Motivation and Expectancy Factors in Symptom Perception: A Laboratory Study of the Placebo Effect

MARK P. JENSEN, P H D AND PAUL KAROLY, P H D Placebos are widely recognized as having a potentially powerful effect on health and health behavior, yet our understanding of the placebo phenomenon is limited. Many factors are considered to be associated with placebo responding. This study was designed to examine, in a laboratory context, the effects of two of these factors, motivation and expectation, on one kind of placebo response, that which involves the perception of symptoms. The results point to the importance of motivation in influencing the perception of symptom change, and provide limited support for the importance of expectation. The findings underscore the necessity for examining placebo effects from a multicomponent perspective, and appear to mandate the inclusion of both expectancy and motivation factors in future process and outcome studies.

INTRODUCTION

Nonspecific treatment procedures, or placebos, are perhaps the most consistently effective and widely used interventions in the history of medicine (1. 2). Placebos have been shown to influence both attributions of the causes of symptom change (3) as well as actual physical symptoms and behaviors (4-7). However, most research on placebos has focused on the effects of placebo inductions on the perception of symptoms (such as pain, 8; depression, 9; and stimulation/sedation, 10). Many explanations have been proposed for placebo effects. Unfortunately, these explanations are often judged to be inadequate (11-13). Most focus on only one

determinant (such as expectancy, e.g. 1417), while the research evidence suggests that the placebo phenomenon is much more complex (13). Many factors, such as the patient population (14, 18), the symptom to be measured (10, 19), demographic variables (20), and the method of symptom measurement (21, 22), have been implicated in placebo responding. The present study sought to examine the effects of placebos on symptom perception from a multicomponent perspective by testing for the effects of two factors, expectancy and motivation, on measures of subjective symptom change.

From the Department of Rehabilitation, University of Washington, Seattle, Washington (M.P.J.) and the Department of Psychology, Arizona State University. Tempe, Arizona (P.K.). Address reprint requests to: Mark Jensen, Ph.D., Department of Rehabilitation Medicine, RJ-30. University of Washington School of Medicine, Seattle. WA 98195. Received for publication October 1, 1990; revision received January 3, 1991

Expectancy and Placebo Responding The most common explanation for placebo effects is that placebos alter peoples' expectations. Over 1700 years ago, Galen wrote about the physician that "He cures most in whom most are confident (quoted by Evans, 23)". In the present context, an expectancy may be defined as a belief that a symptom will occur. A great deal of support for the expectancy/belief hypothesis comes from double-blind drug outcome studies. Five trends in these studies,
Psychosomatic Medicine 53:144-152 (1991)

144
0033-3174/91/5302-0144$03.00/0 Copyright 1991 by Ihe American Psychoson ntic Society

MOTIVATION AND EXPECTANCY

summarized by Ross and Olson (19), reveal that: (a) the direction of placebo effects parallels the effects of the drug under study, (bj the strength of the placebo effect is proportional to that of the active drug, (c) the reported side effects of the placebo and drug are often similar, (d) the timeeffect curves (the time it takes for the drug to become maximally effective and to decrease in effectiveness) for placebos are similar to those of comparison drugs, and (e) placebos and drugs have similar dosage effects. Ross and Olson conclude that: . . . the five patterns in the data from pharmacological studies illustrate the importance of subjects' expectancies in placebo effects. Most studies find that an administered placebo will alter the recipient's condition (or, in some instances, self-reports of the condition) in accordance with the placebo's expected effect (p. 419). Additional support for the expectancy hypothesis is found in those experiments that assess the effects of placebic instructions upon the perception of symptom change. In two such experiments, subjects given placebos reputed to be analgesics reported less pain than no-treatment control subjects (8, 24). In another study, Gryll and Katahn (25) found that, out of several experimental manipulations, the content of the placebo instructions had the most powerful effect on pain and anxiety among dental patients. Finally, Frankenhaeuser et al. (10) were able to significantly influence the perception of sedation by altering the instructions given with placebo capsules. Although most research supports the expectancy hypothesis, a few studies have produced results that are inconsistPsychosomatic Medicine 53:144-152 (1991)

ent with it. Penick and Fisher (26), for example, reported that placebo responding (as measured by a symptom checklist) was not significantly affected by instructions which described the placebos as stimulants and sedatives, although the average response under the two conditions was in the direction predicted by the expectancy hypothesis. Penick and Fisher suggested that significant effects were not found because the subjects were medical students (and therefore biologically sophisticated) and because the expectancy manipulation was relatively weak. In another study, Shipman et al. (27) examined the effects of analgesic placebos administered with an "enthusiastic" endorsement versus a "normal" endorsement. The subjects in these studies were classified on the basis of psychological tests and an interview with a clinical psychologist as either "normal" (people relatively free of emotional problems), "hypernormal" (physiologically tense and "psychologically rigid" people), or "psychoneurotic" ("emotionally troubled" people). The hypernormal and psychoneurotic subjects responded more to the placebos given an enthusiastic endorsement, as would be predicted by the expectancy hypothesis. However, the socalled normal subjects responded best to the placebo given a normal endorsement. Shipman et al. (27) explained their findings in a manner consistent with the expectancy hypothesis. That is, they argued that the "normals" may have been more sensitive to the "hard sell" techniques used in the enthusiastic endorsement, so that this form of endorsement may have created a reactive decrease in expectancy for these subjects. In sum, it is clear that the bulk of experimental studies using placebos support the expectancy hypothesis. Researchers
145

M. P. JENSEN AND P. KAROLY

who have obtained data inconsistent with the expectancy hypothesis have suggested that the manipulations did not engage the subjects' expectancy either in the indicated direction or to the predicted degree. Although this interpretation makes sense in light of the strong support for the expectancy hypothesis, the manipulation checks, which could have been performed to lend empirical support for such explanations, were not included in these studies.

Motivation and Placebo Responding The second variable to be explored in this study, as it relates to placebo responding, is motivation. For the purposes of this experiment, motivation refers to the degree to which subjects desire to experience a symptom change. Motivation to experience certain symptoms (or symptom relief) could derive from a number of sources including societal (in the form of compensation for disability, cf. 28), social (in the form of familial support for displaying illness behavior, cf. 28, 29), or individual (having a personal need met by experiencing a symptom, cf. 12). There has been only one experimental test of the motivational hypothesis for symptom perception placebo responding. Totman (30) provided two experimental groups with an analgesic "injection" that was actually just a slight pin prick. Then, the members of one group were told that they would be paid for their participation in the experiment, and those of the other were told that they would not be paid. Cognitive dissonance theory (31) would predict that the subjects who were not expecting to be paid for their participation would have a greater motivation to respond to the placebo, since a response
146

would provide them with a justification for experiencing the discomfort of the injection. If motivation is related to placebo responding, these subjects should then display a greater analgesic effect. As predicted, those subjects told that they would not be paid displayed the greatest analgesic effect. Unfortunately, Totman did not provide a check to determine if the manipulation was effective in altering the subjects' perceived motivation to respond to the placebo. The current study sought to extend our knowledge regarding the roles of expectancy and motivation on placebo responding within a single experiment. Both factors were manipulated experimentally, and the effects of these manipulations on placebo responding were assessed. To the degree that the manipulations were effectively instituted, it was hypothesized that subjects would respond more when they are led to believe they are receiving a high dose of a drug than when they are led to believe they are receiving a low dose, and that motivated subjects would respond more than those who are relatively unmotivated.

METHOD Subjects
The subjects for this study consisted of 86 (39 males and 47 females) individuals recruited from a pool of introductory psychology students at a large southwestern university.

Procedure
Motivation Manipulation. Subjects were randomly assigned to one of two conditions. Subjects in the high motivation condition were asked to read instructions designed to increase their desire to respond to the "sedative" placebo pill. These instruc-

Psychosomatic Medicine 53:144-152 (1991)

MOTIVATION AND EXPECTANCY

tions described the characteristics of one kind of personality ("Type B") in a positive way. and another kind of personality ("Type A") in a negative way. A fictional rationale was then provided for a strong relationship between response to the pill and personality type. A response to the pill, it was explained, would indicate that the person is a Type B. and no response that the person is a Type A. Subjects should thus be motivated to respond to the placebo under this condition because being a Type B was described as desirable. Subjects in the low motivation condition read similar instructions as did those in the high motivation group, except that the relationship between personality type and response to the pill was described as "very weak." Under this condition, subjects were led to believe that response to the pill has little, if any, bearing on their personality. Subjects were not told whether they were Type A or Type B. Expectancy Manipulation. The expectancy manipulation was crossed within the groups factor. All subjects received a high dose and a low dose manipulation so that they could act as their own controls, thereby minimizing chance variance between the expectancy conditions. In order to control for order effects, half of the subjects were assigned to receive the high expectancy condition during the first session (Day 1) and the low expectancy condition during the second session (Day 2), and half were assigned to receive the expectancy conditions in reverse order. Within the high expectancy condition, the placebo was described as a high dose of a common over-the-counter sedative. The low expectancy condition consisted of a description of the placebo as a low dose of the same drug. The "drug" descriptions were followed by preplacebo assessment of the subjects' level of sedation. This pretest consisted of an adjective checklist containing 38 adjectives that have demonstrated their ability to discriminate between sedatives, stimulants, and placebos in a previous study (32). Sedation level was scored in accordance with the method outlined by Brodeur (33). Specifically, checking each adjective associated with stimulants received a score of +1. and checking each adjective associated with depressants received a score of 1. The sum of these scores was used for the base rate of sedation. As a manipulation check, the subjects were then asked to indicate their "motivation" (i.e., desire to respond), and "expectancy" (i.e., how they believed they actually would respond), on two 11-point scales, with = "I want [or expect] to have many 5 sedative responses" and 5 = "I want [or expect] to have many stimulant responses." Subjects were then asked to ingest a placebo.

RESULTS Manipulation Checks Expectancy A four factor mixed design analysis of variance (Sex x Motivation X Order x Expectancy) performed on the expectancy ratings assessed before ingestion of the placebo yielded three significant findings: a main effect for expectancy (F (1,78] = 8.12, p < 0.01), an Expectancy X Order interaction (F(l,78) = 14.91, p < 0.001),
TABLE 1. Mean Ratings of Expectancy. Collapsing over Motivation, for Sex, Order, and Day Expectancy
Croup Low Dose M Males Order Order Females Order Order 1 2 1 2 20 19 24 23 -0.45 -1.00 -0.17 -2.04 SD 0.88 2.24 0.87 1.43 High Dose M -1.35 -1.26 -1.75 -1.43 SD 1.92 2.08 1.87 1.41

Note: Expectancy rating has a possible range of +5 (indicating a strong expectation to experience stimulation) to 5 (indicating a strong expectation to experience sedation). Under Order 1, the high expectancy manipulation was given first and the low expectation manipulation given second. The opposite order was administered in the Order 2 group.

Psychosomatic Medicine 53:144-152 (1991)

147

M. P. JENSEN A N D P. KAROLY

and an Expectancy x Sex x Order interaction (F(l,78) = 4.06, p < 0.05). To interpret the significant higher order interactions, the data were analyzed systematically according to the pattern outlined by Keppel (34, pp. 303-315). These analyses showed that the expectancy manipulation primarily affected females, with the high dose manipulation producing a significantly greater expectancy when given first (F(l,23) = 28.93, p < 0.001), and producing a lower expectancy rating when given second (F(l,22) = 5.96, p < 0.05; see Table 1).

x Expectancy) was performed using the ratings of general response and adjective checklist measures as the dependent variables and the preplacebo adjective checklist measure as the covariate. These analyses revealed several significant outcomes: a significant main effect for motivation (multivariate F = 3.72, p < 0.05), a significant main effect for time (multivariate F = 4.25, p < 0.01), and significant Sex X Time (multivariate F = 2.49, p < 0.05), Sex X Order x Time (multivariate F = 2.67, p < 0.05), and Sex X

Motivation A four-factor mixed design analysis of variance (Sex X Motivation x Order X Expectancy) performed on the motivation to respond measure produced only one significant effecta significant main effect for motivation (F(l,78) = 4.25, p < 0.05). From the summary table of means (Table 2), it can be seen that subjects in the high motivation condition rated themselves as having a greater desire to respond to the sedative than subjects in the low (neutral) motivation condition. These results indicate that the motivation manipulation was effective in changing the subjects' reported desire to respond to the placebo.

TABLE 2. Mean Ratings of Motivation, Collapsing over Sex, Order, and Day
Croup High motivation Neutral motivation 78 94 -0.92 -0.12 SD 2.43 1.61

Note: Motivation rating has a possible range of +5 (indicating a strong desire to experience stimulation) to 5 (indicating a strong desire to experience sedation).

0.0-0.2-0.4-0.6-0.8-1.0-

Neutral Motivation

The Influence of the Manipulations on Measures of Sedation: Placebo Effects Omnibus Multivariate Test An omnibus five-factor mixed design analysis of covariance (Sex X Motivation X Order X Time (measurement occasion)
148

TIME SINCE INGESTION (In minutes)

Fig. 1. Mean general response ratings following ingestion of the "sedative" placebo at three points in time. General response has a possible range of +5 (indicating a strong stimulant response) to (indicating a strong 5 sedative response).

Psychosomatic Medicine 53:144-152 (1991)

MOTIVATION AND EXPECTANCY

Motivation X Order X Time (multivariate tivation condition, and (b) the measures of symptom change increased over time F = 3.06, p < 0.05] interactions. (within sessions) for both motivation groups. Adjective Checkiist Responses. A fiveUnivariate Tests factor mixed design analysis of covariance Because the omnibus test yielded sig- (Sex x Motivation x Order x Time x nificant findings, separate univariate Expectancy) on the adjective checklist analyses were performed in order to ex- measures of sedation using the preplacebo plore the impact of the manipulations on adjective checklist score as the covariate revealed two significant results: a main the individual dependent measures. Genera] Response Ratings. A five-factor effect for time (multivariate F = 3.35, p = mixed design analysis of variance (Sex X 0.05) and a main effect for motivation Motivation x Order x Time x Expect- (F(l,78) = 5.33, p < 0.05). To illustrate ancy) performed on the rating of general these effects, change scores were comresponse yielded a significant main effect puted for every subject for each time for time (multivariate F = 7.85, p < 0.001), period, and the average of these change and a significant main effect for motiva- scores were plotted over time (Figure 2). tion (F(l,78) = 4.88, p < 0.05). These main As can be seen, a larger symptom change effects are reflected in Figure 1, which occurred among the high motivation subshows that (a) subjects in the high moti- jects, and a general increase in the pervation condition reported more symptom ception of sedation occurred over time change than subjects in the neutral mo- (within the session) for both motivation groups.

0.0-0.4-0.8-1.2-i.a-2.0-

Nautral Motlvatlo High Motivation

DISCUSSION

TIME SINCE INGESTION (In minutes) Fig. 2. Mean change in adjective checklist measure of sedation following ingestion of the "sedative" placebo at three points in time. Change in adjective checklist has a possible range of +38 (indicating a strong stimulant response) to 38 (indicating a strong sedative response).

The results of the present study provide clear support for a motivational view of placebo responding. Our findings suggest that subjects who are encouraged to have a stronger desire to respond to a placebo sedative tend to report greater sedation than those subjects who are not equally motivated to respond. This finding extends the work of Totman (30) from placebo analgesia to placebo sedation. While the expectancy induction was apparently successful, there was no overall main effect of this manipulation on the measures of placebo responding. Our analyses of the effect of the expectancy manipulation suggests that it was less ef149

Psychosomatic Medicine 53:144-152 (1991)

M. P. JENSEN AND P. KAROLY

fective on the second day. A possible explanation of this unexpected finding is consistent with the expectancy hypothesis. That is, people use their experience with drugs as a data source for developing expectations about their future effects. In this experiment, having experienced the effects of the "sedative" placebo in the first session, the subjects may have been less likely to believe it was a powerful drug in the second session. Research is now needed to examine the manner in which motivation and expectancy influence the perception of symptoms in clinical settings. For example, it is unlikely that these two factors will influence placebo responding to the same degree or with equal potency across all occasions or clinical groups. The potential impact of population type on placebo responding can be illustrated by a comparison of individuals with chronic and laboratory pain. On one hand, chronic pain patients are described as being influenced by a number of motivating factors such as social reinforcement and financial compensation for pain behavior (28). In fact, the operant approach to pain treatment attempts to treat pain primarily through the assessment and control of such motivational factors (28). On the other hand, subjects involved in laboratory pain studies may have less investment in the outcome of treatment, so that motivational factors may play less of a role in such acute pain contexts. It is interesting to note that a large difference has been found between placebo responding of clinical pain patients and laboratory pain patients. Beecher (35), in a review of 23 studies, found an average of 34.6% placebo response in the former and 3.2% placebo response in the latter group of subjects. He concluded that placebos are more effective when greater levels of
150

stress are involved. The motivational hypothesis provides a compatible alternative explanation for this finding. If the findings of this study extend to other symptoms and situations, then they have important implications for the understanding and delivery of medical and psychological services. In order to test for the effects of treatments over and above those produced by a "placebo," control interventions would need to match "active" therapy not only in terms of procedural or outcome credibility (cf. 36), but also in terms of the degree of motivationfor-change engendered in the therapeutic context (over time). The present study has several limitations which should be made explicit. First, the sample was recruited from a group of healthy introductory psychology studentsa group which may differ from many clinical populations in a number of important ways (e.g., age, health status, social economic status). Thus, the results may not generalize to clinical populations. There is a need to extend this research in order to determine the populations and symptoms which are most strongly influenced by motivational factors. Also, although a strength of the current design was the inclusion of manipulation checks, their inclusion may also have inadvertently influenced the results by encouraging the subjects to "report" those symptoms which they had previously indicated an expectancy or motivation to experience. Future investigations would do well to include groups of subjects who do and do not receive manipulation check measures. Finally, the reader will recall that the subjects in this study had relatively little time to report on their experience following placebo ingestion (they had four minutes to reflect upon and report their symptoms on three
Psychosomatic Medicine 53:144-152 (1991)

MOTIVATION AND EXPECTANCY

occasions). This limited amount of time may not have been sufficient for them to reflect upon and accurately report their "true" experience. It would be interesting, and instructive, to vary the amount of time subjects are allowed to "listen to" their body to see if this influences the level of placebo responding. To the extent that any active treatment has a placebo component, and there is a consensus that nearly all active treatments are susceptible to placebo factors (e.g., 1, 11, 13, 37-39), it seems reasonable to suggest that practitioners seek to maximize treatment effectiveness by providing the optimal mix of expectancy and motivation to respond. Of course, the difficulty will be to choose the "optimal"

levels. We do not yet know whether the relationships between expectancy, motivation, and response to treatment are linear or curvilinear (recall the Shipman et al. (27) study in which it was shown that a very strong expectancy was associated with a decrease in treatment response for one population of subjects). Research examining the impact of motivation and expectancies on various treatments is now indicated in order to discern how this information may be used to be of most assistance to individuals experiencing distressing symptoms. We wish to thank Drs. Anne Harris and Sandy Braver for their advice and assistance.

REFERENCES
1. Modell W: The Relief of Symptoms. Philadelphia, Saunders, 1955 2. Shapiro AK: A contribution to the history of the placebo effect. Behav Sci 5:109-135, I960 3. Gibbins FX, Hormuth SE: Motivational factors in placebo responsivity. Psychopharmacol Bull 17:77-79. 1981 4. Klopfer B: Psychological variables in human cancer. J Proj Tech 21:331-340, 1957 5. McFadden ER Jr, Luparello T, Lyons HA, Bleecker E: The mechanism of action of suggestion in the induction of acute asthma attacks. Psychosom Med 31:134-143, 1969 6. Sternbach RA: The effects of instructional sets on autonomic responsivity. Psychophysiology 1:67-72. 1964 7. WolfS, Pinsky RH: Effects of placebo administration and occurrence of toxic reactions. JAMA 155:339341,1954 8. Liberman R: An experimental study of the placebo phenomenon under three different situations of pain. J Psychiatr Res 2:233-246, 1964 9. Honigfeld G: Physician and patient attitudes as factors influencing the placebo response in depression. Dis Nerv Syst 24:343-347, 1963 10. Frankenhaeuser M, Jarpe G, Svan H, Wrangsjo B: Psychophysiological reactions to two different placebo treatments. Scand J Psycho] 4:245-250, 1963 11. Gallimore RG, Turner JL: Contemporary studies of the placebo phenomenon. In ME Jarvik (ed), Psychopharmacology in the Practice of Medicine. New York, Appleton-Century-Crofts, 1977, 47-57 12. Jensen MP, Karoly, P: Control theory and multiple placebo effects. Int J Psychiatry Med 15.137-147, 1985 13. Shapiro AK, Morris LA: The placebo effect in medical and psychological therapies. In Garfield SL, Bergin AE (eds), Handbook of Psychotherapy and Behavior Change (2nd ed). New York, Wiley, 1978, 369-410 14. Evans FJ: The placebo response in pain reduction. Adv Neurol 4:289-296, 1974a

Psychosomatic Medicine 53:144-152 (1991)

151

M. P. JENSEN AND P. KAROLY 15. )ones RA: Self-Fulfilling Prophesies: Social Psychological and Physiological Effects of Expectancies. Hillsdale, NJ, Erlbaum, 1977, 204-238 16. Peek C): A critical look at the theory of placebo. Biofeedback Self-Regul 2:327-335, 1977 17. Ross M, Olson JM: An expectancy-attribution model of the effects of placebos. Psychol Rev 88:408-437, 1981 18. Beecher HK: The powerful placebo. JAMA 159:1602-1606, 1955 19. Donovan TR, Gershman L: Experimental anxiety reduction: Systematic desensitization versus a falsefeedback expectancy manipulation. J Behav Ther Exp Psychiatry 10.173-179. 1979 20. Sato TL, Turnbull CD, Davidson JRT, Madakasira S: Depressive illness and placebo response, lnt ] Psychiatry Med 14:171-179, 1984 21. Evans FJ: Placebo response: Relationship to suggestibility and hypnotizability. Proc Ann Conv Am Psychol Assoc 77:889-890, 1969 22. Nash MM, Zimring FM: Prediction of reaction to placebo. J Abnor Psychol 74:568-573, 1969 23. Evans FJ: The power of a sugar pill. Psychol Today 7:54-59, 1974b 24. Gelfand S, Ullman LP, Krasner L: The placebo response: An experimental approach. J Nerv Ment Dis 136:379-387, 1963 25. Gryll SL, Katahn M: Situational factors contributing to the placebo effect. Psychopharmacology 57:253261, 1978 26. Penick SB, Fisher S: Drug set interaction: Psychological and physiological effects of epinephrine under differential expectations. Psychosom Med 27:177-182, 1965 27. Shipman WG, Green CS, Laskin DM: Correlation of placebo responses and personality characteristics of myofacial pain-dysfunction (MPD) patients. J Psychosom Res 18:475-483, 1974 28. Fordyce WE: Behavior Methods for Chronic Pain and Illness. St Louis, Mosby, 1976 29. Minuchin S: Families and family therapy. Cambridge, MA, Harvard University press, 1974 30. Totman R: Cognitive dissonance and the placebo response: The effect of differential justification for undergoing dummy injections. Eur J Soc Psychol 5:441-456, 1975 31. Festinger L: A theory of cognitive dissonance. Evanston, Row, 1957 32. Smith GM, Beecher HK: Amphetamine secobarbital and athletic performance: II. Subjective evaluations of performances, mood states, and physical states. JAMA 172:1502-1514, 1960 33. Brodeur DW: The effects of stimulant and tranquilizer placebos on healthy subjects in a real-life situation. Psychopharmacologia 7:444-452, 1965 34. Keppel G: Design and analysis: A researchers handbook (2nd ed). Englewood Cliffs, New Jersey, PrenticeHall, 1982 35. Beecher HK: Increased stress and effectiveness of placebos and "active" drugs. Science 132:91-92, 1960 36. Lick J, Bootzin R: Expectancy factors in the treatment of fear: Methodological and theoretical issues. Psychol Bull 82:917-931, 1975 37. Benson H, Epstein MD: The placebo effect: A neglected asset in the care of patients. JAMA 232:12251227, 1975 38. Benson H, McCallie DP: Angina pectoris and the placebo effect. N Engl J Med 300:1424-1429, 1979
39. Lasagna L: Placebos. Sci A m 193:68-71, 1955

152

Psychosomatic Medicine 53:144-152 (1991)