Jump to ContentJump to Main Navigation
Placebo EffectsUnderstanding the mechanisms in health and disease$

Fabrizio Benedetti

Print publication date: 2008

Print ISBN-13: 9780199559121

Published to Oxford Scholarship Online: September 2009

DOI: 10.1093/acprof:oso/9780199559121.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 21 June 2018

A modern view of placebo and placebo-related effects

A modern view of placebo and placebo-related effects

(p.19) Chapter 2 A modern view of placebo and placebo-related effects
Placebo Effects

Fabrizio Benedetti

Oxford University Press

Abstract and Keywords

The placebo effect, or placebo response, is a psychobiological phenomenon that must not be confounded with other phenomena, such as spontaneous remission and statistical regression to the mean. The nocebo effect, or nocebo response, is a negative placebo effect, which goes in the opposite direction compared to the placebo effect. There are many placebo effects with different biological mechanisms and in different systems and apparatuses that are triggered by the psychosocial context around the patient and the therapy. Expectation of a future outcome plays a central role, and may act through different mechanisms, such as reduction in anxiety and activation of reward circuits. Learning plays a crucial role, and powerful placebo effects may be induced through a conditioning procedure. Placebo and placebo-related effects may be related to other self-regulatory processes and may have emerged during evolution as a defence mechanism of the body.

Keywords:   neurobiological mechanisms, spontaneous remission, regression to the mean, expectation, learning, reward mechanisms, anxiety, self-regulatory processes

  • By definition, the placebo effect is the effect that follows the administration of an inert treatment (the placebo), whereas in a placebo-related effect no placebo is given.

  • The placebo effect, or placebo response, is a psychobiological phenomenon that must not be confounded by other phenomena, such as spontaneous remission and statistical regression to the mean.

  • The nocebo effect, or nocebo response, is a negative placebo effect, which goes in the opposite direction of the placebo effect.

  • There are many placebo effects with different biological mechanisms and in different systems and apparatuses, which are triggered by the psychosocial context around the patient and the therapy.

  • Expectation of a future outcome plays a central role, and may act through different mechanisms, such as reduction in anxiety and activation of reward circuits.

  • Learning plays a crucial role, and powerful placebo effects may be induced through a conditioning procedure.

  • Modern interpretations of classical conditioning suggest that many conditioned placebo responses are consciously mediated by complex cognitive factors.

  • Placebo and placebo-related effects may be related to other self-regulatory processes and may have emerged during evolution as a defence mechanism of the body.


2.1 What they are not

2.1.1 Many phenomena are mistakenly taken for placebo effects

As already mentioned in Chapter 1, a first and important source of confusion about the placebo effect is in its definition. By looking through the online resource PubMed and inserting the word ‘placebo’, approximately 120,000 papers can be found, and the word placebo is often associated with the terms ‘effect’ or ‘response’. Most of these papers, particularly those involving clinical trials, use these words inappropriately or in a very broad and confusing sense. In fact, if we followed the definition of most clinical trial studies, the placebo response and/or effect would be: the outcome that is found in those groups of patients who receive a placebo (the inert treatment). It is very common to find titles of studies like ‘High rate of placebo responses in clinical trials on …’ or ‘Analysis of the placebo response in clinical trials on …’. Going through these papers, you soon realize that most of the studies analyze the time course of some symptoms in placebo groups. As a point of fact, the reduction of a symptom in the group that received placebo can be due to many causes that have nothing to do with the real placebo effect, or response (Kienle and Kiene 1997; Benedetti and Colloca 2004), as shown in Fig. 1.1 and described in detail in Table 2.1.

Spontaneous remission, habituation, patient and/or observer bias, the effects of unidentified co-interventions, and many other factors are all phenomena that are sometimes taken for placebo effects (Table 2.1). In other words, if a clinical improvement occurs in the group of patients who received the placebo, most studies describe this effect as ‘placebo effect’ or ‘placebo response’, regardless of whether the improvement is attributable to a spontaneous reduction of symptom severity or to a real psychobiological placebo effect in which the brain anticipates the benefit.

As this is the most common source of misunderstanding about the placebo effect, we should abandon the terms ‘placebo effect’ and ‘placebo response’ when describing the outcome of a placebo group in a clinical trial, and should replace them with words like ‘improvement in the group that received the placebo’. In this way, we make it clear that those patients who took the placebo did improve, although the cause of the improvement is unknown. It might be due to a spontaneous regression of a symptom, to a real active involvement of the brain in anticipating the outcome, or it might represent a biased report of the patients who wants to please the doctor. (p.21)

                   A modern view of placebo and placebo-related effects

Fig. 2.1 Graph (A) shows the natural fluctuation of pain in a given painful condition. (B) shows an ineffective therapy (dashed line) administered at the time indicated bythe arrow, compared to natural history (solid line). If it is not known that a remission in severity might occur without treatment, then any remission will be mistakenly attributed to that treatment. In (C) the natural history of pain intensity (i) is compared with that following placebo administration (ii) and active therapy (iii). The analgesia attributable to the active therapy is represented by the difference between (ii) and (iii), whereas the analgesia attributable to the placebo therapy is representedby the difference between (i) and (ii). Modified from Fields and Levine 1984, with permission from Elsevier, Copyright 1984.

2.1.2 Spontaneous remission is frequently and erroneously defined as placebo effect

Most chronic conditions show a spontaneous variation in symptom intensity that is known as natural history (Fields and Levine 1984). If someone takes a placebo just before his symptom starts decreasing, he may believe that the placebo is effective, although that decrease would have occurred anyway. Clearly, this is not a placebo effect but a misinterpretation of the cause–effect relationship, due to a spontaneous remission of the symptom.


Table 2.1 Factors that can cause the false impression of a placebo effect. From Kienle and Kiene 1997, with permission from Elsevier, Copyright 1997.

Natural course of a disease

Spontaneous improvement

Fluctuation of symptoms

Regression to the mean


Additional treatment

Observer bias

Conditional switching of treatment

Scaling bias

Poor definition of drug efficacy

Irrelevant response values

Subsiding toxic effect of previous medication

Patient bias

Answer of politentess and experimental subordination

Conditioned answers

Neurotic or psychotic misjudgement

No placebo given at all


Psychosomatic phenomena

Voodoo medicine

Uncritical reporting of anecdotes


False assumption of toxic placebo effects created by:

Everyday symptoms


Persistance of symptoms

Spontaneous fluctuations and remissions of symptoms are probably the most common source of confusion about placebo effects. For example, in his classic work on the power of placebos, Beecher (1955) was likely to interpret the improvement of common cold after 6 days as a placebo effect following administration of a placebo. He did not consider that many patients with the common cold get better spontaneously within 6 days (Diehl 1953). Similarly, in two studies (Keats and Beecher 1950; Keats et al. 1951) in which the reduction of postoperative pain was claimed to be a placebo effect by Beecher (1955), subsequent analysis showed that it was actually the spontaneous decline of postoperative pain (Kienle and Kiene 1997).

In this book, it will become clear that many clinical trials claiming high placebo responses do not consider spontaneous remission as an important (p.23) contributing factor in symptom reduction. For example, in the genitourinary apparatus (see Chapter 8), although placebo groups have been analyzed in some detail, no definitive conclusion can be drawn because spontaneous remission cannot be ruled out. The only way to rule out spontaneous fluctuations is to use an untreated group as a control. With a no-treatment group it is possible to assess the natural course, or natural history, of the disease; comparison between a placebo group and a natural history (no-treatment) group allows us to identify the component of the symptom that is subject to spontaneous fluctuations.

Fig. 2.1 is an example of spontaneous fluctuation of pain, the so-called natural history of pain. If a placebo, or an otherwise ineffective treatment, is administered at the time indicated by the arrow, the reduction in pain can be mistakenly attributed to the placebo, or the ineffective treatment. In other words, pain would have subsided anyway, regardless of placebo administration. To avoid this mistake, the natural history is compared with a placebo treatment and an active treatment. The difference between the natural history and the placebo treatment represents the placebo component of the therapy, whereas the difference between the placebo treatment and the active treatment represents the specific component of the therapy. Therefore, it is important to realize that a real placebo effect has nothing to do with the natural history of a symptom, although, as previously mentioned, many clinical trials fail to underscore this difference.

2.1.3 Regression to the mean is common in clinical trials

Spontaneous improvement can be considered a special case of regression to the mean, whereby there is a tendency for extreme values to move closer to the mean after repeated measurements. Indeed, regression to the mean is a statistical phenomenon that assumes individual people tend to have extreme values of a physiological parameter (e.g. glucose levels) when enrolled in a clinical trial, and that these extreme values tend to be lower at a second measurement (Davis 2002). In such cases, the improvement cannot be attributed to any intervention they might have undergone. One very important factor in this phenomenon is the selection criteria used for inclusion in a clinical trial. Fig. 2.2 shows an example with random numbers. It is sufficient to randomly generate numbers from 0.00 to 1.00 in order to obtain a model of regression to the mean (Ruck and Sylven 2006), which indicates that regression to the mean is not a biological phenomenon but rather a mathematical model. It can be seen that there is no difference between the means of the first and second series of numbers. However, if a selection is done, in this case values ≥ 0.70, the sample regresses to the mean with statistical significance. (p.24)

                   A modern view of placebo and placebo-related effects

Fig. 2.2 Selection bias may be introduced by factors such as the severity of symptoms, the selection of the diagnostic test, or the definition of the baseline. Every selection gives a model with regression to the mean. In (A) regression to the mean is illustrated by two random series of numbers with values between 0.00 and 1.00. The means of both sets of figures are given, and the statistical difference between the means is not significant (p = N.S.). The figures in bold have values ≥ 0.70. These selected figures were analyzed again on the right side; now their means show a significant difference (p < 0.04). In (B) the means for both sets of data are represented graphically. Modified from Ruck and Sylven 2006, with permission from Elsevier, Copyright 2006.

(p.25) Therefore, every selection bias gives a model with regression to the mean. If these random numbers are replaced with a physiological parameter, such as the plasma level of glucose, the selection of glucose levels above a given value (e.g.≥ 500 mg/100 mL) would produce similar results, giving the impression of improvement at the second measurement.

When assessing the possibility that an improvement is due to regression to the mean, rather than to a real placebo effect, adequate variables must be carefully selected. For example, in a trial of cholestyramine versus placebo (for the management of high levels of cholesterol), patients were selected on the basis of high plasma levels of cholesterol (Lipid Research Clinics Program 1984). One of the known side effects of cholestyramine is constipation, and 3% of patients were found to have constipation before the treatment. After 1 year of treatment, 39% of those who received cholestyramine had constipation, whereas 10% of those who received the placebo had constipation. The crucial point here is that these patients were not selected on the basis of constipation, thus regression to the mean is unlikely to have occurred. These data seem to be consistent with a real placebo effect, whereby constipation increases from 3% to 10% after placebo treatment (see also Davis 2002).

2.1.4 Signal detection ambiguity can sometimes explain symptom ‘reduction’

A particular type of error made by either the patient or the physician, a false positive error, may explain the illusory improvement occurring in some circumstances. This is known as ‘signal detection theory’. It is based on the occurrence of errors in detection of ambiguous signals (Clark 1969; Allan and Siegel 2002). For example, if the patient or the doctor is trying to detect a change in a symptom, say pain, then he or she can fail to identify it (a false rejection) or they can erroneously identify something that turns out to be something else (a false positive). As indicated by Allan and Siegel (2002), false positives are more likely to occur when the consequences of a false rejection are deemed to be greater than those of a false positive. A patient may erroneously detect symptomatic relief in response to an inert treatment, and certainly the ambiguity of the symptom's intensity, like fluctuations in pain, is important. In other words, small and large fluctuations represent a noisy environment that makes a judgment about a symptom difficult.

Signal detection theory is a model that was developed to explain decision making in noisy environments, and it simply states that some signals cannot be distinguished from noise on the basis of their magnitude. Clark (1969) investigated the effects of placebo administration on painful radiant thermal (p.26) stimulation and found that the placebo condition yielded fewer painful responses compared with a control. However, analysis in the context of signal detection theory showed that thermal sensitivity was the same in the placebo and control conditions. The effect of the placebo was to alter the subject's response criteria and not thermal sensitivity. Signal detection theory was also used in a study by Feather et al. (1972), who concluded that placebos did not produce real decreases in pain sensitivity.

Although signal detection theory may explain the effects of placebos in some circumstances, there is now strong neurobiological evidence that symptom reduction is real in some conditions, such as in pain and Parkinson's disease, in which the neural correlates of such reductions have been found (see Chapters 3 and 4). Therefore, signal detection theory must be considered a possible confounding factor in some circumstances, and it needs to be ruled out through the analysis of objective measurements, such as metabolic changes in the brain or changes in neuron activity.

2.1.5 Sometimes patients and doctors give biased reports of the clinical condition

Biases represent an important confounding factor in the evaluation of clinical improvement. This is true for both patients and doctors. For example, there is some evidence that the patient often wants to please the doctor for the time and effort spent helping him, so he may exaggerate his feeling of clinical improvement (Roberts 1995; Sackett 1995; Kienle and Kiene 1997). Other types of bias are represented by the so-called scaling biases, whereby an asymmetrical measurement scale (e.g. many categories for improvement and one or none for deterioration) tempts the patient to give too many positive reports (Kienle and Kiene 1997). It is also important to consider that some patients may exaggerate their symptoms in order to be included in a clinical trial (Kleinman et al. 2002).

Investigator biases are important as well; he or she may become unblinded during the course of a clinical trial, which may affect his or her expectations about the effectiveness of the therapy under study. Indeed, there is evidence that the doctor's expectations may affect the placebo response of his patients. Gracely et al. (1985) devised an experiment in which doctors knew they had a high or low probability of giving the powerful painkiller fentanyl to their patients. When the physicians knew that fentanyl administration was highly probable, the administration of placebo was more effective than when they knew there was a lower chance of administering fentanyl. Thus, the doctor's knowledge about a therapy may affect the outcome in patients through verbal communication and/or attitudes.

(p.27) 2.1.6 Co-interventions can sometimes be the cause of improvement

Co-interventions and additional treatments are other important confounding factors that are sometimes ignored, thus leading to erroneous interpretation. For example, a concomitant diet may be responsible for a clinical improvement during placebo treatment. There are several examples of neglected co-interventions that may have accounted for the reduction of a symptom. As pointed out by Kienle and Kiene (1997) in one trial on angina pectoris in the Beecher's analysis (Beecher 1955), the placebo group also received nitrates (Travell et al. 1949). In another trial on the common cold, patients were allowed to take hot baths, gargles and diets (Diehl 1953). Similarly, the study by Lichstein et al. (1955) claims substantial placebo effects in the treatment of unstable colon, but all these patients had been put on a special diet. In another study, this time on alcoholism, patients in the placebo group received specialized medical and psychological support (Wells 1957). Therefore, it is very important to consider all possible co-interventions when assessing the outcome of a therapy.

An interesting example of co-factors that may contribute to therapeutic outcome after placebo administration is the case of placebos for cough given in syrup form. In this regard, Eccles (2006) analyzed the beneficial role of the sweet taste of cough syrups on the cough itself. In fact, cough syrups contain many substances, including sugar, honey, capsicum and citric acid. These substances may in turn cause reflex salivation and promote mucus secretion, thus lubricating the pharynx and larynx and helping to reduce coughing. In general, sweet gustatory stimuli are also known to have antitussive effects through the facial, glossopharyngeal and vagus nerves (Eccles 2006). It is clear that any reduction of cough associated with these substances in syrups is not a real placebo effect, but rather the effect of a co-intervention (the stimulation of the gustatory receptors) and this is often neglected as a possible beneficial factor influencing coughing.

2.1.7 Classical clinical trials are not good for understanding placebo effects

All the phenomena analyzed in the previous sections are present in clinical trials. For example, if a symptom subsides in a group that takes placebo, this can be due to spontaneous remission, regression to the mean, some co-interven-tion, a real placebo effect (and so on) but there is no way to distinguish between all these factors in the absence of adequate control groups. Therefore, clinical trials do not represent a good model for studying and to understand the mechanisms underlying the placebo effect, unless the target of the trial is (p.28) the understanding of the placebo effect and the use of appropriate controls. Indeed, many results obtained in the classical clinical-trial setting (whereby the only comparison carried out is between an active treatment and a placebo group) differ from those obtained in the laboratory setting. This is not surprising, as in the laboratory setting it is possible to compare a no-treatment group with a placebo group, so that the spontaneous remission of symptoms can be identified. In the laboratory setting, other confounding variables can be controlled as well. For example, the statistical phenomenon of regression to the mean can be ruled out by using ‘experimental pain’, whereby pain intensity can be controlled experimentally and selections can be avoided. Likewise, symptom-detection ambiguity and subject bias can be eliminated through measurement of objective physiological parameters; furthermore, the possible effects of co-interventions can be eliminated by choosing an appropriate experimental model.

In 2001, Hrobjartsson and Goetzsche (2001) conducted a systematic review and meta-analysis of clinical trials in which patients were randomly assigned to either placebo or no treatment. The purpose was to investigate the clinical effect of placebos, discerning whether patients randomized to placebo under blind conditions have better outcomes than those randomized to no treatment. The investigators identified 130 trials in which 40 different clinical outcomes were investigated by selecting binary outcomes (e.g. the proportion of alcohol abusers and non-alcohol abusers) and continuous outcomes (e.g. the amount of alcohol consumed). They considered the effect of three types of placebos: pharmacological (e.g. a pill), physical (e.g. a manipulation), and psychological (e.g. conversation), and calculated the pooled relative risk for binary outcomes as well as the pooled standardized mean differences for continuous outcomes. The pooled relative risk was defined as the ratio of the number of patients with an unwanted outcome to the total number of patients in the placebo group, divided by the same ratio in the untreated group. The standardized mean difference was defined as the difference between the mean values for unwanted outcomes in the placebo and untreated groups divided by the pooled standard deviation. A negative value indicated a beneficial effect of placebo both for binary and continuous outcomes. The findings by Hrobjartsson and Goetzsche (2001) did not detect a significant effect of placebo, as compared with no treatment, in pooled data from trials with subjective or objective binary or continuous objective outcomes. However, they found a significant difference between placebo and no treatment in trials with subjective outcomes and in trials involving the treatment of pain. There also was some evidence that placebos had greater effect in small trials with continuous outcomes than in large trials, with an inverse relation between trial size (p.29) and placebo size. Furthermore, in an update of their first review, Hrobjartsson and Goetzsche (2004) argued that when a large effect of a placebo intervention is not present, small effects on continuous outcomes, for example in pain, could not be clearly distinguished from biases. Therefore, the observed significant effect of placebo on subjective outcomes may have been due to biased reports of the patients rather than to real placebo effects.

The meta-analysis by Hrobjartsson and Goetzsche (2001), albeit of great impact when published, was not successful subsequently and now most researchers do not take it very seriously. In fact, it is important to note that they used very broad inclusion criteria and failed to recognize that placebos are not expected to work uniformly across diseases or disorders (Shapiro and Shapiro 1997; Ader 2001; Brody and Weismantel 2001; Di Nubile 2001; Einarson et al. 2001; Greene et al. 2001; Kaptchuk 2001; Kirsch and Scoboria 2001; Kupers 2001; Lilford and Braunholtz 2001; Miller 2001; Papakostas and Daras 2001; Shrier 2001; Spiegel et al. 2001; Wickramasekera 2001). Aggregating without regard to the heterogeneity of disorders means we cannot discern whether a placebo really works. It is as if we wanted to test the effects of morphine across all medical conditions, like pain, schizophrenia, marital discord, asthma, nephritis, and other diseases. Of course a pooled analysis would find no effect of morphine. Another problematic aspect of Hrobjartsson and Goetzsche's study it that it is impossible to consider the critical factors involved in placebo responses, such as patient and physician expectations, the healing context, and the cues and factors that can influence the effectiveness of a therapeutic intervention (Di Blasi et al. 2001; Benedetti 2002; Moerman and Jonas 2002; Moerman 2003; Benedetti et al. 2005; Colloca and Benedetti 2005).

In 2002, Vase et al. (2002) conducted another meta-analysis that included 23 of the 29 clinical trials from the meta-analysis by Hrobjartsson and Goetzsche (2001) and an additional meta-analysis of 14 studies investigating placebo analgesic mechanisms. Although this study has been criticized by Hrobjartsson and Goetzsche (2006), Vase et al. (2002) found that the magnitudes of the placebo analgesic effects were higher in studies that investigated placebo analgesic mechanisms compared with clinical trials where the placebo was used only as a control condition. These authors suggest that this difference might be due to the different placebo instructions and suggestions given in the clinical trial setting compared to the experimental setting (Vase et al. 2002). In fact, clinical trial investigators typically avoid giving oral suggestions of analgesia in favour of neutral instructions, whereas investigators of the placebo effect typically emphasize the analgesic suggestions.

In general, these two meta-analyses are worthy of consideration because they present the scenario for two different ways of investigating the placebo (p.30) effect: on the one hand the randomized clinical trial, and on the other the clinical/experimental setting specifically designed to investigate the placebo effect. Clearly in order to understand the magnitude and mechanisms of the placebo effect across different medical conditions and therapeutic interventions, investigations must be carried out under strictly controlled conditions in the laboratory setting, or at least in a clinical trial that is intentionally devised to analyze the placebo group.

Even if the placebo effect is a phenomenon of small magnitude in some circumstances and in some medical conditions, understanding it is necessary within routine medical practice, the clinical trial setting and the experimental approach to mind–brain–body issues. For example, a recent debate focused on the possible small or large magnitude of placebo effects, omitting some crucial questions about the underlying mechanisms. Starting from the meta-analysis by Hrobjartsson and Goetzsche (2001), Wampold et al. (2005) re-analyzed the same studies and found robust placebo effects both for pharmacotherapies and psychotherapies. In a rebuttal, Hrobjartsson and Goetzsche (2007) presented evidence that claimed that Wampold's re-analysis was wrong.

As emphasized by Hunsley and Westmacott (2007), the meta-analysis results reported by the two sets of authors (Hrobjartsson and Goetzsche, and Wampold's group) are nearly identical, yet their conclusions differ dramatically. Hunsley and Westmacott (2007) also stress that both meta-analyses indicate that placebo effects do exist and cannot be dismissed as unimportant, although their magnitude appears to be small. Interestingly, these authors compared the size of the effect of placebos with those of other treatments by using the NNT index (number needed to treat), which represents the number of patients that one would need to treat in order to have one more successful outcome than would be possible with the control condition. By considering that the NNT value for placebos is approximately 7, Table 2.2 shows that the magnitude of the placebo effect is comparable to that of transfusion to treat stroke in children with sickle-cell anaemia or to the value of adding radiotherapy to tamoxifen in the treatment of breast cancer (Hunsley and Westmacott 2007).

As the context around the patient is the crucial factor in placebo responsiveness, and psychological factors are at the core of its magnitude, it is not surprising that placebo effects in clinical trials are highly variable, and often small. By contrast, as discussed throughout this book, when the context and the patient's psychological factors are manipulated under strictly controlled conditions in the laboratory setting, the magnitude of the placebo effect can be modulated like many other physiological parameters. Therefore, despite the usefulness of all these meta-analyses, the clinical trial setting is not a good model for understanding the placebo effect and is likely to lead to erroneous, (p.31)

Table 2.2 Examples of effect sizes from individual studies expressed as the number needed to treat (NNT). For comparison, NNT values for placebo are around 7 (from the Centre for Evidence-Based Medicine)

Condition or disorder

Intervention vs control



Chronic depression

Nefazodone and psychotherapy vs either treatment alone



Stroke in children with sickle-cell anaemia

Transfusion vs standard care

All strokes


Post-menopausal women with breast cancer

Radiotherapy plus tamoxifen vs tamoxifen alone



Patients resuscitated from ventricular arrhythmias

Implantable defibrillator vs anti-arrhythmic drug therapy

All-cause mortality


Hip fractures in nursing-home patients

External hip protectors vs control

Hip fracture


or at least confusing, interpretations. We should be aware that placebos act through a set of different mechanisms, thus they must be investigated using different approaches for different diseases and therapeutic interventions. For example, if we want to study the placebo effect in the immune and endocrine systems, we cannot elicit it through experimental manipulation of the patient's expectations, as expectation has no effect on immune mediators and hormones. By contrast, if we use a conditioning paradigm, robust placebo effects can be observed in both the immune and endocrine system (see Chapter 6).

2.2 What they are

2.2.1 Is the placebo effect different from the placebo response?

Today placebo researchers tend to use the term ‘placebo effect’ and ‘placebo response’ interchangeably (Colloca et al. 2008). Accordingly, throughout this book I will use the two terms interchangeably. However it should be noted that the term ‘placebo effect’ is sometimes considered to be different from ‘placebo response’. This is because, as seen in Chapter 1, the term ‘placebo effect’ was originally used to mean any improvement in the condition of a group of subjects that has received a placebo, thus including everything (spontaneous remission, regression to the mean, biases, co-interven-tions, real placebo responses, and so forth). By contrast, ‘placebo response'sometimes refers to the change in an individual caused by placebo manipulation, which represents the real psychophysiological placebo response of (p.32) a subject to the inert treatment. Unfortunately, it is not easy to identify a placebo response in a single individual, and many control subjects are sometimes necessary to rule out spontaneous remissions, biases and symptom-detection ambiguities. For example, in a situation where someone reports a pain reduction from 7 to 6 on a numerical rating scale of 0 to 10 after the administration of a placebo, there is no way to say whether this small reduction is attributable to a placebo effect or to a spontaneous remission or a bias. For this reason, the placebo effect is better described as a group effect and, indeed, most of the studies on placebo mechanisms consider the mean reduction of a symptom in a group of subjects after placebo administration. There are, however, a few instances in which the natural history of a symptom is straightforward and guarantees safe identification of the placebo response in a single individual. For example, in this sense, postoperative pain is a good model, as this kind of pain either increases or is constant over time during the first hours after surgery. Therefore, it is safe to assume that a reduction of postoperative pain after placebo administration during the first hours after surgery is not due to a spontaneous fluctuation.

I believe that the difference between the words ‘effect’ and ‘response’, which is often present in the literature, is only a matter of definition, and is worthy of consideration only in part. I think that distinguishing between placebo effects and placebo responses is neither advantageous nor useful, because we risk going back to the old concept of placebo effect, which included everything. Therefore, I suggest that we use placebo effect and response interchangeably to mean a psychobiological phenomenon occurring in an individual or in a group of individuals.

2.2.2 The psychosocial context around the therapy is the crucial factor

For a long time, the word placebo has been equated with ‘sugar pill’, as it was widespread practice to give a carbohydrate tablet as a means of detecting ‘mystifiers’ (identified through the success of the sham therapy) or as a compassionate remedy for the terminally ill. However, the aim of ‘pleasing’ the patient, as the etymology of the word suggests, can clearly be achieved not only with drugs but also with any medical treatment ranging from physical cures to psychotherapy. What matters is not the sugar, of course, but its symbolic significance, which can be attached to practically anything (Brody, 2000). Moerman (2002) went as far as proposing to substitute the term ‘placebo response’ with ‘meaning response’, to underscore the importance of the patient's beliefs about the treatment and stress what is present (something in the environment inducing the expectation of a benefit) rather than what is (p.33) absent (a chemical or manipulation of proven specific efficacy). Therefore, the concept of placebo has shifted from the ‘inert’ content of the placebo agent (e.g. sugar pills) to the concept of a simulation of an active therapy within a psychosocial context.

On the basis of these considerations, when a medical treatment (for instance, a drug) is given to a patient, it is not administered in a vacuum, but in a complex set of psychological states that varies from patient to patient and from situation to situation. For example, when morphine is given to relieve pain, it is administered along with a complex set of psychosocial stimuli which tell the patient that a benefit or worsening may occur shortly (Fig. 2.3). These psychosocial stimuli represent the context around the therapy and the patient and such a context may be as important as the specific pharmacodynamic effect of a drug. Di Blasi et al. (2001) listed a series of contextual factors that might affect the therapeutic outcome. These range from the treatment characteristics (colour and shape of a pill) to the patient's and provider's characteristics (treatment and illness beliefs, status and sex) and from the patient–provider relationship (suggestion, reassurance and compassion) to the healthcare setting (home or hospital) and room layout. Thus, the context is made up of anything that surrounds the patient under treatment, including doctors, nurses, hospitals, syringes, pills, machines, and such like, but certainly doctors and nurses are a very important component of the context, as they can transmit a lot of information to the patient through their words, attitudes and

                   A modern view of placebo and placebo-related effects

Fig. 2.3 When a medical treatment is being administered, it is surrounded by a complex psychosocial context—the sight of medical personnel and the hospital environment, the touch of machines, the words used by doctors and nurses, and the smell of drugs. All these sensory stimuli tell the patient that a therapy is being carried out. A positive context may improve the therapeutic outcome, whereas a negative context may worsen it.

(p.34) behaviours (Benedetti 2002). By using a single word, Balint (1955) referred to this context as the whole atmosphere around the treatment.

Thomas (1987) found that positive and negative consultations in general practice have an important impact on patients who present with minor illness (see section 9.6). Likewise, Di Blasi et al. (2001) examined 25 randomized clinical trials in which the context about the treatment and the patient's expectations about the therapeutic outcome were manipulated. A consistent finding was that those doctors who adopt a friendly and reassuring manner are more effective than those who are formal and do not offer reassurance.

Interestingly, marketing may affect the magnitude of the placebo effect. In a study in 835 patients suffering from headache, the participants were randomized to branded aspirin, unbranded aspirin, branded placebos, and unbranded placebos (Branthwaite and Cooper 1981). Pills were dispensed in either a plain bottle or a bottle with a prominent brand name on the label. Results showed that branded aspirin was the most effective, followed by the unbranded aspirin, then by branded placebo, and finally by unbranded placebo. More recently, Waber et al. (2008) investigated the effects of price on analgesic response to placebo pills. After randomization, half the participants were told that the drug (actually a placebo) had a regular price of US$2.50 per pill and half were told that the price had been discounted to US$0.10 per pill. Overall, the participants in the regular price group experienced a larger reduction in pain compared to the low-price (discounted) group. These results are consistent with described phenomena of commercial variables affecting patients’ expectations, and expectations influencing therapeutic efficacy. This may explain the popularity of high-cost medical therapies over inexpensive, widely available alternatives such as over-the-counter drugs (Waber et al. 2008).

The placebo effect is therefore a context effect (Di Blasi et al. 2001; Benedetti 2002). Not only is the context around a treatment associated with positive outcomes, but to negative outcomes as well. For example, distrust in a therapy and/or in medical personnel can make a patient expect a negative outcome, which is called nocebo effect (see below). It is important, therefore, to realize that the study of the placebo effect is the study of the psychosocial context around the treatment—whether it is positive or negative—and how these social stimuli may affect the patient's brain. This, in turn, may have either beneficial or negative effects on the course of a disease and/or the response to a therapy.

2.2.3 Placebo and nocebo effects occur when inerttreatments are given

By definition, a placebo effect is the effect that follows the administration of a placebo, that is, of an inert treatment. Therefore, any psychobiological effect (p.35)

Table 2.3 Main differences between placebo and placebo-related effects

Placebo effects

Placebo-related effects

Inert treatments are given

No inert treatments are given

Expectation induced by administration of inert treatments

Suggestions of improvement or worsening without administration of inert treatments

Conditioning, whereby the conditioned stimulus is neutral (inert)

Difference between expected (open) and unexpected (hidden) treatments

Nocebo effect, whereby the administration of inert treatments induces negative expectations

Expectation of belonging to either the placebo or active treatment group in a clinical trial

on the brain and/or the body that follows the administration of a placebo can be called, in its own rights, placebo effect or placebo response (Table 2.3). It is important to stress that the inert treatment is given along with contextual stimuli, for example verbal suggestions of clinical improvement which make the patient believe that the treatment is real and effective. As described above, it is crucial to point out that the inert treatment (e.g. a sugar pill or a saline solution) will never acquire therapeutic properties, so that a pharmacologically inert substance will always remain inert. What matters is the context and the verbal suggestions of clinical benefit. Therefore, a placebo would be better defined as an inert treatment plus the context that tells the patient a therapeutic act is being performed.

The nocebo effect is a placebo effect because an inert substance is administered. However, in order to induce a nocebo effect, the inert substance is given along with a negative context, for example verbal suggestions of clinical worsening, so as to induce negative expectations about the outcome. The term nocebo (‘I shall harm’) was introduced in contrast to the term placebo (‘I shall please’) by some authors to distinguish the pleasing from the noxious effects of placebos (Kennedy 1961; Kissel and Barrucand 1964; Hahn 1985, 1997). If the positive psychosocial context, which is typical of the placebo effect, is reversed in the opposite direction, the nocebo effect can be studied.

From an ethical point of view, the investigation of the nocebo effect is difficult to carry out. In fact, whereas the induction of placebo responses is certainly ethical in many circumstances (Benedetti and Colloca 2004), the induction of nocebo responses represents a stressful and anxiogenic procedure, because verbally induced negative expectations of symptom worsening may lead to a real worsening. Certainly, a nocebo procedure is unethical in patients, and this is one of the main reasons why much less is known about nocebo phenomena.

(p.36) 2.2.4 Placebo- and nocebo-related effects do not involve the administration of inert treatments

By definition, if no placebo is administered, the effect that follows its administration cannot be called a placebo or nocebo effect. However, it has become clear in recent times that the term placebo effect is too restrictive and does not help explain the underlying mechanisms (Benedetti 2008). Indeed, there are several placebo-like effects, whereby no placebo is given, which are due to the influence of the context surrounding the treatment on the patient's brain (Table 2.3). For example, although a placebo is usually given along with verbal suggestions of clinical improvement, verbal suggestions of either improvement or worsening can be given alone, without administering any inert treatment, so as to induce expectancies about the outcome.

Another example of placebo-related effect is represented by the decreased effectiveness of hidden treatments. It is possible to eliminate the placebo (psychosocial) component and analyze the specific effect of the treatment, free of any psychological contamination, by making the patient completely unaware that a medical therapy is being carried out (Colloca et al. 2004). To do this, drugs are administered through hidden infusions by machines. A hidden infusion of a drug is delivered by a computer-controlled infusion pump, pre-programmed to deliver the drug at the desired time. The crucial point here is that the patients do not know that any drug is being injected, so that they do not expect anything. By contrast, an open administration represents routine medical practice, whereby drugs are given overtly and the patients expect a clinical benefit. Therefore, an open injection of a drug represents an expected treatment, whereas a hidden injection represents an unexpected therapy. The difference between the outcomes following administration of the expected and unexpected therapy is the placebo (psychological) component, even though no placebo has been given (Colloca et al. 2004).

Regardless of placebo administration, expectations can make a big difference in a clinical trial. In fact, what the subjects expect in a clinical trial may influence the outcome, irrespective of whether they belong to the placebo group or to the active treatment group (Benedetti 2005, 2007). For example, in one clinical trial, real acupuncture was compared with placebo acupuncture and patients were asked which group they believed they belonged to (either placebo or real treatment). Patients who believed they belonged to the real acupuncture group experienced greater clinical improvement than those who believed they belonged to the placebo acupuncture group (Bausell et al. 2005) (see also sections 3.1.3 and 4.1.2).

(p.37) Placebo- and nocebo-related effects, therefore, do not involve the administration of any inert treatment (placebo). However, it appears clear that in many circumstances similar mechanisms are at work. For example, is there any difference between placebo administration along with suggestions of analgesia and suggestions of analgesia alone? Maybe, the administration of a placebo, which is nothing but the simulation of the therapeutic act, enhances the suggestions of improvement because it makes the verbal suggestions more credible. Although this point is still unknown, it appears clear that it is neither useful nor advantageous to restrict the study of the placebo effect to the procedure of giving a placebo.

2.2.5 Are subjective outcomes different from objective outcomes?

One of the most debated issues in the history of placebo research is about subjective versus objective measurements. The specific reason for this debate is that the subjective reports by patients about their clinical condition are often considered to be strongly influenced by biases (see section 2.1.5). Indeed, Hrobjartsson and Goetzsche (2001) found a difference between subjective and objective measurements. They did not detect a significant effect of placebo (as compared with no treatment) in trials with subjective or objective binary or continuous objective outcomes, but they did find a significant difference between placebo and no treatment in trials with subjective outcomes and in trials involving the treatment of pain. There also was some evidence that placebos had a greater effect in small trials with continuous outcomes than in large trials, which may be due to biased reports of patients rather than to real placebo effects.

Although this issue is clearly an important one, today there is no reason to believe that subjective placebo responses can be dismissed as report biases. In fact, with a typical subjective symptom like pain, there are now a number of brain imaging studies that indicate the subjective experience of pain reduction following placebo administration is accompanied by objective changes in several brain regions where pain transmission is inhibited (see Chapter 3). Likewise, substantial placebo responses can be observed in motor disorders such as Parkinson's disease, in which motor performance can be measured objectively (see Chapter 4), and in the immune and endocrine systems, in which objective measurements of immune mediators and hormone plasma concentrations can be obtained.

Therefore, the question of whether placebos induce subjective or objective changes has been reframed with respect to identifying the neural substrates of the subjective changes. In other words, subjective reports by patients (p.38) are as important as the objective changes, provided that other phenomena, like the patient's biases, can be ruled out. For example, a subjective report of well-being after placebo administration is worthy of scientific inquiry in the same way as an objective change in motor performance. One good example is seen in a study in Parkinson's disease, in which subjective reports of well-being were recorded following administration of a placebo, such as ‘I feel good, I want to stand up’ or ‘I feel like the usual therapy’ or ‘I feel much better’ (Benedetti et al. 2004). Without the appropriate controls and objective measurements, these reports could not be distinguished from biases, such as the patient's desire to please the experimenter. An in-depth analysis of these subjective effects revealed that the subjective reports of well-being were accompanied by the objective reduction of muscle rigidity, as assessed by a blinded neurologist, as well as by a reduction of the firing rate and bursting activity of subthalamic nucleus neurons, as assessed by intraoperative single-unit recording (Benedetti et al. 2004).

Thus, a clearcut distinction between subjective and objective measurements is neither advantageous nor useful, provided that the correct methodology is applied to rule out possible biases and other phenomena. Whenever a subjective report of well-being is reported by patients after placebo administration, before dismissing it as a bias or a subjective experience that is not worthy of consideration we should devise the appropriate experimental approach in order to uncover the underlying mechanisms.

2.3 How they work

2.3.1 There is not a single placebo effect but many

There is not a single mechanism of the placebo effect and not a single placebo effect—but many. So we have to look for different mechanisms in different medical conditions and in different therapeutic interventions. Expectation and anticipation of clinical benefit play a crucial role when conscious physiological functions are involved, whereas classical conditioning takes place in unconscious physiological functions (Benedetti et al. 2003) (Fig. 2.4). For example, expectations have no effect on hormone secretion, whereas a conditioning procedure can induce conditioned placebo hormonal responses. Therefore, different systems and apparatuses as well as different diseases and treatments are affected by placebos in different ways. Indeed, this book aims to describe the different placebo and placebo-like effects across different medical conditions in order to make it clear that the placebo effect is a general phenomenon that involves different mechanisms.


                   A modern view of placebo and placebo-related effects

Fig. 2.4 The psychosocial context around the treatment may act on the patient's brain through unconscious and conscious mechanisms. Conscious mechanisms involve complex cognitive factors, such as expectation and anticipation of benefit, belief in the treatment, and trust and hope. Unconscious mechanisms involve classical conditioning whereby, after repeated pairings between a conditioned contextual stimulus (e.g. the colour and the shape of a pill) with an unconditioned stimulus (the pharmacological agent inside the pill), the conditioned stimulus alone can produce an effect (a conditioned response).

The existence and occurrence of many placebo effects across diseases has also an important methodological meaning. It is necessary to adopt the appropriate methodology and procedure in order to study different placebo effects. For example, if a procedure that induces strong expectations is used to elicit hormonal placebo responses, no effect will be observed, as hormone secretion is not affected by expectations (see section 6.3). Of course, until all the mechanisms involved in different conditions are identified, it is not possible to rule out the possibility of some common mechanisms across medical conditions. For example, reward mechanisms as well as classical conditioning may represent a common substrate in different diseases and different systems.

2.3.2 Expectation of a future outcome is one of the principal mechanisms

Most of the research on placebos has focused on expectations as the main factor involved in placebo responsiveness (Fig. 2.4). Indeed, the literature is (p.40) full of studies that analyze expectations, and the terms ‘effects of placebos’ and ‘effects of expectations’ are frequently used interchangeably, as described many times throughout this book.

In general, expectations of a future outcome and of a future response—the so-called response expectancies—are held by each individual about his or her own emotional and physiological responses such as pain, anxiety and sexual arousal (Kirsch 1985, 1990, 1999). Expectations may lead to a cognitive readjustment of the appropriate behaviour. Thus, it is not surprising that positive expectations lead to adoption of a particular behaviour, for instance resuming a normal daily schedule, whereas negative expectations lead to its inhibition (Bandura 1977, 1997; Bootzin 1985). Otherwise, the effects of expectations may be mediated by changes in other cognitions, such as a decrease in self-defeating thoughts when expecting analgesia (Stewart-Williams and Podd 2004). Expectations are unlikely to operate alone, and several other factors have been identified and described, such as memory and motivation (Price et al. 1985, 2001, 2008; Price and Barrell 2000; Geers et al. 2005a) and meaning of the illness experience (Pennebaker 1997; Brody 2000). According to Brody (2000) the meaning precedes other causal mechanisms, like expectation. In an attempt to explain the causal mechanisms of the placebo effect, Bootzin and Caspi (2002) and Caspi and Bootzin (2002) integrate multiple explanatory mechanisms in a model that involves factors which are both internal and external to the individual, including expectation.

Expectation is certainly a difficult issue, as it can involve different factors and mechanisms. For example, Frank (1961, 1971, 1981) analyzed the healing process within the context of patient's expectations, though he proposed that hope is the primary mechanism of change in the folk tradition of healing and in psychotherapy. Indeed, hope can be defined as the desire and expectation that the future will be better than the present. Expectations may also play a role in the so-called Hawthorne effect. This describes the clinical improvement in a group of patients in a clinical trial that is attributable to the fact they are being observed (Last 1983). In other words, a patient who knows he is being studied may expect a better therapeutic benefit because of the many examinations he undergoes, the special attention he receives from the medical personnel, and the trust in the new therapy under investigation. It appears therefore clear that expectation is a general term that can be considered from many different perspectives.

From a neuroscientific point of view, expecting a future event may involve several brain mechanisms aimed at preparing the body to anticipate that event. For example, the expectation of a future positive outcome may reduce anxiety and/or activate the neuronal networks of reward mechanisms, whereas (p.41) the expectation of a negative outcome may result in anticipation of a possible threat, thus increasing anxiety. Indeed, anxiety has been found to be reduced after placebo administration in some studies. In other words, if one expects a distressing symptom to subside shortly, anxiety tends to decrease. For example, McGlashan et al. (1969) and Evans (1977) studied experimental pain in both ‘trait’ and ‘state’ anxiety subjects. Trait anxiety represents a personality trait which can be found throughout life; state anxiety may be present in specific stressful situations, representing an adaptive and transitory response to stress. These researchers gave the subjects a placebo which they believed to be a painkiller. There was no correlation between trait anxiety and pain tolerance after placebo administration, but there was a strict correlation between situational anxiety and pain tolerance during the placebo session. In fact, subjects who showed decreased anxiety had better pain tolerance than those who experienced increased anxiety. Similar results were obtained more recently by Vase et al. (2005), who found decreased anxiety in patients with irritable bowel syndrome who received a placebo treatment. Moreover, brain imaging studies have found reduced activation of anxiety-related areas during a placebo response (Petrovic et al. 2005) (see also section 5.2.2).

Expectations may also induce changes through reward mechanisms, which ensure future reward acquisition. These mechanisms are mediated by specific neuronal circuits that link cognitive, emotional and motor responses; they are traditionally studied in the context of the pursuit of natural (e.g. food), monetary, and drug rewards (Mogenson and Yang 1991; Kalivas et al. 1999). In animals, dopaminergic cells in the brainstem ventral tegmental area that project to the nucleus accumbens of the ventral basal ganglia respond to both the magnitude of anticipated rewards and deviations from the predicted outcomes, thus representing an adaptive system modulating behavioural responses (Setlow et al. 2003; Tobler et al. 2005; Schultz 2006). A simplified schema of the reward circuitry is shown in Fig. 2.5. The nucleus accumbens plays a central role in the dopamine-mediated reward mechanism, together with the ventral tegmental area, the amygdala, the periaqueductal grey and other areas in the thalamic, hypothalamic, and subthalamic (pallidum) regions.

In 2001, de la Fuente-Fernandez et al. (2001, 2002) used positron emission tomography to show dopamine activation in the nucleus accumbens following administration of a placebo in patients with Parkinson's disease. Interestingly, no relationship was found between dopamine activity in the nucleus accumbens and the actual placebo effects on motor function; this suggests that dopamine activation was better related to the expectation of reward (see section 4.1). Likewise, Mayberg et al. (2002) observed activation of (p.42)

                   A modern view of placebo and placebo-related effects

Fig. 2.5 A simplified schema of the reward system. It has been found to be activated by placebos in some circumstances, as assessed by an increase in dopamine activity in the nucleus accumbens. In the case of placebo administration, the reward is represented by the clinical improvement.

the nucleus accumbens in depressed patients after 1 week of placebo treatment, and Petrovic et al. (2005) described a correlation between ratings of negative-affect improvement during preconditioning with a benzodiazepine and ventral basal ganglia synaptic activity after receiving a placebo (see also sections 5.1 and 5.2). In addition, Scott et al. (2007) found a correlation between responsiveness to placebo and responsiveness to monetary reward in a model of experimental pain in healthy subjects; they also found that placebo responsiveness was related to the activation of dopamine in the nucleus accumbens (see section 3.1.7). All these studies seem to confirm involvement of dopamine activity in the nucleus accumbens, and the general reward circuitry, in the expectations induced by administration of a placebo. Detailed accounts of this are covered in other chapters.

2.3.3 The placebo effect is a learning phenomenon

Subjects who suffer from a painful condition, such as a headache, and who regularly consume aspirin, can associate the shape, colour and taste of that pill to a decrease in their pain. After repeated associations, a sugar pill that looks like aspirin can also decrease their pain (Fig. 2.4). In place of the shape, colour and taste of pills, several other stimuli can be associated with clinical improvement, such as syringes, stethoscopes, white coats, hospitals, doctors, nurses, and so on. The mechanism that underlies this effect is ‘conditioning’ whereby a conditioned (neutral) stimulus (like the colour or shape of (p.43)

                   A modern view of placebo and placebo-related effects

Fig. 2.6 Tolerance to experimental ischaemic arm pain. In (A) a placebo (suggestion of analgesia) is given for the first time on day 2 (days 1 and 3 represent control conditions) as shown by the black column. In (B) the placebo is given on day 4 after administration of the painkiller ketorolac on days 2 and 3 (shaded columns). Days 1 and 5 represent control conditions. It can be seen that the placebo given after pharmacological preconditioning induces greater placebo analgesia than placebo given without preconditioning. From Amanzio and Benedetti 1999 with permission from the Society for Neuroscience, Copyright 1999.

a pill) can be effective in inducing the reduction of a symptom if it is repeatedly associated with an unconditioned stimulus (like the drug inside the pill). This type of associative learning may represent the basis of many placebo effects—the placebo is the conditioned (neutral) stimulus itself. Indeed, in the 1960s, Herrnstein (1962) found that injection of scopolamine induced motor changes in rats, and these motor changes also occurred after injection of a saline solution (placebo) which was performed after the scopolamine injection.

A sequence effect of this sort also occurs in humans. For example, it has long been known that placebos given after drugs are more effective than when they are given for the first time (Sunshine et al. 1964; Batterman 1966; Batterman and Lower 1968; Laska and Sunshine 1973). Fig. 2.6 shows that if a placebo is given for the first time, the placebo response is present but small. If the placebo is administered after two prior administrations of an effective painkiller, the placebo analgesic response is much larger (Amanzio and Benedetti 1999), thus indicating that the placebo effect is a learning phenomenon.

(p.44) These clinical and pharmacological observations are in keeping with studies conducted in the laboratory setting by Voudouris et al. (1989, 1990). They showed that the placebo effect can indeed be conditioned. They applied a neutral non-anaesthetic cream (placebo) to one group of subjects who were assured that it was a local anaesthetic; not surprisingly, some of these subjects showed a placebo response after painful electrical stimulation. In a second group, the application of the same placebo cream was associated with surreptitious reduction of the intensity of stimulation, so as to make them believe that it was a powerful painkiller. These subjects, who had experienced a ‘true analgesic effect’, became strong placebo responders. Voudouris et al. (1989, 1990) concluded that conditioning is the main mechanism involved in the placebo effect.

These experiments were replicated by other authors, however a cognitive component was found to contribute to the conditioning-induced placebo responses. For example, De Jong et al. (1996) found a correlation between the expected and the actual level of analgesia in a similar experimental situation, thus suggesting that expectation is involved. Montgomery and Kirsch (1997) used a design in which subjects were given cutaneous pain via iontophoretic stimuli (see also Chapter 3). Subjects were surreptitiously given stimuli with reduced intensities in the presence of a placebo cream (the conditioning procedure) and were divided into two groups. The first did not know about the manipulation of the stimulus; the second group was informed about the experimental design and learned that the cream was inert. There was no placebo analgesic effect in this second group. This suggests that conscious expectation is necessary for placebo analgesia. This is a very important point because it suggests that expectation plays a major role, even in the presence of a conditioning procedure. In other words, expectation and conditioning are not mutually exclusive—they may represent two sides of the same coin (Stewart-Williams and Podd 2004).

Early in the 1930s, Tolman (1932) dissented from the view that conditioning is an automatic non-conscious event, due to the temporal contiguity between the conditioned and the unconditioned stimulus. Indeed, in the 1960s, conditioning was reinterpreted in cognitive terms on the basis that conditioned learning does not depend simply on pairing of the conditioned and unconditioned stimuli, but on the information that is contained in the conditioned stimulus (Rescorla 1988). In other words, conditioning would lead to the expectation that a given event will follow another event, and this occurs on the basis of the information that the conditioned stimulus provides about the unconditioned stimulus (Reiss 1980; Rescorla 1988; Kirsch et al. 2004).


                   A modern view of placebo and placebo-related effects

Fig. 2.7 Different action of expectation and conditioning on physiological functions. During a placebo procedure, conscious physiological processes like pain and motor performance are affected by verbally induced expectations, even though a conditioning procedure is performed. By contrast, unconscious physiological processes like hormone secretion are totally unaffected by expectations, but are influenced by placebos through unconscious conditioning mechanisms. From Benedetti et al. 2003 with permission from the Society for Neuroscience, Copyright 2003.

Despite the reinterpretation of conditioning in cognitive terms, conditioning in humans is not always cognitively mediated, particularly when considering conditioned placebo responses (Stewart-Williams and Podd 2004). For example, there is experimental evidence in humans that unconscious conditioned placebo responses are present in the immune and endocrine system (Chapter 6) and in the cardiovascular and respiratory system (Chapter 7). It has also been suggested that unconscious conditioning is important in placebo responses that involve unconscious physiological functions, whereas it is cognitively mediated when conscious processes come into play (Benedetti et al. 2003). According to this model, expectation has no effect on unconscious processes (Fig. 2.7). It is also worth noting that even in placebo analgesia some non-cognitive unconscious components may be present. For example, Amanzio and Benedetti (1999) showed that a placebo analgesic response is still present, albeit reduced, in the absence of expectation after prior conditioning, thus suggesting that a small portion of the placebo effect may occur unconsciously.

Therefore, as pointed out many times (Ader 1997; Siegel 2002), many placebo effects can be explained in the context of conditioning theories. In fact, a placebo is by definition a neutral stimulus with no therapeutic effects, in the same way that a conditioned stimulus is by definition neutral. Likewise, a placebo response is by definition elicited by a neutral stimulus, in the same way that a conditioned response is induced by a neutral stimulus. The specific experimental settings and effects are described elsewhere in this book.

A central point in the conditioning mechanisms that underlie the placebo effect is represented by the nature of the unconditioned response (Siegel 2002; (p.46) see also section 6.3.1). There are, in fact, some examples of conditioned responses that do not go in the same direction as the unconditioned response. For example, the unconditioned response to insulin is hypoglycaemia, while the conditioned response has also been hyperglycaemia in some conditions (Siegel 1972, 1975). Whenever the conditioned response is opposite to the unconditioned response, the unconditioned response must be defined correctly. In fact, administration of insulin produces hypoglycaemia which is not, however, the unconditioned response, but rather the unconditioned stimulus that is detected by receptors in the central nervous system. The real unconditioned response is the compensatory hyperglycaemia that the central nervous system adopts once the hypoglycaemia has been detected. Thus, by defining the unconditioned response correctly, the apparent contrasting outcomes can be resolved.

An important point emphasized by Ader (1997) is that the typical studies of conditioning use an experimental approach whereby the unconditioned response is a response that goes outside the boundaries of normal homeostasis. For example, painful stimulation is usually adopted as an unconditioned stimulus in animals and healthy human volunteers, and pain is a typical situation outside normal homeostasis. A patient is usually in the opposite situation. In fact, the starting point of a patient in pain is outside homeostasis, and a placebo tends to restore the normal homeostatic condition, that is, no pain. This might be an important difference between the experimental setting and the real clinical situation.

Conditioning is not the only learning mechanism that might be involved in placebo phenomena. Social learning is another form of learning, where people learn from one another by observational learning and imitation. The placebo effect may involve social learning as well, as expectations of future positive or negative outcomes may have a major effect in social learning (Bootzin and Caspi 2002).

2.3.4 Some personality traits may be associated with placebo responsiveness

Although there are some inconsistencies in findings about the relationship between personality and placebo responsiveness (Geers et al. 2005b), a few recent studies indicate that certain personality traits can predict placebo responsiveness. For example, De Pascalis et al. (2002) found that individual differences in suggestibility contribute significantly to the magnitude of placebo analgesia. The highest placebo effect was found in highly suggestible subjects who received suggestions that were presumed to elicit high expectations for drug efficacy.

(p.47) Geers et al. (2005b) found that personality and situational variables interact to determine placebo responding. In this study, optimists and pessimists were randomly assigned to one of three groups. The first group was told they were to ingest a pill that would make them feel unpleasant (deceptive expectation); the second group was told they were to ingest either a real or an inactive pill (conditional expectation); the third was told they were to ingest an inactive pill (control). The pessimists were more likely than the optimists to follow a negative placebo (nocebo) expectation when given a deceptive expectation, but not when given a conditional expectation. This suggests that the personality variable ‘optimism–pessimism’ relates to placebo responding when people are given a deceptive, but not a conditional, expectation. Thus, personality and situational variables seem to interact to determine placebo responding.

In a subsequent study, Geers et al. (2007) tested people who varied in their level of optimism. In the first condition, they were given the expectation that a placebo sleep treatment would improve their sleep quality. In the second condition, they underwent the same sleep treatment but were not given the positive placebo expectation. In the third condition, they received neither the positive placebo expectation nor underwent the placebo sleep treatment. Optimism was positively associated with better sleep quality in the first condition, suggesting that optimism relates to placebo responding.

2.3.5 What is the difference between placebo responders and non-responders?

In light of the mechanisms reviewed above, it appears that placebo responsiveness may depend on many factors, and on the mechanisms involved in a given type of placebo effect (e.g. learning and reward). It should be emphasized once again, therefore, that there is not a single placebo effect but many, so there also are many reasons why some people respond to placebos while others do not. Learning is certainly an important factor, as people who have had prior positive therapeutic experiences show larger placebo responses than those who have not (Fig. 2.6). In a study by Colloca and Benedetti (2006), small, medium and large placebo responses were observed, depending on several factors like their previous positive or negative experiences of analgesic treatment and the time lag between treatment and the placebo responses, further indicating that placebo analgesia is finely tuned by prior experience (see Fig. 3.1B).

Another important determinant of placebo responsiveness may be individual differences in the efficiency of the neural mechanisms of reward. In fact, a correlation between expectation of reward and dopamine activation in the (p.48) nucleus accumbens has been found in people with Parkinson's disease (de la Fuente-Fernandez et al. 2001, 2002). In another imaging study that used both positron emission tomography and functional magnetic resonance, Scott et al. (2007) tested the correlation between responsiveness to placebo and responsiveness to monetary reward. They found that responsiveness to placebo was related to activation of dopamine in the nucleus accumbens. In addition, these subjects were then tested for monetary responses in the nucleus accumbens. A correlation between placebo and monetary responses was found, thus suggesting that the efficiency of the nucleus accumbens reward system may play a role in placebo responsiveness (see section 3.1.7).

Another approach to understanding the differences between placebo responders and non-responders is to analyze the biological changes occurring in the brain and/or body following placebo administration. However, these phenomenological findings fail to unravel exactly what determines such differences in responsiveness. For example, Lipman et al. (1990) found an increase in plasma endorphins in patients with low back pain who responded to a placebo treatment, but patients who did not respond showed no changes in plasma endorphins. Similarly, Petrovic et al. (2002) found that the anterior cingulate cortex was activated in placebo responders but not in non-respon-ders. Benedetti et al. (2004) observed changes in the firing pattern of neurons in the subthalamic nucleus in placebo responders, but not in non-responders (see also Chapters 3 and 4). As discussed above, although these differences in biochemical and brain responses show that some biological responses to placebos are different in responders and non-responders, they give little information about the determinants of these differences.

2.4 Why they occur

2.4.1 Expectation-mediated placebo effects may be related to other self-regulatory processes

By considering other self-regulatory processes, the brain regions engaged by placebos that involve manipulation of expectations may be part of a general circuit underlying the voluntary regulation of affective–emotional responses. A summary of data obtained from 15 studies on the placebo effect, regulation of emotions and activation by opioid drugs is shown in Fig. 2.8. The superposition of peak coordinates of increased activation in each of these conditions reveals that several frontal areas are consistently engaged during different tasks in which negative affect must be suppressed. On the lateral surface, these regions include the dorsolateral and ventrolateral prefrontal cortex, and the rostral prefrontal cortex. On the medial surface, the midrostral dorsal anterior (p.49) cingulate and the neighbouring superior medial prefrontal cortex are involved. On the orbital surface, the peaks of increased activation are located around the medial orbital sulcus bilaterally. All these regions, with the notable exception of the orbitofrontal cortex and right ventrolateral prefrontal cortex, have been found to be activated following administration of an opioid analgesic (Firestone et al. 1996; Adler et al. 1997; Wagner et al. 2001; Petrovic et al. 2002). Both the dorsal and ventral prefrontal cortex have also been shown to be consistently activated in the voluntary positive reinterpretation of the meaning of aversive visual stimuli (Ochsner et al. 2002, 2004; Levesque et al. 2003; Phan et al. 2005) and correlated with reduced activation of the amygdala (Lieberman et al. 2005) and anxiety (Bishop et al. 2004).

In a study of placebo modulation of affective responses to pictures, Petrovic et al. (2005) found placebo-induced activity in both the dorsolateral and ventrolateral prefrontal cortex and the midrostral cingulate. Both increased activation in these regions and placebo-induced decreases in the amygdala were correlated with larger placebo effects in reported emotion. Interestingly, Rainville et al. (1997) found that the same region of the cingulate cortex was modulated by hypnotic analgesia. It is important to point out that these different varieties of self-regulation have not been tested in the same study. Nonetheless, the co-localization shown in Fig. 2.8 suggests there may be a general system for self-regulation that applies to both emotions and pain, as well as both voluntary strategies and the externally generated appraisals that produce placebo effects.

Taking all these factors into account, one possibility is that placebo and placebo-related effects, in which expectations are involved, are driven by executive control. For example, the dorsal and ventral prefrontal cortex is activated by a large class of cognitively demanding conditions. Distraction from pain also produces activation in these regions (Petrovic and Ingvar 2002) and working memory and executive attention have also revealed similar activation patterns (Wager and Smith 2003; Wager et al. 2004a). Another possibility is that this neural network subserves the process of meaning generation and appraisal of current and predicted events (Lazarus 1991). Effective placebo treatment may engender and active re-evaluation of the significance of pain, which engages both the orbitofrontal cortex and the lateral prefrontal systems in generation and maintenance of short-term context that biases ongoing nociceptive and affective processing (Miller and Cohen 2001).

The executive functions of the prefrontal areas may generate these self-regulatory processes, including expectation-related placebo effects; this is supported by the disruption of placebo-like responses in people with Alzheimer's disease, who show impairment of their prefrontal executive control (p.50)

                   A modern view of placebo and placebo-related effects

Fig. 2.8 Regions of the frontal lobes showing increased activity in recent studies of self-regulation. Increases are shown for delivery of opiate analgesics compared with resting non-drug control states (blue letters), downregulation of aversive emotional experience (green letters) through emotional reappraisal, and placebo effects on pain or emotional processing (red letters). Some peaks reflect regions for which increases in activity are correlated with reductions in negative emotional experience or pain. One exception is the study by Bishop et al. (2004) in which frontal activation was correlated with a reduced state of anxiety. Peak locations from the same study within 12 mm were averaged together for clarity of presentation. From Benedetti et al. 2005 with permission from the Society for Neuroscience, Copyright 2005.

Studies on opioid increases

Studies on placebo

F—Firestone et al. 1996

W—Wager et al. 2004b (anticipation)

A—Adler et al. 1997

G—Wager et al. 2004b (pain)

N—Wagner et al. 2001

I—Lieberman et al. 2004

P—Petrovic and Ingvar 2002

V—Petrovic et al. 2002

Studies on emotion regulation

T—Petrovic et al. 2005

L—Levesque et al. 2003

M—Mayberg et al. 2002

C—Ochsner et al. 2002

O—Ochsner et al. 2004

H—Phan et al. 2005

B—Bishop et al. 2004

(p.51) (as assessed by the Frontal Assessment Battery) and functional disconnection of their prefrontal lobes (as assessed by electroencephalographic connectivity analysis) (Benedetti et al. 2006) (see section 5.3.2). Therefore, placebo and placebo-like effects that involve expectation mechanisms may be phenomena that emerge from the ability of the prefrontal self-regulatory network to suppress negative emotions.

2.4.2 Are placebo and placebo-related effects a product of evolution?

There are many endogenous defence mechanisms in the body, ranging from cellular immune responses and antibody production to wound healing and nerve regeneration. One of the crucial issues for understanding why placebo and placebo-related effects exist at all is whether they represent a sort of endogenous healthcare system that has emerged in humans during evolution. As explained above, there is not a single placebo effect, but many. For example, when the placebo is a neutral stimulus that is repeatedly associated with an unconditioned stimulus, the conditioned placebo response must be viewed in the context of conditioning mechanisms. In this case, associations between stimuli (conditioned and unconditioned) occur in the clinical setting, but the meaning is the same as in other contexts—that is, learning to anticipate an event.

When placebo and placebo-related effects are related to expectations, the role of cognition in social interaction plays a key role. It has been suggested, for example, that the facial expression of pain evolved for eliciting medical attention from others (Williams 2002). In this respect, as suggested by Evans (2002), it is important to know when medical care first emerged, when special care to a sick individual was provided for the first time. Evans (2002) suggests that ‘medicine’ must have originated some time between 5,000,000 and 10,000 years ago (a huge span of time). In fact, the remains of skulls that have undergone complex surgery have been found, dating back more than 7,000 years, which suggests sophisticated forms of medical care had evolved at that time. Maybe the first provision of special care to the sick dates back to the very early hominids, the Australopithecus or Homo habilis. Indeed, non-human primates demonstrate forms of social interactions that resemble medical care, such as grooming and picking ticks off each other's backs (Fig. 2.9).

Reasoning in this way, health management may have evolved in a social context among different groups of hominids. The acts of caring and curing must have become a powerful social stimulus that induced beliefs, trust, hope and expectations of recovery. If one member of a social group trusts just one other member of that group, he or she may have an improved quality of life and may (p.52)

                   A modern view of placebo and placebo-related effects

Fig. 2.9 During the course of evolution, social interactions may have evolved from grooming in non-human primates or early hominids (A), to shamanism in primitive societies (B), to modern doctors (C). The act of taking care of somebody in the social group may have been very important in the emergence of placebo and placebo-related effects. As soon as the act of giving special care to a sick individual emerged (what we call medical care today), those who trusted a member of their social group (whether a chimpanzee, shaman or doctor) were likely to have greater advantages than those who did not.

survive longer. In primitive societies, this trusted group member was the ‘shaman’, but in our modern human society this is a ‘doctor’ (Fig. 2.9). Therefore, individuals who trust a member of their social group, whether a chimpanzee, shaman or doctor, are better placed than those who do not. Expectation-related placebo effects may be part of the evolution of these complex social interactions (Humphrey 2002). To become activated, these placebo responses require social contact with the person who is trusted. While the placebo effects require the act of curing (the placebo), the placebo-like effects occur without any act of curing, but simply with the human interaction, both verbal and non-verbal forms.

Interestingly, Wall (1999) claimed that pain is a ‘need'state, which can be terminated by specific consummatory acts like hunger and thirst. According to Wall (1999) the consummatory acts that terminate pain can be, for instance, either withdrawing one's hand from a noxious stimulus or care and attention from others (the act of caring and the placebo). It is this purely social event that represents the evolutionary novelty in humans (Evans 2002). A person whose brain is capable of shutting down pain when the presence of medical help is detected may have an advantage over someone whose brain lacks this capacity (Evans 2003). In Chapter 3, we will see that specific endogenous mechanisms are activated by this social act, resulting in suppressed pain transmission.

(p.53) 2.5 Points for further discussion

  1. 1) We need to know where, when and how placebo effects work. In particular, it is crucial to understand whether some mechanisms, such as reward, are common in different conditions.

  2. 2) We know that conditioning is important in unconscious physiological functions like immune responses and hormone secretion. However, according to modern theories of learning, cognitive factors are essential for classical conditioning. It would be interesting to address this issue, particularly within the context of the broader literature on conditioning. Beyond the placebo effect, a key question is: Does unconscious conditioning exist in humans?

  3. 3) Although it is clear that prior experience plays a role in many placebo effects, we need to know how this occurs. For example, is it possible to increase the magnitude and duration of placebo responses through learning?

  4. 4) Placebo-related effects may need to be extended further to include a variety of psychosocial factors. Harrington (2002) tried to compare placebo effects with a broad range of phenomena, for example the influence of hospital rooms on recovery, or the influence of cultural and religious beliefs on mortality.

  5. 5) Today we know that placebo responders and non-responders may differ in terms of either previous experience (learning) or efficiency of reward mechanisms. However, it is necessary to further investigate the genetic, psychological and social variables in placebo responsiveness.

  6. 6) A further challenge is to determine whether socially activated placebo mechanisms emerged during evolution as a defence system of the body (in the same way as the immune system and wound healing evolved). Are placebo and placebo-like effects related to trust, beliefs and hope that emerged in social groups?


Bibliography references:

Ader R (1997). The role of conditioning in pharmacotherapy. In: A Harrington, ed. The Placebo Effect: An Interdisciplinary Exploration, pp. 138–65. Harvard University Press, Cambridge, MA.

Ader R (2001). Much ado about nothing. Advances in Mind–Body Medicine, 17, 293–95.

Adler LJ, Gyulai FE, Diehl DJ, Mintun MA, Winter PM and Firestone LL (1997). Regional brain activity changes associated with fentanyl analgesia elucidated by positron emission tomography. Anesthesiology and Analgesia, 84, 120–26.

Allan LG and Siegel S (2002). A signal detection theory analysis of the placebo effect. Evaluation and the Health Professions, 25, 410–20.

(p.54) Amanzio M and Benedetti F (1999). Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioning-activated specific subsystems. Journal of Neuroscience, 19, 484–94.

Balint M (1955). The doctor, his patient, and the illness. Lancet, 1, 683–88.

Bandura A (1977). Self-efficacy: toward a unifying theory of behavior change. Psychological Reviews, 84, 191–215.

Bandura A (1997). Self-Efficacy: the Exercise of Control. Cambridge University Press, New York.

Batterman RC (1966). Persistence of responsiveness with placebo therapy following an effective drug trial. Journal of New Drugs, 6, 137–41.

Batterman RC and Lower WR (1968). Placebo responsiveness—influence of previous therapy. Current Therapeutic Research, 10, 136–43.

Bausell RB, Lao L, Bergman S, Lee WL and Berman BM (2005). Is acupuncture analgesia an expectancy effect? Preliminary evidence based on participants' perceived assignments in two placebo-controlled trials. Evaluation and the Health Professions, 28, 9–26.

Beecher HK (1955). The powerful placebo. Journal of the American Medical Association, 159, 1602–06.

Benedetti F (2002). How the doctor's words affect the patient's brain. Evaluation and the Health Professions, 25, 369–86.

Benedetti F (2005). The importance of considering the effects of perceived group assignment in placebo-controlled trials. Evaluation and the Health Professions, 28, 5–6.

Benedetti F (2007). What do you expect from this treatment? Changing our mind about clinical trials. Pain, 128, 193–94.

Benedetti F (2008). Mechanisms of placebo and placebo-related effects across diseases and treatments. Annual Review of Pharmacology and Toxicology, 48, 33–60.

Benedetti F, Arduino C, Costa S, et al. (2006). Loss of expectation-related mechanisms in Alzheimer's disease makes analgesic therapies less effective. Pain, 121, 133–44.

Benedetti F and Colloca L (2004). Placebo-induced analgesia: methodology, neurobiology, clinical use, and ethics. Reviews in Analgesia, 7, 129–43.

Benedetti F, Colloca L, Torre E et al. (2004). Placebo-responsive Parkinson patients show decreased activity in single neurons of subthalamic nucleus. Nature Neuroscience, 7, 587–88.

Benedetti F, Mayberg HS, Wager TD et al. (2005). Neurobiological mechanisms of the placebo effect. Journal of Neuroscience, 25, 10390–402.

Benedetti F, Pollo A, Lopiano L, Lanotte M, Vighetti S and Rainero I (2003). Conscious expectation and unconscious conditioning in analgesic, motor and hormonal placebo/nocebo responses. Journal of Neuroscience, 23, 4315–23.

Bishop S, Duncan J, Brett M and Lawrence AD (2004). Prefrontal cortical function and anxiety: controlling attention to threat-related stimuli. Nature Neuroscience, 7, 184–88.

Bootzin RR (1985). The role of expectancy in behavior change. In: L White, B Tursky and GE Schwartz, eds. Placebo: Theory, Research, and Mechanisms, pp.196–210. Guilford Press, New York.

Bootzin RR and Caspi O (2002). Explanatory mechanisms for placebo effects: cognition, personality and social learning. In: HA Guess, A Kleinman, JW Kusek and LW Engel, eds. The Science of the Placebo: Toward an Interdisciplinary Research Agenda, pp. 108–32. BMJ Books, London.

(p.55) Branthwaite A and Cooper P (1981). Analgesic effects of branding in treatment of headaches. British Medical Journal, 282, 1576–78.

Brody H (2000). The Placebo Response. Harper Collins, New York.

Brody H and Weismantel D (2001). A challenge to core beliefs. Advances in Mind–Body Medicine, 17, 296–98.

Caspi O and Bootzin RR (2002). Evaluating how placebos produce change. Evaluation and the Health Professions, 25, 436–64.

Clark WC (1969). Sensory-decision theory analysis of the placebo effect on the criterion for pain and thermal sensitivity. Journal of Abnormal Psychology, 74, 363–71.

Colloca L and Benedetti F (2005). Placebos and painkillers: is mind as real as matter? Nature Reviews Neuroscience, 6, 545–52.

Colloca L and Benedetti F (2006). How prior experience shapes placebo analgesia. Pain, 124, 126–33.

Colloca L, Finniss DG and Benedetti F (2008). Placebo and nocebo. In: Textbook of Clinical Pain Management, Hodder Arnold, in press.

Colloca L, Lopiano L, Lanotte M and Benedetti F (2004). Overt versus covert treatment for pain, anxiety, and Parkinson's disease. LancetNeurology, 3, 679–84.

Davis CE (2002). Regression to the mean or placebo effect? In: HA Guess, A Kleinman, JW Kusek and LW Engel, eds. The Science of the Placebo: Toward an Interdisciplinary Research Agenda pp. 158–66. BMJ Books, London.

De Jong PJ, van Baast R, Arntz A and Merkelbach H (1996). The placebo effect in pain reduction: the influence of conditioning experiences and response expectancies. International Journal of Behavioral Medicine, 3, 14–29.

de la Fuente-Fernandez R, Phillips AG, Zamburlini, et al. (2002). Dopamine release in human ventral striatum and expectation of reward. Behavioural Brain Research, 136, 359–63.

de la Fuente-Fernandez R, Ruth TJ, Sossi V, Schulzer M, Calne DB and Stoessl AJ (2001). Expectation and dopamine release: mechanism of the placebo effect in Parkinson's disease. Science, 293, 1164–66.

De Pascalis V, Chiaradia C and Carotenuto E (2002). The contribution of suggestibility and expectation to placebo analgesia phenomenon in an experimental setting. Pain, 96, 393–402.

Di Blasi Z, Harkness E, Ernst E, Georgiou A and Kleijnen J (2001). Influence of context effect on health outcomes: a systematic review. Lancet, 357, 757–62.

Di Nubile MJ (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1278.

Diehl HS (1953). Medicinal treatment of the common cold. Journal of the American Medical Association, 101, 2042–49.

Eccles R (2006). Mechanisms of the placebo effect of sweet cough syrups. Respiratory Physiology and Neurobiology, 152, 340–48.

Einarson TE, Helmes M and Stolk P (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1277.

Evans D (2002). Pain, evolution, and the placebo response. Behavioral and Brain Sciences, 25, 459–60.

Evans D (2003). Placebo: The Belief Effect. Harper Collins, London.

(p.56) Evans FJ (1977). The placebo control of pain: A paradigm for investigating non-specific effects in psychotherapy. In: JP Brady, J Mendels, WR Reiger and MT Orne, eds. Psychiatry: Areas of Promise and Advancement, pp. 249–71. Plenum Press, New York.

Feather BW, Chapman CR and Fisher SB (1972). The effect of a placebo on the perception of painful radiant heat stimuli. Psychosomatic Medicine, 34, 290–94.

Fields HL and Levine JD (1984). Placebo analgesia—a role for endorphins? Trends in Neuroscience, 7, 271–73.

Firestone L, Gyulai F, Mintun M, Adler L, Urso K and Winter P (1996). Human brain activity response to fentanyl imaged by positron emission tomography. Anesthesia and Analgesia, 82, 1247–51.

Frank JD (1961). Persuasion and Healing: a Comparative Study of Psychotherapy.Schocken Books, New York. Johns Hopkins University Press, Baltimore.

Frank JD (1971). Therapeutic factors in psychotherapy. American Journal of Psychotherapy, 25, 350–61.

Frank JD (1981). Therapeutic components shared by all psychotherapies. In: JH Hawey and MM Parks, eds. Psychotherapy Research and Behavior Change. American Psychological Association, Washington, DC.

Geers AL, Helfer SG, Kosbab K, Weiland PE and Landry SJ (2005b). Reconsidering the role of personality in placebo effects: dispositional optimism, situational expectations, and the placebo response. Journal of Psychosomatic Research, 58, 121–27.

Geers AL, Kosbab K, Helfer SG, Weiland PE and Wellman JA (2007). Further evidence for individual differences in placebo responding: an interactionist perspective. Journal of Psychosomatic Research, 62, 563–70.

Geers AL, Weiland PE, Kosbab K, Landry SJ and Helfer SG (2005a). Goal activation, expectations, and the placebo effect. Journal of Personality and Social Psychology, 89, 143–59.

Gracely RH, Dubner R, Deeter WD and Wolskee PJ (1985). Clinicians' expectations influence placebo analgesia. Lancet, 1, 43.

Greene PJ, Wayne PM, Kerr CE, et al. (2001). The powerful placebo: doubting the doubters. Advances in Mind–Body Medicine, 17, 298–307.

Hahn RA (1985). A sociocultural model of illness and healing. In: L White, B Tursky, GE Schwartz, eds. Placebo: Theory, Research, and Mechanisms, pp. 332. Guilford, New York.

Hahn RA (1997). The nocebo phenomenon: concept, evidence, and implications for public health. Preventive Medicine, 26, 607–11.

Harrington A (2002). ‘Seeing’ the placebo effect: historical legacies and present opportunities. In: HA Guess, A Kleinman, JW Kusek, LW Engel, eds. The Science of the Placebo: Toward an Interdisciplinary Research Agenda, pp. 35–52. BMJ Books, London.

Herrnstein RJ (1962). Placebo effect in the rat. Science, 138, 677–78.

Hrobjartsson A and Goetzsche PC (2001). Is the placebo powerless? An analysis of clinical trials comparing placebo with no treatment. New England Journal of Medicine, 344, 1594–602.

Hrobjartsson A and Goetzsche PC (2004). Is the placebo powerless? Update of a systematic review with 52 new randomized trials comparing placebo with no treatment. Journal of Internal Medicine, 256, 91–100.

Hrobjartsson A and Goetzsche PC (2006). Unsubstantiated claims of large effects of placebo on pain. Serious errors in meta-analysis of placebo analgesia mechanism studies. Journal of Clinical Epidemiology, 59, 336–38.

(p.57) Hrobjartsson A and Goetzsche PC (2007). Powerful spin in the conclusion of Wampold et al.'s re-analysis of placebo versus no-treatment trials despite similar results as in original review. Journal of Clinical Psychology, 63, 373–77.

Humphrey N (2002). Great expectations: the evolutionary psychology of faith-healing and the placebo effect. In: C von Hofsten, L Backman, eds. Psychology at the Turn of the Millennium, Vol. 2: Social, Developmental, and Clinical Perspectives, pp. 225–46. Psychology Press, Hove.

Hunsley J and Westmacott R (2007). Interpreting the magnitude of the placebo effect: mountain or molehill? Journal of Clinical Psychology, 63, 391–99.

Kalivas PW, Churchill L and Romanides A (1999). Involvement of the pallidal–thalamocor-tical circuit in adaptive behavior. Annals of the New York Academy of Sciences, 877, 64–70.

Kaptchuk TJ (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1277.

Keats AS and Beecher HK (1950). Pain relief with hypnotic doses of barbiturates and a hypothesis. Journal of Pharmacology and Experimental Therapeutics, 100, 1–13.

Keats AS, D’ Alessandro GL and Beecher HK (1951). A controlled study of pain relief by intravenous procaine. Journal of the American Medical Association, 147, 1761–63.

Kennedy WP (1961). The nocebo reaction. Medicina Experimentalis International Journal of Experimental Medicine, 95, 203–05.

Kienle GS and Kiene H (1997). The powerful placebo effect: fact or fiction? Journal of Clinical Epidemiology, 50, 1311–18.

Kirsch I (1985). Response expectancy as determinant of experience and behavior. American Psychologist, 40, 1189–1202.

Kirsch I (1990). Changing Expectations: A Key to Effective Psychotherapy. Brooks Cole, Pacific Grove, CA.

Kirsch I (1999). How Expectancies Shape Experience. American Psychological Association, Washington, DC.

Kirsch I, Lynn SJ, Vigorito M and Miller RR (2004). The role of cognition in classical and operant conditioning. Journal of Clinical Psychology, 60, 369–92.

Kirsch I and Scoboria A (2001). Apples, oranges, and placebos: heterogeneity in a meta-analysis of placebo effects. Advances in Mind–Body Medicine, 17, 307–09.

Kissel P and Barrucand D (1964). Placebos et Effect Placebo en Medicine. Masson, Paris.

Kleinman A, Guess HA and Wilentz JS (2002). An overview. In: HA Guess, A Kleinman, JW Kusek and LW Engel, eds. The Science of the Placebo Effect: Toward anInterdisciplinary Research Agenda, pp. 1–32. BMJ Books, London.

Kupers R (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1278.

Laska E and Sunshine A (1973). Anticipation of analgesia: a placebo effect. Headache, 13, 1–11.

Last JM (1983). A Dictionary of Epidemiology. Oxford University Press, New York.

Lazarus RS (1991). Cognition and motivation in emotion. American Psychologist, 46, 352–67.

Levesque J, Eugene F, Joanette Y, et al. (2003). Neural circuitry underlying voluntary suppression of sadness. Biological Psychiatry, 53, 502–10.

Lichstein J, DeCosta Mayer J and Hauch EW (1955). Efficacy of methantheline (banthine) bromide in therapy of the unstable colon. Journal of the American Medical Association, 25, 634–37.

(p.58) Lieberman MD, Jarcho JM, Berman S, et al. (2004). The neural correlates of placebo effects: a disruption account. NeuroImage 22:447– 55.

Lieberman MD, Hariri A, Jarcho JM, Eisenberger NI and Bookheimer SY (2005). An fMRI investigation of race-related amygdala activity in African–American and Caucasian–American individuals. Nature Neuroscience, 8, 720–22.

Lilford RJ and Braunholtz DA (2001). Is the placebo powerful? New England Journal of Medicine, 345, 1277–78.

Lipid Research Clinics Program (1984). The lipid research clinics coronary primary prevention trial results. Journal of the American Medical Association, 25, 351–64.

Lipman JJ, Miller BE, Mays KS, et al. (1990). Peak B endorphin concentration in cerebrospinal fluid: reduced in chronic pain patients and increased during the placebo response. Psychopharmacology, 102, 112–16.

Mayberg HS, Silva JA, Brannan SK, et al. (2002). The functional neuroanatomy of the placebo effect. American Journal of Psychiatry, 159, 728–37.

McGlashan TH, Evans FJ and Orne MT (1969). The nature of hypnotic analgesia and placebo response to experimental pain. Psychosomatic Medicine, 31, 227–46.

Miller EK and Cohen JD (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167–202.

Miller FG (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1277.

Moerman DE (2002). Meaning, Medicine and the Placebo Effect. Cambridge University Press, Cambridge.

Moerman DE (2003). Doctors and patients: the role of clinicians in the placebo effect. Advances in Mind–Body Medicine, 19, 14–22.

Moerman DE and Jonas WB (2002). Deconstructing the placebo effect and finding the meaning response. Annals of Internal Medicine; 136, 471–6.

Mogenson GJ and Yang CA (1991). The contribution of basal forebrain to limbic–motor integration and the mediation of motivation to action. Advances in Experimental Medicine and Biology, 295, 267–90.

Montgomery GH and Kirsch I (1997). Classical conditioning and the placebo effect. Pain, 72, 107–13.

Ochsner KN, Bunge SA, Gross JJ and Gabrieli JD (2002). Rethinking feelings: an fMRI study of the cognitive regulation of emotion. Journal of Cognitive Neuroscience, 14, 1215–29.

Ochsner KN, Ray RD, Cooper JC, et al. (2004). For better or for worse: neural systems supporting the cognitive down- and up-regulation of negative emotion. Neuroimage, 23, 483–99.

Papakostas YG and Daras MD (2001). Placebos, placebo effect, and the response to the healing situation: the evolution of a concept. Epilepsia, 42, 1614–25.

Pennebaker JW (1997). Writing about emotional experiences as a therapeutic process. Psychological Science, 8, 162–66.

Petrovic P, Dietrich T, Fransson P, Andersson J and Carlsson K (2005). Placebo in emotional processing-induced expectations of anxiety relief activate a generalized modulatory network. Neuron,46, 957–69.

Petrovic P and Ingvar M (2002). Imaging cognitive modulation of pain processing. Pain, 95, 1–5.

(p.59) Petrovic P, Kalso E, Petersson KM and Ingvar M (2002). Placebo and opioid analgesia—imaging a shared neuronal network. Science, 295, 1737–40.

Phan KL, Fitzgerald DA, Nathan PJ, Moore GJ, Uhde TW and Tancer ME (2005). Neural substrates for voluntary suppression of negative affect: a functional magnetic resonance imaging study. Biological Psychiatry, 57, 210–19.

Price DD and Barrell JJ (2000). Mechanisms of analgesia produced by hypnosis and placebo suggestions. Progress in Brain Research, 122, 255–71.

Price DD, Barrell JE and Barrell JJ (1985). A quantitative–experiential analysis of human emotions. Motivation and Emotion, 9, 19–38.

Price DD, Finniss DG and Benedetti F (2008). A comprehensive review of the placebo effect: recent advances and current thought. Annual Review of Psychology, 59, 565–90.

Price DD, Riley J and Barrell JJ (2001). Are lived choices based on emotional processes? Cognition and Emotion, 15, 365–79.

Rainville P, Duncan GH, Price DD, Carrier B and Bushnell MC (1997). Pain affect encoded in human anterior cingulate but not somatosensory cortex. Science, 277, 968–71.

Reiss S (1980). Pavlovian conditioning and human fear: an expectancy model. Behavioral Therapy, 11, 380–96.

Rescorla RA (1988). Pavlovian conditioning: it's not what you think it is. American Psychologist, 43, 151–60.

Roberts AH (1995). The powerful placebo revisited: the magnitude of non-specific effects. Mind–Body Medicine, 3, 1–10.

Ruck A and Sylven C (2006). ‘Improvement’ in the placebo group could be due to regression to the mean as well as to sociobiologic factors. American Journal of Cardiology, 97, 152–53.

Sackett DL (1995). Randomized trials in individual patients. In: G Antes, L Edler, R Holle, W Koepcke, R Lorenz and J Windeler, eds. Biometrie und Unkonventionelle Medizine, pp. 19–33. Munster-Hiltrup, Landwirtschaftsverlag.

Schultz W (2006). Behavioral theories and the neurophysiology of reward. Annual Review of Psychology, 57, 87–115.

Scott DJ, Stohler CS, Egnatuk CM, Wang H, Koeppe RA and Zubieta J-K (2007). Individual differences in reward responding explain placebo-induced expectations and effects. Neuron, 55, 325–36.

Setlow B, Schoenbaum G and Gallagher M (2003). Neural encoding in ventral striatum during olfactory discrimination learning. Neuron,38, 625–36.

Shapiro AK and Shapiro E (1997). The placebo: Is it much ado about nothing. In: A Harrington, ed. The Placebo Effect: An Interdisciplinary Exploration, pp.12–36. Harvard University Press, Cambridge, MA.

Shrier I (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1278.

Siegel S (1972). Conditioning of insulin-induced glycemia. Journal of Comparative Physiology and Psychology, 78, 233–41.

Siegel S (1975). Conditioning insulin effects. Journal of Comparative Physiology and Psychology, 89, 189–99.

(p.60) Siegel S (2002). Explanatory mechanisms for placebo effects: Pavlovian conditioning. In: HA Guess, A Kleinman, JW Kusek, LW Engel, eds. The Science of the Placebo: Toward an Interdisciplinary Research Agenda, pp. 133–57. BMJ Books, London.

Spiegel D, Kraemer H and Carlson RW (2001). Is the placebo powerless? New England Journal of Medicine, 345, 1276.

Stewart-Williams S and Podd J (2004). The placebo effect: dissolving the expectancy versus conditioning debate. Psychological Bulletin, 130, 324–40.

Sunshine A, Laska E, Meisner M and Morgan S (1964). Analgesic studies of indomethacin as analyzed by computer techniques. Clinical Pharmacology and Therapeutics, 5, 699–707.

Thomas KB (1987). General practice consultations: is there any point in being positive? British Medical Journal, 294, 1200–02.

Tobler PN, Fiorillo CD and Schultz W (2005). Adaptive coding of reward value by dopamine neurons. Science, 307, 1642–45.

Tolman EC (1932). Purposive Behavior in Animals and Men. Appleton Century Crofts, New York.

Travell J, Rinzler SH, Bakst H, Benjamin ZH and Bobb A (1949). Comparison of effects of alpha-tocopherol and a matching placebo on chest pain in patients with heart disease. Annals of the New York Academy of Sciences, 52, 345–53.

Vase L, Riley JL 3rd and Price DD (2002). A comparison of placebo effects in clinical analgesic trials versus studies of placebo analgesia. Pain, 99, 443–52.

Vase L, Robinson ME, Verne GN and Price DD (2005). Increased placebo analgesia over time in irritable bowel syndrome (IBS) patients is associated with desire and expectation but not endogenous opioid mechanisms. Pain, 115, 338–47.

Voudouris NJ, Peck CL and Coleman G (1989). Conditioned response models of placebo phenomena: further support. Pain, 38, 109–16.

Voudouris NJ, Peck CL and Coleman G (1990). The role of conditioning and verbal expectancy in the placebo response. Pain, 43, 121–28.

Waber RL, Shiv B, Carmon Z and Ariely D (2008). Commercial features of placebo and therapeutic efficacy. Journal of the American Medical Association, 299, 1016–17.

Wager TD, Reading S and Jonides J (2004a). Neuroimaging studies of shifting attention: a meta-analysis. Neuroimage, 22, 1679–93.

Wager TD, Billing JK, Smith EE, et al. (2004b). Placebo-induced changes in fMRI in the anticipation and experience of pain. Science, 303, 1162–66.

Wager TD and Smith EE (2003). Neuroimaging studies of working memory: a meta-analysis. Cognitive, Affective and Behavioral Neuroscience, 3, 255–74.

Wagner KJ, Willoch F, Kochs EF, et al. (2001). Dose-dependent regional cerebral blood flow changes during remifentanil infusion in humans: a positron emission tomography study. Anesthesiology, 94, 732–39.

Wall PD (1999). Pain: The Science of Suffering. Weidenfeld and Nicholson, London.

Wampold BE, Minami T, Tierney SC, Baskin TW and Bhati KS (2005). The placebo is powerful: estimating placebo effects in medicine and psychotherapy from randomized clinical trials. Journal of Clinical Psychology, 61, 835–54.

Wells RE (1957). Use of reserpine (Serpasil) in the management of chronic alcoholism. Journal of the American Medical Association, 163, 426–29.

(p.61) Wickramasekera I (2001). The placebo efficacy study: problems with the definition of the placebo and the mechanisms of placebo efficacy. Advances in Mind–Body Medicine, 17, 309–12.

Williams AC (2002). Facial expression of pain: an evolutionary account. Behavioral and Brain Sciences, 25, 439–55. (p.62)