AJ and Dickins DW: Nitrous oxide-induced conditioned taste aversions in rats: The role of duration of drug exposure and its relation to the taste aversion-selfadministration "paradox." Pharmacology, Biochemistry and Behavior 9:587-592, 1978

Goudie AJ and Thornton EW: Role of drug metabolism in the aversive properties of d-amphetamine. !RCS Medical Science 5:93, 1977

Goudie Al, Thornton EW and Wheeler TJ: Drug pretreatment effects in drug induced taste aversions; effects of dose and duration of pretreatment. Pharmacology, Biochemistry and Behavior 4:629-633, 1976

Goudie AJ, Dickins DW and Thornton EW: Cocaine-induced conditioned taste aversion in rats. Pharmacology, Biochemistry and Behavior 8:757-761, 1978

Goudie AJ, Stolerman IP, Demellweek C and D'Mello GD: Does conditioned nausea mediate drug-induced conditioned taste aversion? Psychopharmacology 78: 277-281, 1982

Greenshaw AJ and Buresova 0: Learned taste aversion to saccharin following intra-

ventricular or intraperitoneal administration of d,l-amphetamine. Pharmacol Biochem Behav 17:1129-1133, 1982

Heikkila RE, Cabbat FS, Manzino L and Duvoisin RC: Rotational behavior induced by cocaine analogs in rats with unilateral 6-hydroxydopamine lesions of the substantia nigra: Dependence upon dopamine uptake inhibition. J Pharmacol Exp Ther 211:189-194, 1979

Hill SY and Powell BJ: Acquired preference for morphine but not d-amphetamine as a result of saccharine adulteration. Psychopharmacol 50:309-312, 1976

Katz JL: A comparison of responding maintained under second-order schedules of intramuscular cocaine injection or food presentation in squirrel monkeys. Journal of the Experimental Analysis of Behavior 32:419-431, 1979

Kumar R and Stolerman IP: Experimental and clinical aspects of drug dependence. In: LL Iversen, SD Iversen and SH Snyder (eds.), Handbook of Psychopharmacology, Volume 7. New York: Plenum, pp 321-367, 1977

Le Magnen J: Peripheral and systemic actions of food in the caloric regulation of intake. Ann NY Acad Sci 157:1126-1156, 1969

Lorden JF, Callahan M and Dawson R: Depletion of central catecholamines alters amphetamine- and fenfluramine-induced taste aversions in the rat. J Comp Physiol Psycho! 94:99-114, 1980

Marcus R, Carnathan G, Meyer RE and Cochin J: Morphine-based secondary reinforcement: Effects of different doses of naloxone. Psychopharmacology 48:247-250, 1976

Martin GM and Storlien LH: Anorexia and conditioned taste aversions in the rat. Learning and Motivation 7:274-282, 1976

Martin JC and Ellinwood EH: Conditioned aversion in spatial paradigms following methamphetamine injection. Psychopharmacologia 36:323-335, 1974

McKearney JW and Barrett JE: Schedule-controlled behavior and the effects of drugs.

In: DE Blackman and DJ Sanger (eds.), Contemporary Research in Behavioral Pharmacology. New York: Plenum, pp 1-68, 1978

Morse WH and Kelleher RT: Determinants of reinforcement and punishment. In:

WK Honig and JER Staddon (eds.), Handbook of Operant Behavior. New Jersey: Prentice-Hall, pp 174-200, 1977

Mumford L, Teixeira AR and Kumar R: Resistance of morphine-seeking behaviour in rats to pharmacological and behavioural "treatments." Nature 272:167-168, 1978 Nakajima S: Conditioned aversion of water produced by cycloheximide injection. Physiological Psychology 2:484-486, 1974

Nathan BA and Vogel JR: Taste aversions induced by d-amphetamine: Dose-response relationship. Bulletin of the Psychonomic Society 6:287-288, 1975

Numan R, Banerjee U, Smith N and Lal H: Secondary reinforcement property of a stimulus paired with morphine administration in the rat. Pharmacology, Biochemistry and Behavior 5:395-399, 1976

Orme-Johnson DW and Yarczower M: Conditioned suppression, punishment and aversion. Journal of the Experimental Analysis of Behavior 21:57-74, 1974

Panksepp J and Booth DA: Tolerance in the depression of intake when amphetamine is added to the rat's food. Psychopharmacologia 29:45-54, 1973

Parker LF and Radow BL: Morphine-like physical dependence: A pharmacologic method

for drug assessment using the rat. Pharmacology, Biochemistry and Behavior 2:613-618, 1974

Pilcher CWT and Stolerman IP: Conditioned flavor aversions for assessing precipitated morphine abstinence in rats. Pharmacology, Biochemistry and Behavior 4:159-163, 1976a

Pitcher CWT and Stolerman IP: Recent approaches to assessing opiate dependence in rats. In: HW Kosterlitz (ed.), Opiates and Endogenous Opioid Peptides. Amsterdam: Elsevier/North Holland, pp 327-334, 197613

Rabin BM and Hunt WA: Effects of antiemetics on the acquisition and recall of radiation- and lithium chloride-induced conditioned taste aversions. Pharmacol Biochem Behav 18:629-635, 1983

Reicher MA and Holman EW: Location preference and flavor aversion reinforced by amphetamine in rats. Animal Learning and Behavior 5:343-346, 1977

Revusky S and Garcia J: Learned association over long delays. In: GH Bower (ed.), The Psychology of Learning and Motivation: Advances in Research and Theory. New York: Academic Press, pp 1-84, 1970

Riley AL, Lotter EC and Kulkosky PJ: The effects of conditioned taste aversions on the acquisition and maintenance of schedule-induced polydipsia. Animal Learning and Behavior 7:3-12, 1979

Seligman MEP and Hager JL: Biological Boundaries of Learning. New York: AppletonCentury-Crofts, 1972

Spealman RD, Goldberg SR, Kelleher RT, Goldberg DM and Charlton JP: Some effects of cocaine and two cocaine analogs on schedule-controlled behavior of squirrel monkeys. J Pharmacol Exp Ther 202:500-509, 1977

Spiker VA: Taste aversion: A procedural analysis and an alternative paradigmatic classification. The Psychological Record 27:753-769, 1977

Steiner SS. Beer B and Shaffer MM: Escape from self-produced rates of brain stimulation. Science 163:90-91, 1969

Stolerman IP and D'Mello GD: Amphetamine-induced taste aversion demonstrated with operant behaviour. Pharmacology, Biochemistry and Behavior 8:107-111, 1978a

Stolerman IP and D'Mello GD: Oral self-administration and the relevance of conditioned taste aversion. In: T Thompson, PB Dews and WA McKim (eds.), Advances in Behavioral Pharmacology 3. New York: Academic Press, 1981 pp 169-214

Stolerman IP and Kumar R: Preferences for morphine in rats: Validation of an experimental model of dependence. Psychopharmacologia 17:137-150, 1970

Stolerman IP, Kumar R and Steinberg H: Development of morphine dependence in rats; lack of effect of previous ingestion of other drugs. Psychopharmacologia 20:321-336, 1971

Ternes IV: Conditioned aversion to morphine with naloxone. Bull Psychonomic Society 5:292-294, 1975

Testa TJ and Ternes JW: Specificity of conditioning mechanisms in the modification of food preferences. In: LM Barker, MR Best and M Domjan (eds.), Learning Mechanisms in Food Selection. Waco, Texas: Baylor University Press, pp 229-253, 1977

Vogel JR and Nathan BA: Learned taste aversions induced by hypnotic drugs. Pharmacology, Biochemistry and Behavior 3:189-194, 1975

Wise RA, Yokel RA and deWit H: Both positive reinforcement and conditioned aversion from amphetamine and from apomorphine in rats. Science 191:1273-1275, 1976