Psychopharmacology (1996) 127 : 359- 364 �9 Springer-Verlag 1996

Mark T. Fil lmore �9 M . VogeI-Sprott

Social drinking history, behavioral tolerance and the expectation of alcohol

Received: 9 January 1996 / Final version: 13 June 1996

Abstract This research tested the hypothesis that a compensatory response to cues for alcohol contributes to the greater behavioral tolerance displayed by more experienced social drinkers. Sixty male social drinkers, ranging from 19 to 24 years of age, participated in the study. Thirty subjects had been drinking for 20 months or less (mean-" II.0 months), and were classified as novice (N) drinkers. The remaining 30 subjects had been drinking for 24 months or more (mean = 40.8 months), and were classified as experienced (E) drinkers. All subjects practiced a pursuit rotor task that measured psychomotor skill. Equal numbers of E and N subjects were then assigned to one of three groups (n = 10). Two groups were led to expect alco- hol and performed the task after receiving either 0.56g/kg alcohol, or a placebo. The third group received no beverage. E subjects displayed more behav- ioral tolerance to alcohol than did N drinkers. In accord with the hypothesis, E drinkers displayed a drug- opposite improvement in performance under a placebo, whereas N drinkers showed no appreciable change in performance. These observations support the hypoth- esis that the acquisition of a compensatory response to cues predicting the administration of alcohol con- tributes to the greater behavioral tolerance of more experienced drinkers.

Key words Tolerance - Alcohol - Placebo �9 Drinking history - Learning - Human

This work was supported by a grant from the Alcoholic Beverage Medical Research Foundation

M.T. Fillmore (LS:~) �9 M. Vogel-Sprott Department of Psychology, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1

Introduction

Drug tolerance is defined by a diminished response to a given dose of a drug. One explanation of tolerance attributes it to the development of an adaptive drug- opposite response that serves to counter and reduce the initial drug effect (e.g., Kalant et al. 1971; Siegel 1989; Vogel-Sprott 1992). Research on animals has shown that, after tolerance is established with repeated drug administrations, drug-opposite responses are elicited by pre-drug cues that signal the administration of the drug (e.g., Le et al. 1979; Crowell et al. 1981). In addition, studies have shown that maximum drug tolerance is observed when the drug is taken in the presence of pre- drug cues that reliably signal its administration (e.g., Siegel 1989). These findings are consistent with the notion that drug-opposite responses contribute to tol- erance, and they can be associated with cues that pre- dict the administration of the drug.

It has long been known that alcoholics and others who have a long history of heavy alcohol consumption display little behavioral impairment after drinking amounts of alcohol that would greatly impair lighter drinkers (e.g., Mendelson and Mello 1966; Nathan et al. 1971). Such observations may be attributed to differences among individuals in the strength of a drug- opposite response that compensates for behavioral impairment. A longer history of greater alcohol use on the part of alcoholics and heavy drinkers may serve to strengthen this drug-opposite response and its associ- ation with pre-drug cues, so that greater compensation for impairment (i.e., more behavioral tolerance) is shown.

Little is known about the possible development of behavioral tolerance in social drinkers. However, a recent study of the behaviorally impairing effect of a challenge dose of alcohol in young social drinkers showed that those who had been drinking regularly for 2 years or more displayed less impairment (i.e., more tolerance) than novice drinkers who had been using

360

a lcoho l for less than 1 yea r ( F i l l m o r e and Vogel-Sprot t 1995a). These g roups d id not differ in their ages, b lood a l coho l concen t r a t i ons (BACs) , o r in their cur ren t d r ink ing habi ts (e.g., typical q u a n t i t y c o n s u m e d per occas ion and weekly f requency o f a lcoho l use). Thus, it a p p e a r e d that the d i sp lay o f to le rance was related to subjec ts ' d r i n k i n g histories , in te rms o f how long they had been using a l coho l ra ther than to their cur ren t quan t i t y and f requency o f a l coho l use.

M a n y factors might be sugges ted to accoun t for the re la t ionsh ip between the to le rance d i sp layed in social d r inkers and the d u r a t i o n o f their d r ink ing experience. However , one poss ib i l i ty is associa t ive learning. Ind iv idua l s engage in a var ie ty o f act ivi t ies in social d r ink ing s i tuat ions, where they m a y exper ience the behaviora l effects o f the d rug and the associa ted envi- ronmen ta l consequences . Wi th grea te r exper ience o f regula r a lcoho l use, env i ronmen ta l cues, as well as the scent and taste o f a lcohol , come to predic t the admin - i s t ra t ion o f a lcohol and the ensuing effect o f the drug. Research has shown a d r u g - o p p o s i t e response that serves to reduce behaviora l i m p a i r m e n t (i.e., yield tol- erance) is d i sp layed by social d r inker s in a s i tua t ion that provides a favorable consequence for c ompe n - sa tory pe r fo rmance (Vogel-Sprot t 1992). This f inding suggests that grea ter to le rance on the par t o f social d r inkers with grea te r exper ience o f regula r a lcoho l use may, in par t , be due to the acqu is i t ion o f a c ompe n - sa to ry response to cues for a lcohol . This hypothes is can be tested by admin i s t e r ing a p lacebo in the pres- ence o f cues for a lcoho l admin i s t r a t i on . In the absence o f a lcohol , the cues should lead to a c o m p e n s a t o r y response that should be observed as improved perfor- mance.

The present exper imen t examined social d r inkers ' m o t o r skill pe r fo rmance to test the hypothes is that "exper i enced" d r inkers would d i sp lay a c o m p e n s a t o r y response to p lacebo c o m p a r e d with "novice" dr inkers , and be less impai red by a lcoho l (i.e., more to lerant ) . The d r ink ing histories o f social d r inker s were classified as e i ther novice o r exper ienced in te rms o f the number o f m o n t h s that they had been d r i n k i n g on a regula r basis. These two g roups pe r fo rmed a m o t o r skill task under a lcohol , p lacebo, or no beverage. It was pred ic ted that when a lcohol was expected but a p l acebo was received, exper ienced dr inkers would c o m p e n s a t e and thus d i sp lay supe r io r p e r f o r m a n c e c o m p a r e d with novice drinkers. Similarly, exper ienced d r inkers would be less impai red under a lcohol , and thus pe r fo rm bet- ter under the d rug than would novice dr inkers . Thus under a p lacebo o r under a lcohol , the p e r f o r m a n c e o f exper ienced dr inkers should be s u p e r i o r to novice drinkers. The s tudy also inc luded a no beverage con- trol t r ea tment that served to verify that the pe r fo rmance o f exper ienced and novice d r inkers d id not differ when no a lcohol was expected or received, thus ind ica t ing that d r ink ing his tory is i r re levant to p e r f o r m a n c e in a n o n - d r i n k i n g context .

Materials and methods

Subjects

Sixty male university students between the ages of 19 and 24 par- ticipated in the experiment. Data from 28 participants in the pre- sent experiment have been published in a previous report of a study that examined individual differences in expectancies about alcohol effects (Fillmore and Vogel-Sprott 1995a). All subjects were selected from a pool of university student volunteers who indicated a will- ingness to participate in psychology experiments. These individu- als completed a drinking habit questionnaire that measured their current typical drinking habits, as well as their drinking history in terms of the length of time that they had been regularly drinking alcohol (see Apparatus and materials). In accord with prior research, the response to this item was used to classify the volun- teers in terms of their drinking history (Fillmore and Vogel-Sprott 1995a). Volunteers were classified as either "novice" (i.e., drinking for 20 months or less) or "experienced" (drinking for 24 months or more) social drinkers.

Thirty novice and 30 experienced male volunteers were randomly selected from the volunteer pool to participate as subjects in the experiment. All subjects were in good health and were taking no medication. All subjects were social drinkers who reported no prior treatment for alcohol abuse or alcohol-related offenses (e.g., dri- ving while intoxicated). They fasted for 4 h prior to the experiment and abstained from alcohol for 24 h. They received $6 for their par- ticipation. Ethical approval for the project was obtained from the University Office of Human Research.

Apparatus and materials

Pttrsuil rotor task

A computerized pursuit rotor (PR) tracking task requiring psy- chomotor coordination was used to measure subjects' performance. The task has been used in other research that examined the behav- ioral effect of alcohol (e.g., Fillmore and Vogel-Sprott 1995a). It consisted of a computer and monitor on a table top, 75 cm above the floor. The subjects sat in a chair directly in front of a computer screen that displayed a rectangular track (14 cm • 11.5 cm) and an on-screen target (diameter = 1.3 cm) that moved clockwise at 23 RPM arotmd the track. The subjects tracked the target by moving a computer mouse on the table top to control an on-screen circu- lar cross-hair sight (diameter = 1.3 cm). The subject was required to keep the sight over the rotating target as long as possible dur- ing a trial. The computer measured performance as the percentage of time on target (%TOT) during each trial and stored the trial scores on a computer disk.

In view of evidence showing that environmental consequences or evaluative feedback about performance under alcohol can alter sub- jects" responses (Vogel-Sprott 1992), the PR task excluded these fac- tors. No feedback was provided to the subject about his perlbrmance. The computer task controlled the entire test proce- dure. so subjects always performed the task alone in the room. This served to reduce any possible experimenter or other environmental influences that may affect the subject's performance. The experi- menter was only with the subject in the test room to introduce and explain the task at the outset of the study.

Drhtk#tg habit questionmthe

The drinking habit questionnaire (Vogel-Sprott 1992) yielded two measures of a subject's current drinking habits: frequency (num- ber of drinking occasions per week), and dose (ml of absolute alco- hol per kg body weight typically consumed during a single drinking

361

occasion). An additional item measured drinking history in terms of the number of months that the subject had been drinking alco- hol on a regular basis. This was defined as drinking alcohol on a weekly basis or customarily on social occasions.

Expected type of effect

Subjects rated the expected effect of alcohol on their PR perfor- mance using a scale to report how they expected "two beers drank in 1 h" would affect their performance on the PR task (Fillmore and Vogel-Sprott 1995b). Subjects provided a single rating on a 13- point scale that ranged in 5-point increments from, - 30 "Extremely Impair" to +30 "Extremely Enhance", with 0 indicating that "No Effect" was expected.

Previous research using this scale found that social drinkers tend to expect some degree of slight or moderate impairment from alco- hol and thus give expectancy ratings ranging from - 5 "slight impairment" to - 2 0 "moderate impairment" (Fillmore and Vogel- Sprott 1995b). The present study aimed to obtain the most nor- mative group of drinkers in terms of their expectancies, and to match the range of expectancies among the groups. Thus only sub- jects whose ratings ranged from - 5 to - 2 0 were included in the present study, and novice and experienced drinkers were randomly assigned to the various treatment groups, with the constraint that the expectancy ratings of subjects in each group ranged from - 5 to - 2 0 (see Procedure, Treatment Assignment).

Other research indicates that individuals who report expecting alcohol to enhance motor functioning also tend to report heavy alcohol consumption that may place them at a greater risk for devel- oping alcohol-related problems (Goldman et al. 1987; Mann et al. 1987). Because the present study was concerned with examining expectancies of social drinkers, individuals who expected either no impairment or enhancement were considered unrepresentative of this population and were excluded. Only two novice and three expe- rienced drinkers in a total of 60 volunteers who arrived to partici- pate in the experiment reported expecting either no effect, or improved performance. These five individuals were dismissed prior to receiving any treatment, and were replaced with additional vol- unteers.

Procedure

Subjects were told that the experiment aimed to investigate the effects of alcohol on psychomotor performance, and a general expla- nation of the study was provided before written informed consent was obtained. After subjects were weighed, they again completed the Drinking Habits Questionnaire in order to check the consis- tency of their reports of months of regular drinking. All of their reports on the second administration of the questionnaire verified their classification as novice or experienced drinkers.

Baseline training

Each subject was escorted to a testing room that contained the PR task. After the task was explained, the subject performed a 50-s familiarization trial and the experimenter answered any additional questions regarding the task requirements. Baseline training on the PR consisted of 12, 50-s trials separated by 30-s rests. Trials were performed in three blocks of four trials each, with 2-min rests between blocks. During the first block of trials, the experimenter remained in the room to ensure that the task instructions had been understood. Thereafter the subject performed the task alone. The baseline training required approximately 20 rain. All subjects then completed the Expected Type of Effect Scale and returned to the waiting room.

Treatment assignment

The 30 novice (N) drinkers were assigned to three equal sized treat- ment groups (group n = 10): alcohol (NA), placebo (NP), or no beverage control (NC). The 30 experienced (E) drinkers were also assigned to equal-sized groups that received alcohol (EA), placebo (EP), or no beverage (EC). Group assignment ensured that the range of the subjects' expectancy ratings were equal in each group. For example, the first novice drinker who reported an expectancy rat- ing of - 2 0 was randomly assigned to one treatment condition and the next novice drinker reporting the same expectancy rating was assigned to another treatment group. The same assignment process was used for experienced drinkers.

Treatment

Groups NA and EA received 0.56 g/kg absolute alcohol in a bev- erage containing 2 parts tonic water and 1 part alcohol. An equal volume of the beverage was divided into 2 glasses. The two drinks were served 5 min apart, and subjects drank each drink within l min. Eighteen minutes after drinking commenced, the experi- menter measured the subjects' BACs (Smith and Wesson Breathalyzer, Model 900A). This measure required approximately 2 min.

Groups EP and NP were told that they would receive alcohol but instead received an alcohol placebo consisting of tonic water with a few drops of alcohol on top of the drink. It was served in two drinks of equal volume. The glasses were sprayed with an alco- hol mist that appeared as condensation and provided a strong alco- hol scent as the beverage was consumed. In addition, just before the subject returned to the waiting room, the room was lightly sprayed with an alcohol mist that produced a scent of alcohol. These procedures provided taste and olfactory cues to enhance the belief that the beverage contained alcohol (Rohsenow and Marlatt 1981). Previous research has shown that subjects believe that this placebo beverage contains alcohol (Fillmore et al. 1994; Fillmore and Vogel- Sprott 1995b). The two drinks were served 5 rain apart, and sub- jects drank each drink within 1 rain. Eighteen minutes after drinking commenced they provided breath samples, ostensibly to measure BAC.

Groups NC and EC received no beverage. These subjects pro- vided a measure of performance when no drug was expected or received. When they returned to the waiting room after their base- line training, they were told that they had been assigned to a no- treatment condition and would receive no alcohol during the experiment. They read magazines while they remained in the wait- ing room for a 20-min period. This rest interval was provided so that the time away from the task in groups NC and EC was iden- tical to that of the alcohol and placebo treatment groups.

All subjects subsequently entered the test room alone and per- formed 12 trials on the PR task. The trial schedule was identical to the baseline training (i.e., three blocks of four trials). When the tri- als were completed, breath samples were provided by subjects in the alcohol and placebo groups.

Manipulation checks

To ensure that subjects who received a placebo felt completely free to report honestly on the contents of their beverage, a post-exper- imental inquiry was conducted after subjects were paid. Groups EP and NP were told that some participants had received a placebo instead of alcohol. Subjects were then asked if they knew whether or not they had a placebo, and if so at what time did they realize that no alcohol had been received. All subjects in groups EP and NP reported they believed they had alcohol. All subjects were then fully debriefed.

362

Criterion measures and data analyses

Subjects' scores on the PR improved as they learned the task dur- ing baseline training. The mean of the three highest scores on the last block of training trials represented asymptotic performance, and was used to measure a subject's pre-treatment baseline perfor- mance (i.e., mean %TOT). Because BACs would rise throughout the treatment trials, the mean %TOT of all 12 trials provided a subject's treatment score on the task. The change in performance under each treatment was calculated by subtracting a subject's pre- treatment %TOT from his treatment %TOT. Therefore, a negative change in %TOT indicated poorer (impaired) treatment perfor- mance.

The effects of drinking history on the performance under each treatment were analyzed by a 2 (Drinking History) • 3 (Treatment) analysis of variance (ANOVA) of subjects' change in %TOT scores. This analysis tested the effects of drinking history under the three treatments, and provided the most representative variance estimate for testing hypotheses by a priori planned comparisons based on tests of simple effects. In addition to a 2 (Drinking History) • 3 (Treatment) ANOVA using performance change scores, a 2 (Drinking History) • 3 (Treatment) analysis of covariance (ANCOVA) was performed on subjects' treatment scores, using their pre-treatment scores as a covariate. These two types of analy- ses produced identical conclusions. Because the ANCOVA yields adjusted scores, and change scores provide a direct indication of the actual observed response to treatment, only analyses of change scores are reported.

Results

Subject characteristics

The sample of subjects in the study were young drinkers, ranging in age from 19 to 24 years. The novice drinkers had a history of drinking on a regular basis for a mean of 11.0 months (SD = 6.5). The period of time that they had been drinking regularly ranged from 1 to 20 months. The experienced drinkers had a his- tory o f drinking on a regular basis for a mean of 40 .Smonths ( S D = 15.8). Their periods of regular drinking ranged from 24 to 84 months. The mean ages of the experienced and novice drinkers were 21.0 (SD = 1.5) and 19.4 (SD -- 0.6) years, respectively, and an independent t test revealed a significant difference in age between novice and experienced subjects [t(58) = -5 .34 , P < 0.001].

Independent t-tests obtained no significant diff- erences between experienced and novice drinkers on their measures of current typical dose (P = 0.938), and frequency of drinking (P = 0.122). The overall sample mean frequency of drinking was 1.2 times per week (SD =0.8) , with an average dose per occasion of 1.1 m l / k g (SD = 0.6). This dose is the equivalent o f 4.8 bottles o f 5% alcohol beer for a 75 kg person. These quantity and frequency measures are typical for male social drinkers, aged 19-25 years (Vogel-Sprott 1992, p. 195).

Subjects were assigned to each group so that the range of expectancy ratings (i.e., - 5 to - 20) were iden- tical. A one-way ANOVA of the ratings showed no

significant differences among the groups [F(5, 54) = 1.0, P = 0.452], and confirmed the success o f matching expectancies among groups. In addition, the range of the expectancy ratings was equal among all groups.

Pre-treatment baseline per formance

A one-way ANOVA of the mean pre- t reatment base- line % T O T scores obtained no group effect [F(5, 54) = 0.7, P = 0.651]. Thus the per formance of the six groups did not differ prior to treatment. The mean pre-treat- ment baseline score of all 60 subjects was 58.5 % T O T (SD = 9.2).

Treatment effects

The changes in % T O T scores were analyzed by 2 (Drinking H i s to ry )x 3 (Treatment) ANOVA. The analysis revealed significant main effects o f t reatment [F(2,54) = 21.6, P < 0.001], and drinking history [F(1, 54) = 8.0, P = 0.006], and no treatment x drink- ing history interaction [F(2, 54) = 1.9, P = 0.150]. The means of the change in % T O T for each group are illus- trated in Fig. I. Zero on the ordinate represents the pre-treatment baseline performance, and negative change scores indicate poorer (i.e., impaired) perfor- mance. The figure shows that the mean per formance of the placebo groups (EP and NP) showed some improvement. The alcohol groups (EA and NA) dis- played impaired performance, and the control groups

O 2

Z 0

z -2

-4 z < I.t.t -6

-8

-10

6

41- DRINKING HISTORY

EXPERIENCED I I I NOVICE

PRE~TREATMENT BASELINE

EP NP EC NC EA NA

PLACEBO CONTROL ALCOHOL

TREATMENTS

Fig. 1 Mean change in performance of groups with experienced or novice drinking histories under either placebo, control, or alcohol treatments (group n = I0). Solid bars represent experienced drinkers and striped bars represent novice drinkers. Each bar shows the mean (+ SEM). Zero on the vertical axis represents subjects' pre-treatment percentage of time on target (%TOT) scores. A negative change indicates impaired performance, and a positive change indicates improvement

363

(EC and NC) essentially maintained their pre-treat- merit level of performance.

The prediction that experienced drinkers should dis- play a compensatory response (i.e., improved perfor- mance) when they expect alcohol but receive a placebo was tested by comparing the change in performance displayed by groups EP and NR Figure 1 indicates that group EP displayed greater improvement in perfor- mance than group NP, and this was confirmed by a one-tailed simple effects analysis based on the mean square error term from the ANOVA [F(1, 54)= 3.3, P = 0.038]. This conclusion was also verified by a sep- arate one-tailed independent t-test that revealed group EP displayed a greater improvement under placebo compared with group NP [t(18) = 2.0, P = 0.029]. Thus both the simple effects comparisons from the overall ANOVA and a separate independent t-test supported the hypothesis that experienced drinkers would display superior performance under placebo compared with novice drinkers. The extent of improvement under placebo in each group was also tested by the degree to which the mean change in the performance of a group differed from its zero pre-treatment baseline. Dependent t-tests showed significant improvement in group EP [t(9)= 3.6, P = 0.006], and no significant improvement in group NP (P = 0.392). Thus, the receipt of a placebo resulted in significant compensatory improvement in the performance of experienced drinkers whereas novice drinkers failed to display any significant improvement.

In accord with previous research (Fillmore and Vogel-Sprott 1995a), the figure indicates that group EA displayed less impairment (i.e., more tolerance) than group NA under alcohol. A one-tailed simple effects analysis based on the mean square error term form the ANOVA confirmed this difference [F(1,54)=8.7, P=0.002]. Groups EA and NA did not differ significantly in their mean BACs at the beginning of the treatment trials (P = 0.357), and the mean BAC of both groups at this time was 50.5mg/100ml (SD = 13.5). A group difference in mean BACs measured at the conclusion of the treatment trials did approach significance (P = 0.085). Group EA, who displayed the least impairment, had a higher mean BAC of 71.8mg/100ml (SD=II .5 ) . Group NA had a lower mean BAC of 63.5mg/100ml (SD=8.6) . Thus EA subjects were more tolerant despite their somewhat higher BACs during the course of the treat- ment trials.

A simple effects analysis compared the mean treat- ment change in %TOT of control groups (EC and NC), and revealed no significant group difference [F(I, 54) --- 0.2, P=0.876]. Thus when alcohol was neither expected nor received, subjects' drinking histories did not affect performance.

Because the mean ages of experienced and novice drinkers differed in this experiment, it was important to determine that the differences in performance

between experienced and novice drinkers under the treatments could not be attributed to differences in their ages. This was confirmed by 2 (Drinking History) • 3 (Treatment) ANCOVA of subjects' changes in %TOT, using age as a covariate. This ANCOVA produced con- clusions identical to the ANOVA, and also showed that age was not a significant covariate [F(1,53)=0.4, P = 0.523]. Thus the ANCOVA indicated that age bore no relationship to subjects' change in %TOT. These results confirm that the conclusion that differences in performance between experienced and novice drinkers under the treatments were due to differences in their drinking history rather than their age.

Discussion

The results of the present experiment showed that experienced social drinkers displayed greater tolerance to alcohol-induced impairment than did novice drinkers. The evidence also showed that only the expe- rienced drinkers displayed a drug-opposite improve- ment in performance when they expected alcohol and received a placebo. These observations support the hypothesis that a compensatory response to the expectation of alcohol contributes to the greater behav- ioral tolerance displayed by more experienced social drinkers.

Evidence that a compensatory response contributes to the display of tolerance among social drinkers is consistent with laboratory studies of environmental events that affect alcohol tolerance (e.g., Sdao-Jarvie and Vogel-Sprott 1991; Vogel-Sprott 1992). In that research, different groups of social drinkers performed a psychomotor task under equal doses of alcohol. Some groups were reinforced by receiving a favorable out- come for the display of non-impaired performance. Other groups received no such outcome. The results showed that only groups who were reinforced displayed behavioral tolerance to the impairing effects of alcohol and drug-opposite improvements when a placebo was received. In the present study, no environmental out- comes were contingent upon drinkers' performance under alcohol or placebo. The only factor in the study that appeared to account for the display of tolerance and the compensatory response to placebo was the extent of drinking experience in terms of the number of months that drinkers had been using alcohol. However, a longer history of regular alcohol use could have also exposed the experienced drinkers to more drinking situations in which nonimpaired (compen- satory) behavior was reinforced. This would provide greater opportunity to acquire a compensatory response to cues for alcohol, and this learning may account for the drug-opposite improvement in response to placebo and tolerance in response alcohol displayed by the experienced drinkers.

364

The evidence has implications for interpreting differences in response to a challenge dose of alcohol among social drinkers. Differences in the degree of behavioral impairment under alcohol have been con- sidered as possible behavioral markers that correspond to other known risk factors for the development of alcohol-related problems, such as family history of alcoholism (e.g., Schuckit 1994). However, the present research suggests that prior experience with the drug may also alter the responses of comparatively young social drinkers. In particular, it appears that a com- pensatory response to the expectation of alcohol can develop early during the natural course of regular social drinking. This compensatory response may diminish the degree of behavioral impairment observed under alcohol. As a result, the response to a challenge dose of alcohol may be affected by drinkers' prior learning experiences under alcohol as well as by pre-existing dis- positional characteristics.

The characteristics of drinkers prior to the onset of drinking may be an important consideration in the interpretation of different responses to alcohol and placebo. In the present study, the experienced drinkers had a mean age of 21 years and reported drinking alco- hol regularly for an average of 40 months. Thus the average age of onset for regular drinking was ] 7.7 years. In contrast, the novice drinkers had a mean age of 19.4 years and reported drinking alcohol regularly for an average of 11 months, indicating an average onset age of 18.3 years. Thus the experienced drinkers on aver- age began drinking alcohol at a somewhat earlier age than the novice drinkers. It is possible that individual characteristics that lead to an earlier onset of drinking may contribute to reduced behavioral impairment under alcohol, but it seems unlikely to explain the compensatory improvement to the placebo. Further research that investigates this important issue is warranted.

References

Crowell CR, Hinson RE, Siegel S (1981) The role of conditional drug responses in tolerance to the hypothermic effect of ethanol. Psychopharmacology 73: 51- 54

Fillmore MT, Vogel-Sprott M (1995a) Behavioral effects of alco- hol in novice and experienced drinkers: alcohol expectancies and impairment. Psychopharmacology 122:175-181

Fillmore MT, Vogel-Sprott M (1995b) Expectancies about alcohol- induced motor impairment predict individual differences in responses to alcohol and placebo. J Stud Alcohol 56:90-98

Fillmore MT, Mulvihill LE, Vogel-Sprott M (1994) The expected drug and its expected effect interact to determine placebo responses to alcohol and caffeine. Psychopharmacology 115: 383-388

Goldman M, Brown S, Christiansen B (1987) Expectancy theory: thinking about drinking. In: Blane H, Leonard K (eds) Psychological theories of drinking and alcoholism. Guilford Press, New York, pp 181-226

Kalant H, LeBlanc AE, Gibbons RJ (1971 ) Tolerance to, and depen- dence on, some non-opiate psychotropic drugs. Pharmacol Rev 23:135-191

Le AD, Poulos CX, Cappell H (1979) Conditioned tolerance to the hypothermia effect of alcohol. Science 206:1109-1110

Mann LM, Chassin L, Sher KJ (1987) Alcohol expectancies and the risk for alcoholism. J Consult Clin Psychol 55:411-417

Mendelson JH, Mello NK (1966) Experimental analysis of drink- ing behavior of chronic alcoholics. Ann NY Acad Sci t33: 828-845

Nathan PE, O'Brien JS, Lowenstein LM (1971) Operant studies of chronic alcoholism: interaction of alcohol and alcoholics. In: Mclsaac W, Creaven P (eds) Biological aspects of alcohol. University of Texas Press, Austin, Texas, pp 341-370

Rohscnow D, Marlatt GA (1981) The balanced placebo design: methodological considerations. Addict Behav 6:107-122

Schuckit MA (1994) Low level of response to alcohol as a predic- tor of future alcoholism. Am J Psychiatry 151 : 184-189

Sdao-Jarvie K, Vogel-Sprott M (1991) Response expectancies affect the acquisition and display of behavioral tolerance to alcohol. Alcohol 8:491-498

Siegel S (1989) Pharmacological conditioning and drug effects. In: Goudie A J, Emmett-Oglesby MW (eds) Psychoactive drugs: tolerance and sensitization. Humana Press, Clifton, N.J., pp 115-180

VogeI-Sprott M (1992) Alcohol tolerance and social drinking: learn- ing the consequences. Guilford Press, New York