The placebo effect has long been something of a mystery. The response of researchers has broadly been of two kinds: either to dismiss it as an artefact; or to view it as an important, but unknown, component in treatment. The thesis of this article is that the latter view is supported by the weight of evidence. After describing and evaluating a range of evidence relating to the placebo effect, it is concluded that placebos are active and potent elements in a complete theory of healing. Central to the effect is the interaction of the patient's and the clinician's beliefs in the efficacy of the process of healing. This permits a redefiniton of the placebo effect in terms of the therapeutic alliance and a shift of emphasis away from the passive treatment of susceptible patients by clinicians to the mutuality of what is termed a joint 'credebo' effect.

Placebo ('I shall please') is defined as:

. . . any therapeutic procedure (or that component of any therapeutic procedure) which is given deliberately to have an effect, or unknowingly has an effect on a patient, symptom, syndrome, or disease, but is objectively without specific activity for the condition being treated.( n1)

The above definition is widely cited in the literature and expresses a common understanding of the nature of the placebo effect. It will be argued below that this definition is ambiguous and that the ambiguity reflects alternative and incompatible views of the placebo effect. Nevertheless and despite this, it provides a good starting point for discussion.

Traditionally, placebos have been used in drug studies to control for the effects of expectancy in treatment. They are also used in psychological therapy research as a means of controlling for socalled 'nonspecific' factors in the therapy. In the following discussion, the role of placebos in phar-macological and psychological research will be reviewed. The aim will be to examine the nature of placebos and some of the explanations that have been proposed for their effects. It will be argued that placebos are a missing link in theories of the healing process. Their status in the theory of healing has changed over the years from a nuisance, through a mystery, to an important ingredient in the process of treatment. As the mechanisms of placebo effects become increasingly articulated, so it is apparent that the wider understanding and application of the principles of these effects have the potential to enhance health care generally.

One of the most remarkable characteristics of placebos is how potent they appear to be. Oh( n2) suggests that they work best with pain conditions, disorders of the autonomic nervous system, and processes under neurohumoral control. Ross and Olsen( n3) list evidence of their efficacy in the 'treatment of radiation sickness, dental pain, headaches, coughing, asthmas, postoperative pain, multiple sclerosis, the common cold, diabetes, ulcers, arthritis, sea sickness, parkinsonism, and much more' (p. 409). Indeed, such ubiquitous effects have prompted O'Donnell( n4) to make the rather ironical comment that placebo is 'the most effective medication known to science, subjected to more clinical trials than any other medicament yet [it] nearly always does better than anticipated. The range of susceptible conditions appears to be limitless.'

Ross and Olsen( n3) distinguish five trends in the literature on the pharmacological placebo effect. The first is direction of effect. Placebos tend to work in the same therapeutic direction as the active treatment with which they are being compared. Thus, a placebo, compared with a drug that lowers blood pressure, will also have a pressure-lowering effect. In the same way, if the active treatment produces an increase in blood pressure, then the placebo will have the same direction of effect.

Secondly, there is equivalence of strength. If a placebo is compared with a strong analgesic, such as morphine, then it will replicate the strong analgesia. If the placebo is compared with a mild analgesic such as aspirin, then the magnitude of analgesia will be comparable with that provided by the aspirin.

Thirdly, placebo side-effects are similar to the side-effects of the active drugs under analysis. Any active drug is likely to have known side-effects (e.g. stomach discomfort, nausea, headaches, effects on appetite or on sexual performance, etc.). Placebos have been found to produce broadly similar effects.

Fourthly, the time course of the placebo mimics the time course of the effects of the active drug. If the effects of the active drug last for a number of minutes or hours, the placebo will tend to follow the same pattern.

Finally, the 'therapeutic window' of placebos seems to be the same as the active drug under comparison. Any active drug requires a specific dosage to provide an optimal concentration of the drug in the blood system. There is a level under which the drug is ineffective and above which it is redundant. The therapeutic window is somewhere in between these extremes and differs from drug to drug. Placebos appear to follow the same pattern as the active drug in question.

The extent of the placebo effect can be estimated in two ways. One is in relation to the percentage of patients who will be responsive to placebo treatment. The other is in terms of the degree of relief of symptoms. Regarding the first, much of the early work on placebos was under-taken in the context of pain relief. Thus, the commonly quoted rate of the placebo effect is generally said to be around 35%, meaning that about 35% of patients will respond to placebo treatment. This is derived from Beecher's( n5) seminal study, which investigated the placebo response in a variety of pain-inducing medical conditions. Of the 1082 patients studied, between 15% and 58% (mean 35.2) had their symptoms 'satisfactorily relieved' by a placebo.

In terms of the degree of relief, Evans( n6) reviewed a number of studies comparing a placebo with active pain relief and found that an average of 36% of patients claimed at least 50% relief from pain after placebo administration. This represents approximately 50-60% of the degree of relief obtained from strong analgesia, such as morphine.

Around 30-40%, therefore, is the generally quoted rate for the proportion of patients likely to respond to placebo treatment; in these patients there is generally found to be something of the order of 50-60% symptom relief. However, there are a number of caveats. First of all, the percentages must be treated with caution because the scales used to measure placebo effects are not always well conceived. For example, in measuring pain relief from active and placebo treatments, a common technique is to use four equally spaced and weighted categories, where 1 = no relief, 2 = slight relief, 3 = moderate relief and 4 = complete relief. As Tursky( n7) points out, both of the extreme categories are highly unlikely to apply: 'almost any treatment (real or imagined) will produce at least some indication of slight relief, and complete pain relief is almost never achieved by analgesics' (p. 232). This leaves just the two middle categories and these are not good discriminators of the degree of pain relief. It should not be considered surprising if a larger number of patients who are receiving the active treatment score 3. In addition, bearing in mind the nature of placebos, a notable number of patients receiving the placebo treatment will also score 3. The significance of these findings, however, may just be an artefact of the scoring procedure.

It must also be noted that average figures mask the fact that some patients report relatively greater relief and some relatively little. There are, therefore, individual patient factors that must also be taken into consideration in interpreting such data. It is also interesting to note that there does not appear to be a 'placebo-prone' personality. Patients who are responsive to placebo on one occasion may not be on another. Furthermore, an individual who has been unresponsive initially may yet be responsive subsequently. In addition, few placebo studies include a no-treatment condition that would control for spontaneous remission and regression to the mean, so that some of the apparent relief may have been due to nonplacebo factors.( n8, n9) This variability has led Shapiro and Shapiro( n10) to conclude that 'no reliable and valid estimate of placebo reactivity is currently available' (p. 80).

However, the difficulty in attaching a precise figure to the extent of the phenomenon should not detract from the fact that a potent effect exists.( n10) The conclusion, therefore, must be that, despite the above reservations, it seems clear that not all instances of the placebo effect can be discounted as an artefact of deficient measuring procedures or a statistical anomaly. A powerful effect exists; the question is when, how and to whom does it occur? An under-standing of the placebo effect is essential to the evaluation of medical treatments and is particularly relevant to the investigation of psychological approaches to treatment: both, 'pure' psychological treatments and psychological aspects of medical treatments. It is important, therefore, to define the phenomenon as clearly as possible and identify the underlying mechanisms that are thought to be responsible for it.

Explanatory models of the placebo effect
Brody( n11) identifies three possibilities in explaining the effects of any medical or psychological treatment:

The patient got better due to the 'vis medicatrix naturae' (the 'natural history') of the condition in an organism with intact healing and therapeutic powers.
The patient got better due to the symbolic effect of the treatment --that is, its impact on his or her imagination, beliefs, and/or emotions.
The patient got better due to some 'specific' or 'characteristic' feature of the treatment that can be studied, isolated, and predicted within the context of contemporary medical theory (p. 42).
Each of these possibilities entails its own set of hypotheses.

The vis medicatrix naturae explanation attributes placebos to a naturally occurring propensity of the organism to heal itself. In more prosaic terms, this is the basis of 'spontaneous remission' and 'regression to the mean'. The natural way of controlling for this in experimental studies is to have a no-treatment group amongst whom 'natural history' factors will exert an influence in relation to which any additional effect of the treatment can be measured. With respect to placebos, however, even if they can be explained as due to a process of natural history, the question remains whether this process is fully independent of any external influence, or whether it can be triggered by certain characteristics of treatment (i.e. can placebo factors expedite natural history processes?). The placebo effect may be amenable to experimental control, but its mechanism is not fully explained by the vis medicatrix naturae hypothesis.

The 'symbolic' explanation is based upon the view that every medical treatment entails a symbolic effect, and some treatments may depend entirely upon this. This is to say that the patient upon whom any treatment is focused must inevitably have a perceptual, cognitive and affective response to the treatment. Thus, in addition to the 'active' ingredient, there is something within the patient that is stimulated by something outwith him or her (namely, the process of treatment). The something within is his or her imagi-nation, beliefs and emotions. These may be the portrayed as the 'incidental' artefacts of the active treatment or the symbolic effects of a sham treatment (e.g. a sugar pill). In both cases, it is appropriate to provide an 'attention' control that has the same characteristics as the active treatment in terms of time, attention and interaction with the clinician.

Finally, the 'specific' explanation is the feature of the treatment that is amenable to scientific scrutiny and is interpretable within a valid theoretical framework. With regard to placebos there is some debate about whether or not they belong in this category. This is more than a semantic quibble and is fundamental to whether placebos are dismissed as marginal epiphenomena or central ingredients of theory. This question will be discussed more fully below.

The central question raised by the above hypotheses is whether the theory of placebo effect should be relegated to those circumstances where there is no known active ingredient in the treatment, so that the explanation of response to treatment is entirely symbolic, or should the placebo level of explanation also be taken into account, even when a fully explicated active treatment is administered. In other words, does the symbolic element add anything to the theory of active treatment? For reasons that it is hoped will become apparent below, it will be argued that placebos do indeed form part of the complete theory of treatment. The neglect of a full understanding of placebo factors in, say, an active drug treatment, will lead to an incomplete theory of the action of the drug. The same principle applies to psychological treatments. It will be argued below that the specific and the symbolic explanations are not exclusive alternatives; rather, they are different aspects of the same explanation. To oppose specific with symbolic explanations is to fall into the trap of 'mind-body' dualism.

Placebos and the reduction of anxiety
The anxiety reduction hypothesis posits that placebos exert their influence solely by changing the affective response to the condition being treated. This is best illustrated in relation to pain. Pain is defined as 'an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage.'( n12) Pain can further be distinguished in terms of various dimensions, such as:

Sensation: what the pain feels like (e.g. 'burning', 'stabbing', 'tearing', etc.);
Intensity: how strong it is (e.g. visual analogue scale score);
Reactivity: how unpleasant it is (e.g. 'miserable', 'depressing', 'unbearable', etc.).( n13, n14)
Recently, these distinctions appear to be confirmed by developments in neural science, with the discovery of different neural pathways for sensory and affective pain respectively.( n15)

The broad distinction between sensory pain (sensation) and affective pain (intensity and reactivity) provides the basis for the anxiety-reduction hypothesis of placebo effect. One of the earliest writers on the subject, Beecher,( n16) proposed that placebos exert their effect on pain by reducing the emotional distress caused by the pain. Thus, the patient who responds positively to a sugar pill for pain, gains apparent relief by virtue of a diminution of the affective component of the pain. The sensory component may remain the same, but the overall experience of the pain, which comprises both sensory and affective dimensions, is reduced. It should be noted that, if the same explanation is extended to placebo 'cures' for medical complaints, then a further mechanism must be entailed, namely a means by which changes in affectivity (e.g. reduced stress) cause biological changes and thus affect health. The theory underpinning this possibility is reviewed in Evans et al.( n17) and Glaser and Kiercolt-Glaser.( n18)

What is true for pain is true for other conditions subjected to the placebo effect. For example, in a study of insomnia, Storms and Nisbett19 prescribed patients who were suffering from sleeplessness with one of two kinds of placebo pill. One was described as a 'pep' pill and the other as a 'sedative'. Somewhat paradoxically, the patients taking the placebo pep pills experienced reduced insomnia and the patients taking placebo sedatives experienced increased insomnia. The authors explain this by virtue of the fact that the insomniacs given 'pep' pills attributed their bedtime arousal to the effects of the pill rather than some internal emotional process. This had the effect of reducing their distress and worry about their own supposed psychological pathology, enabling them to become more calm more quickly and, consequently, sleep. In contrast, the patients given placebo sedatives interpreted their continued bedtime arousal as further evidence of their psychological pathology, which endured despite the apparent 'sedative'.

Conditioning and expectancy
The classical conditioning view of placebo analgesia is described by Wickramasekera,( n20) who suggests that the active ingredients of analgesics act as unconditioned stimuli for pain reduction. In addition, the panoply of stimuli accompanying the administration of analgesics (the hospital environment, smells, sounds; the sight of a stethoscope or white coat or syringe, etc.) constitute the conditioned stimuli. According to the rules of classical conditioning, the presence of the conditioned stimuli can elicit responses associated with the unconditioned stimuli.

As it stands, however, the classical conditioning approach is insufficient to explain the complexity of the human placebo response. Price and Fields( n21) develop the conditioning model and suggest that these principles can be extended to the human placebo response in which patients are conditioned to associate white coats and other paraphernalia of medicine with pain reduction (and presumably, in some cases, such as nausea associated with chemotherapy, with pain increase). This conditioning, however, is also modulated by other psychological factors such as motivation to seek relief (desire for treatment) and expectation of successful treatment.

The expectation of successful treatment has become a major emphasis in psychological explanations of the placebo effect. Expectancy subsumes the patient's assumptive world and the perceived meaning of events within that context. It is usually portrayed as a characteristic of the patient, but it can also include characteristics of the clinician. The expectancy hypothesis of the placebo effect is that expectations generate beliefs, and beliefs, it is argued, influence the course of healing. An illustration is provided by Ross and Olsen.( n3) They explain the five trends in the placebo literature, described earlier, as being a consequence of the implicit communications of clinicians involved in the administration of the placebo. Thus, for example, a clinician who is blind to whether an administered drug is active or a placebo is likely, nevertheless, to know the direction, strength, side-effects etc. of the active drug and communicate this implicitly to the patient, whether that patient is receiving the active or the placebo drug. This, induces a certain expectation in the patient and, they argue, accounts for the similarities described earlier between active and placebo treatments. It must also be noted that the expectations thus induced have been implanted by the behaviour of the clinician. The placebo effect, therefore, is a product of the interaction between the clinician's and the patient's attitude and beliefs.

Lundh( n22) takes this theme further in defining the placebo effect as consisting in the generation of placebo beliefs. The latter are 'beliefs of the form "This treatment is going to cure me" ' (pp. 138-39). These beliefs have their effect by counteracting the negative affective impact of illness and disease. Thus a person who is seriously ill may be depressed or anxious because of the consequences of the illness on his or her life. If circumstances can be developed whereby a belief in the possibility of cure can be engendered, then improvement will follow (the 'credebo' effect?). It can be seen, then, that the anxiety reduction hypothesis of placebo and the expectancy hypothesis are compatible. The expectation of cure is likely to engender confidence and thus counter negative affective states such as depression and anxiety.

The neurobiology of the placebo effect
The work of Levine et al.( n23-n25) has stimulated interest in the possible biological substrate to the placebo effect. The basic thesis of this approach is that endogenous opioid peptides and opioid receptors modulate the transmission of pain signals to the brain. It is known that exogenous opioids exert their effects via the same pain modulatory system. In addition, it can be shown that opioid antagonists, such as naloxone, block the action of endogenous and exogenous opioids and produce hyperalgesia.

In relation to placebos, Levine et al.( n23) reported that patients who had responded to a placebo intervention subsequently showed hyperalgesia when administered naloxone. In other words, an opioid antagonist had reversedthe analgesic effects of the placebo. It was concluded from this that the placebo analgesia must have obtained its effect via the endogenous opioid receptors.

However, the opioid receptor hypothesis remains controversial.( n10, n26, n27) For example, Shapiro and Shapiro( n10) point out that there are many other forms of placebo effects than analgesia. Furthermore, even within the sphere of analgesia, the mechanisms posited by Levine et al. may explain some forms of analgesia but not all. For example, do the same biological mechanisms account for both sensory and emotional aspects of pain relief?

An empirical demonstration of the limitations of the opioid receptor hypothesis is provided by Montgomery and Kirsch.( n28) They applied a placebo local anaesthetic to one part of the body of volunteers and then applied pain stimuli simultaneously to treated and untreated parts of the same person's body. The opiate receptor hypothesis predicts global changes in analgesia. However, in this case there was found to be selective anaesthesia, reflecting differences between the placebo-anaesthetized and nonanaesthetized parts of the body. Montgomery and Kirsch explain their findings in terms of the well-attested tendency for placebos to mimic the active treatment in question. As mentioned above,( n3) this tendency has been attributed to expectancy in both the clinician and the person receiving the placebo.

Redefining the placebo effect
It will be recalled that the much-cited definition with which this article began referred to the placebo effect as 'objectively without specific activity'.( n1) Unfortunately, this terminology is ambiguous and the ambiguity is apparent in the literature. The term 'without specific activity' has at least two senses in this context. On one hand, 'without specific activity' is often taken to mean inactive and, on the other hand, it is used to refer to general, but not inactive, ingredients of treatment. The 'inactive' interpretation of a placebo effect underlies the often expressed view that this effect is an epiphenomenon and an inconvenient and illegitimate distraction from the 'real' effects of treatment: 'Although the placebo response in medicine has been widely recognized . . . it has generally been regarded as a nuisance variable'( n26); and: 'They (placebos) were first ignored, then treated as contaminants to be controlled, and finally investigated as variables of interest in their own right'.( n29) In addition, Frank( n30, n31) has pointed out that the assumption that a placebo is an inactive ingredient in therapy is mistaken; they are known to have substantial physiological as well as psychological effects.

In contrast, the 'general' interpretation of a placebo effect describes active features that a particular treatment shares with a range of other treatments. Thus, Critelli and Neumann( n32) define placebos as 'factors that are not specific to particular treatments, that is, factors that are common to most types of therapy'. These features may not yet be fully understood and articulated, but the possibility exists that at some point in the development of theory, they will. This view is advanced by Grunbaum,( n33) who suggests that the placebo effect may, in fact, be specific, but the nature of the effect may simply be poorly under-stood at present. Many forms of innovative treatments contain elements that have not yet been fully articulated owing to the undeveloped state of current theory. They may, nevertheless, have a specific remedial effect. To term these as non-specific, therefore, is to underestimate their potential importance. They may be unknown, but this does not justify calling them nonspecific.

Thus, two ostensibly conflicting views are expressed in the literature. On one hand, placebos are portrayed as 'incidental', 'inactive', 'artefactual' and essentially peripheral to the main theory (i.e. a 'nuisance' that confounds the 'real' treatment). On the other hand, they are portrayed as an active, potent, general, but poorly understood, phenomenon that, nevertheless, ultimately can be understood and incorporated within an expanded theory of the treatment in question.

Clearly, these views are contradictory. The question, therefore, is what evidence is there to support one view over the other? The starting point is to acknowledge that even the most 'powerful' active ingredient of a pharmacological or a psychological treatment cannot have its full effect without the placebo elements. Placebo ingredients are a precondition of effective treatment insofar as they define the conditions upon which successful treatment is founded. The proof of this, of course, would be if a clearly effective active treatment were given without the placebo elements; in this case, one would predict diminished efficacy. Precisely this line of reasoning underlies the finding that some clinicians are more effective as individuals than others in providing exactly the same active treatment.

This point is well illustrated by Freund et al.( n34) They undertook a placebo evaluation of the effects of a weight-reducing drug. The finding of interest for present purposes was that even the effects of the active drug varied according to the characteristics of the physician administering the treatment. Freund et al.'s conclusions are worth quoting at length:

The eight physicians participating in the study were randomly selected from a resident population. They were then 'conditioned' to relate to the patients in a predetermined manner. Nevertheless, even under these standardized conditions, the first week data demonstrated that the participating physicians were not uniformly capable of producing similar therapeutic results. This supports the long-held belief that physicians differ in the 'art of medicine' and in the quality of 'bedside manner' . . If either Doctor C or Doctor D had been the sole investigator . . . entirely different conclusions might have been reached as to the effectiveness of the test drug. Apart from the action of the drug and the patient's age, sex, education and social status, the effect of the physician's personality and the manner in which he related to the patients had a significant effect on therapeutic outcome (p. 177).

Similar conclusions can be drawn from other medical procedures. In the realm of surgery, for instance, the attitude of the surgeon can have a profound effect on outcome. Benson and McCallie( n35) provide a dramatic example of this with a procedure known as internal mammary ligation. This procedure was originally used in cardiac surgery as an active treatment for angina and involves re-routing the blood flow by tying one of the arteries. It was enthusiastically promoted as the (active) treatment of choice for angina and a theoretical rationale was presented in terms of improvements in the blood flow, which reduced ischaemia. Numerous operations were performed with a high rate of apparent success: '38 per cent found complete relief, and 65 to 75 per cent showed considerable improvement' (p. 1426).( n35) Later, however, more sceptical investigators performed the operation with a placebo control in which an anaesthetic was given and an incision in the chest wall was made, but without the ligation. This, it transpired, produced the same outcome. The conclusion was that the efficacy of internal mammary ligation derives from the enthusiasm of the doctor rather than the objective consequences of the operation.

There appears, then, to be a demonstrable effect relating to the attitude and behaviour of the clinician, in conjunction with the expectation of the patient, which can influence the potency and efficacy of both active and placebo treatments. This is an important addition to existing definitions of the placebo effect, which is usually described as something that happens to patients. However, it appears that the effect in patients cannot be separated from the attitudes and behaviour of clinicians.

Placebo effect and the therapeutic alliance
The importance of the attitude and behaviour of the clinician also has a long history in psychotherapy research. For example, Frank31 argues that the placebo effect:

. . . contains the necessary, and possibly sufficient, ingredient for much of the beneficial effect of all forms of psychotherapy. This is a helping person who listens to the patient's complaints and offers a procedure to relieve them, thereby inspiring the patient's hopes and combating demoralization (p. 291).

Indeed, Frank suggests that a significant proportion of patients undergoing psychotherapy are suffering solely from what he terms 'demoralization' and it is this that responds to the factors identified above. In the remaining subset of patients the nature of the problem is greater than can be encompassed by the notion of 'demoralization' and thus the same factors will not, in these cases, be effective as agents of treatment; more structured interventions are required.

Indeed, it may help to recast the notion of placebo factors into alternative terminology, which avoids some of the negative connotations noted above. One candidate term would be 'therapeutic alliance'.( n36) This term describes the relationship between therapist and client and the ingredients of this relationship. Horvath, et al.( n37) list three aspects of the therapeutic alliance:

The client's perception of the relevance and potency of the intervention offered.
The client's agreement with the therapist on reasonable and important expectations of the therapy in the short and medium term.
A cognitive and affective component, influenced by the client's ability to forge a personal bond with the therapist and the therapist's ability to present himself or herself as a caring, sensitive and sympathetic, helping figure.
It is interesting that the effect size (26%) attributed by Horvath et al.'s( n37) meta-analytical study to these characteristics is close to that attributed to the placebo effect in a broad range of other contexts (see above). It is also important to note that the therapist's belief in the efficacy of the intervention is taken as read. Thus the attitudes both of therapist and the client are positive; this is the basis for the effect.

While psychotherapeutic approaches frequently address the therapeutic alliance explicitly, more traditional medical treatments depend on the implicit presence of these characteristics. It nevertheless seems reasonable to suppose that the perceived authority of a medical doctor, the patient's expectation of benign help, and the power of the institution of medicine, generate comparable conditions for a workable therapeutic alliance in medical treatments.( n38, n39)

Synthesis and conclusions
Since the earliest reports, the placebo effect has been something of a mystery. The response to this has broadly been of two kinds: either to dismiss it as an artefact; or to view it as an important, but unknown, component of treatment. The thesis of this article is that the latter view is supported by the weight of evidence. The alternative view that the placebo effect is a mere artefact is based on a common confusion in the placebo literature between definitions of placebo as inactive, on the one hand, and general, but active, on the other. This ambiguity, in turn, has its origins in a further confusion between levels of the theory of what may be termed 'healing', a term that subsumes all endeavours in which a 'healer' (doctor, psychologist, psychotherapist, shaman, etc.) attempts to alleviate symptoms of disease or distress in a designated recipient of healing. One may distinguish a broad or macro-theory of healing from a narrow or microtheory of the components of healing. The microtheory of placebos is different from the microtheory of, say, a drug's action, and both are different dimensions of the macrotheory of healing. There may also be some overlap between microtheories (the overlapping areas shown in Figure 1). However, there is also a nonoverlapping and specific microtheory of the action of, say, analgesics as distinct from the specific theory of any other class of drugs such as anti-inflammatories or muscle-relaxants etc. This is illustrated in Figure 1.

The study of placebo effect is the study of this biopsychosocial interaction. Any theory of the placebo effect, therefore, must take account of:

The central nervous system (the 'bio'); The perceptual system and associated cognitions and emotions relating to the perceptions (the 'psycho');
The cultural context, which encompasses the individual's belief system and the social milieu (the 'social').
These dimensions are interconnected. An explanation of the mechanisms of placebo may involve any combination of these three components. Thus the placebo effect can be understood in relation to: the social world of the patient; the meaning attached to events by the patient; and the effects of these on the central nervous system. Any or all of these may be the prime focus of the explanation in a particular case. However, there is no sole explanation that accounts for all cases.

The placebo effect, then, can be redefined. Placebos are nonspecific in the sense of being general to a range of treatments, but they are nonetheless active and potent elements in the macrotheory of healing. Central to the effect is the interaction of the patient's and the clinician's beliefs in the efficacy of the process of healing. The patient must hold a belief in the potential of the treatment to help. In addition, however, the clinician must also share an equivalent belief and this must be expressed through his or her attitude and behaviour. The conjunction of these belief states gives rise to what might be termed a joint 'credebo' effect. We have thus moved from placebo to credebo: from the passive treatment of gullible patients by clinicians to the mutuality of the therapeutic alliance.

In conclusion, it has been argued that the placebo effect, far from being a peripheral arte-fact with only nuisance value, is a central component of the macrotheory of healing. There are a number of distinct components to the macrotheory, and there is also overlap between components. As the Gestalt psychologists used to say: the whole is greater than the sum of its parts. The whole, however, cannot exist without those parts and the theory of placebo/credebo is an integral part of the whole.

I am grateful to Lorraine Nanke for helpful comments on an earlier draft of this article.

DIAGRAM: Figure 1; The macrotheory of healing (outer circle) and the overlapping microtheories of each component of healing (inner circles)

(n1) Shapiro AK. Factors contributing to the placebo effect: their implications for psychotherapy. Am J Psychother 1961; 18: 73-88.

(n2) Oh VMS. The placebo effect: can we use it better? BMJ 1994; 309: 69-70.

(n3) Ross M, Olsen JM. An expectancy attribution model of the effects of placebos. Psychol Rev 1981; 88: 408-37.

(n4) O'Donnell M. Our oath is hypocritical. Monitor Weekly 1995; 1: 44.

(n5) Beecher HK. The powerful placebo. JAMA 1955; 176: 1102-107.

(n6) Evans FJ. The power of a sugar pill. Psychol Today 1974; (April): 55-59.

(n7) Tursky B. The 55% analgesic effect: real or artefact? In: White L, Tursky B, Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 229-34.

(n8) Turner JA, Deyo RA, Loeser JD, VonKorff M, Fordyce WE. The importance of placebo effects in pain treatment and research. JAMA 1994; 271: 1609-14

(n9) Ernst E, Resch KL. Concept of true and perceived placebo effects. BMJ 1995; 311: 551-53.

(n10) Shapiro AK, Shapiro E. The powerful placebo. Baltimore, MD: Johns Hopkins University Press, 1997.

(n11) Brody M. Placebo effect: an examination of Grunbaum's definition. In: White L, Tursky B, Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 37-58.

(n12) International Association for the Study of Pain; Merskey H (Chairperson). Pain terms: a list with definitions and notes on usage. Pain 1979; 6: 249-52.

(n13) Melzack R, Torgerson WS. On the language of pain. Anesthesiology 1971; 34: 50-59.

(n14) Tursky B. Development of a pain perception profile. In: Weisenberg M, and Tursky B eds. Pain: new perspectives in therapy and research. New York: Plenum, 1976: 171-94.

(n15) Rainville P, Duncan GH, Price DD, Carrier B, Bushnell D. Pain affect encoded in human anterior cingulate but not somatosensory cortex. Science 1997; 277: 969-71.

(n16) Beecher HK. Measurement of subjective responses: quantitative effect of drugs. New York: Oxford University Press, 1959.

(n17) Evans P, Crow A, Hucklebridge F. Stress and the immune system. Psychologist 1997; 10: 303-307.

(n18) Glaser R, Kiercolt-Glaser JK. Handbook of human stress and immunity. London: Academic Press, 1994.

(n19) Storms MD, Nisbett RE. Insomnia and the attribution process. J Pers Soc Psychol 1970; 16: 319-28.

(n20) Wickramasekera I. A conditioned response model of the placebo effect: predictions from the model. In: White L, Tursky B, Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 255-87.

(n21) Price D, Fields H. The contribution of desire and expectation to placebo analgesia: implications for new research strategies. In: Harrington A ed. The placebo effect: an interdisciplinary exploration. Cambridge, MA: Harvard University Press, 1997: 117-37.

(n22) Lundh LG. Placebo belief and health. A cognitive-emotional model. Scand J Psychol 1987; 28: 128-43.

(n23) Levine JD, Gordon NC, Field HL. The mechanism of placebo analgesia. Lancet 1978; ii: 654-57

(n24) Levine JD, Gordon NC, Field HL. Naloxone dose independently produces analgesia and hyperalgesia in postoperative pain. Nature 1979; 278: 740-41.

(n25) Levine JD, Lan W, Kwiat G, Goetzl EJ. Leukotriene B4 produces hyperalgesia that is dependent on polymorphonuclear leukocytes. Science 1984; 17: 743-45.

(n26) Evans FJ. Expectancy therapeutic instructions and the placebo response. In: White L, Tursky B, Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 215-28.

(n27) Grevert P, Goldstein A. Placebo analgesia, naloxone and the role of endogenous opioids. In: White L, Tursky B, Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 332-50.

(n28) Montgomery G, Kirsch I. Mechanisms of placebo pain reduction: an empirical investigation. Psychol Sci 1996; 7: 174-76.

(n29) Bootzin RR. The role of expectancy in behaviour change. In: White L, Tursky B and Schwartz GE eds. Placebo: theory research mechanisms. New York: Guilford, 1985: 196-214.

(n30) Frank JD. Therapeutic components of psychotherapy. J Nerv Ment Dis 1974; 159: 325-42.

(n31) Frank JD. The placebo is psychotherapy. Behav Brain Sci 1983; 6: 291-92.

(n32) Critelli JW, Neumann KF. The placebo. Am Psychol 1984; 39: 32-39.

(n33) Grunbaum A. The placebo concept in medicine and psychiatry. Psychol Med 1986; 16: 19-38.

(n34) Freund J, Krupp G, Goodenough D, Preston LW. The doctor-patient relationship and drug effect. Clin Pharmacol Ther 1971; 13: 172-80.

(n35) Benson H, McCallie DP. Angina pectoris and the placebo effect. N Engl J Med 1979; 300: 1424-29.

(n36) Luborsky L, Crits-Christoph P, Mintz J, Auerbach A. Who will benefit from psychotherapy? New York: Basic Books, 1988.

(n37) Horvath AO, Gaston L, Luborsky L. The therapeutic alliance and its measures. In: Miller NE, Luborsky L, Barber JP, Docherty JP eds. Psychotherapy treatment research: a handbook for clinical practice. New York: Basic Books, 1993. Cited in: Roth A, Fonagy P. What works for whom? A critical review of psychotherapy research. New York: Guilford, 1996: 351.

(n38) Balint M. The doctor, his patient and the illness. Tunbridge Wells: Pitman Medical, 1957.

(n39) Parsons T. The social system. New York: Free Press, 1951.

Address for correspondence: J. Mitchell Noon, Consultant in Health Psychology, Pain Management Unit, Treliske Hospital, Truro, Cornwall TR1 3RJ, UK.


By J Mitchell Noon, Pain Management Unit, Treliske Hospital, Truro, UK

Share this with your friends