This essay examines two questions: First, "What is the placebo effect?" (i.e. how can we define it?). Second, "How does the placebo effect work?" (i.e. what mechanisms lie behind it?).
Before proposing a definition of the placebo effect, let me make four points which I believe have important bearings on the definition.
Bearing these points in mind, I propose the following definition:
The placebo effect is defined as the psychological and/or physiological changes that result from the administration of a physiologically inert treatment, or the inert part of a physiologically active treatment; its efficacy relying on the attitudes (cognitive, affective and behavioural) of the person involved to both their condition and the treatment, and which can facilitate powerful and long-lasting improvements in people's health over a wide range of illnesses or perceived illnesses.
Among possible omissions from this definition are any reference to the attitudes of the healer, and the relationship between the patient and healer; factors that have been shown to be key elements in the effectiveness of the placebo treatment (Benson & Epstein, 1975). However, the inclusion of these factors would, I suggest, bias the definition to certain causal mechanisms rather than others, and, more importantly, would exclude cases where a patient-healer interaction was not directly involved but which would fulfil the other criteria defined above (e.g. gaining relief from a cold remedy bought over the counter at the local chemist, or from a herbal remedy obtained from a friend who had bought it on holiday).
Giving a definition which is both necessary and sufficient is well nigh impossible since the placebo effect is almost certainly not one effect, but a blanket term for a complex of causes, mechanisms and outcomes. Unfortunately, giving a single label to a range of effects only encourages the view that this is a single effect with a single causal mechanism. In the words of Richardson (1989) a "spurious impression of homogeneity" is created. Bearing this is mind, let us move on to look at the mechanisms that may be involved.
Many ideas for placebogenesis have been proposed, mostly involving the mechanisms by which the patient's mental state is altered depending on such factors as his/her attitudes to the illness, to himself/herself, to the placebo, to the physician and to the general healing environment (Shapiro & Morris, 1978). However, little is usually said about how these psychological changes lead to biophysiological changes. What actually is it that mediates between, say, reduced feelings of guilt and improved lung efficiency?
Some studies have, of course, examined psychophysiological processes. For example, Levine et al (1978) showed that placebo analgesia for dental post-operative pain (self-reported) was mediated via endorphins (though Wall, 1993, has raised some important provisos). Since it is unlikely that endorphins are the mechanism whereby all placebo effects operate, this evidence may help place one piece in the jigsaw, but only highlights the fact that this is a multidimensional effect. Though Levine et al utilised a double-blind procedure, the problem with placebo analgesia research is that self-reported pain can be open to reporting errors and to self-misattribution. This methodological difficulty is even more of a problem for illnesses which primarily involve psychological symptoms (e.g. schizophrenia and depression). For this reason, and due to lack of space, I will concentrate on those putative mechanisms that cause objectively measurable physiological benefits: e.g. lung function (Butler & Steptoe, 1984) and blood pressure (Vogel et al, 1980). However, I hope most of the proposed ideas will cover psychological illnesses.
Although not explicitly stated in the main literature on placebogenesis, it seems likely that a placebo response follows (either fully, or to some degree) the following three stages:
Stage 1: The placebo treatment causes a significant change in mental state. This may be:-
(i) a reduction in the level of negative attitudes (e.g. guilt, anxiety, fear, pessimism, cognitive conflict, helplessness, deleterious behavioural intentions); and/or
(ii) an enhancement of positive attitudes (e.g. hope, faith, calm, optimism, harmony, control, beneficial behavioural intentions).
Stage 2: The more positive (or less negative) state of mind produces:-
(i) physiological changes via one or more psychoneuroimmunological pathway(s) (e.g. increased immune efficiency, lower corticosteroid levels, altered ANS activity, endorphin release); and/or
(ii) adoption of healthier behaviours (e.g. better food, less alcohol, less cigarettes, more exercise).
Hence, there is either a movement towards normal levels (homeostatic set-points) from an unhealthy state, or a movement away from normal levels towards an enhanced state of well-being (c.f. Antonovsky's continuum model, 1979).
Stage 3: The perceived improvements in symptoms, or those objectively measured and fed back to the patient, caused by stage 2 (in addition to the perceived improvements in symptoms caused by stage 1) lead to an even more positive state of mind. Hence, a positive cycle around stages 1 and 2 is set up (getting better - feeling better - getting better etc.) until some limit is reached set by biopsychosocial factors.
Let me briefly look in more detail at the mechanisms that may be operating in stages 1 and 2, based on current/available evidence.
Shapiro and Morris (1978) identified three broad groupings in which to categorise the many processes of placebogenesis that could produce the psychological changes in stage 1. These are described below.
Social influence effects include suggestion, persuasion, transference, role demands, operant conditioning and guilt reduction. Emphasis here is on the role of the healer, whether this be in the guise of physician, therapist or experimental investigator. The way the patient gains information, gains emotional support and gains feedback (rewards/punishments), can all play a part in their psychological changes (e.g. guilt reduction due to the therapist's interest and concern). The normative role demands expected of "a patient" can also be key.
Expectancy effects include classical conditioning, cognitive dissonance, internal standards and hope/faith. Here, the emphasis is on the part played by the patient's preconceived ideas about the treatment and prognosis based on their previous experience (whether direct or indirect). For instance, there is evidence which indicates that both length of illness and history of taking medication are both negatively correlated with placebo response (Rickels et al, 1966).
Evaluation effects include response artefacts, labelling and misattribution. These effects suggest an interesting route by which stage 3 could be reached. Suppose, after treatment, the patient perceives some improvement in symptoms - which may, in fact, be due to factors "external" to the treatment e.g. natural remission. Not only should the patient feel better due to the improvement, but by attributing the improvement to the placebo treatment, their confidence in the treatment should improve, thus triggering the positive cycle of stage 3.
Moving to mechanisms that may facilitate stage 2, it seems likely that these operate through the same pathways involved in producing psychosomatic illnesses. Indeed, it has been noted that placebo responders are more likely to give a history of psychosomatic symptoms than non-responders (Lewith, 1993) - although most papers report no correlation between placebo efficacy and patient personality (Wall, 1993). Work on psychosomatic illnesses (recently re-christened "psychophysiological disorders") has now largely been subsumed under the blossoming subject of psychoneuroimmunology, which has paid particular attention to the deleterious effects of stress. Evidence for this is abundant: for example, stress has been shown to increase levels of catecholamines and corticosteroids which can damage arteries and the heart, promote atherosclerosis, hypertension and arteriosclerosis, and cause cardiac arrhythmia (Sarafino, 1994). Stress can also cause behavioural changes which seriously disrupt the biological system (e.g. smoking, and excessive drinking of coffee and alcohol). Stress also affects the immune system. For instance, it has recently been show that a brief (40 minutes) period of emotional arousal can lead to transient impairment of lymphocyte reactivity (Knapp et al, 1992), and that a high number of daily hassles are associated with stressor-induced decreases in T cell and NK cell counts (Brosschot et al, 1994). Stress is not the only psychological process that affects the immune system, the impact may be far more broad and pervasive. For example, it has been shown that antibody-mediated and cell-mediated immune processes can be classically conditioned to the intake of saccharin water in rats (Ader & Cohen, 1985).
As research work progresses in discovering more about the psychophysiological pathways that lead to illness, so light will be shed on the mechanisms that mediate the organic placebo responses. Also, as the Biopsychosocial framework is more widely recognised, so will the understanding of how psychological, physiological and behavioural factors interact to create healthy, or unhealthy, conditions.
To summarise, then, placebo treatments can be powerful instigators of a positive cycle of psychological and physiological change which can result in disease remission. The lesson for healers (and the services that support them) should be to treat the patient holistically, with care and compassion, so that positive psychological changes can occur. The main aim of research should be to uncover the psychoneuroimmunological processes that mediate the placebo effect. But one should not forget that truth may not be enough - perhaps a little mystery, magic and ignorance (!) are vital requirements if the healing process is to fully tap the power of the mind over the body.
Antonovsky, A. (1979). Health, Stress and Coping. San Francisco: Jossey-Bass.
Benson, H., & Epstein, M. D. (1975). The Placebo effect: a neglected asset in the care of patients. Journal of the American Medical Association, 232:12, 1225-1227.
Broschott, J. F., Benschop, R. J., et al (1994). Influences of life stress on immunological reactivity to mild psychological stress. Psychosomatic Medicine, 56(3), 216-224.
Knapp, P. H., Levy, E. M., Giorgi, R. G. et al (1992). Short-term immunological effects of induced emotion. Psychosomatic Medicine, 54, 133-148.
Levine, J. D., Gordon, N. C. & Fields, H L. (1978). The Mechanism of placebo analgesia. The Lancet, 1978, 2: 654-657.
Lewith, G. T. (1993). Every doctor a walking placebo. In Clinical research methodology for complementary therapies, Lewith & Aldridge (eds.), Hodder & Stoddon
Richardson, P. (1989). Placebos: their effectiveness and modes of action. In Health Psychology: Processes and Applications, A.K. Broome(ed), Chapman and Hall.
Sarafino, E. P. (1994). Health psychology; biopsychosocial interactions (2nd ed), John Wiley.
Shapiro, A. K. & Morris, L. A. (1978). The placebo effect in medical and psychological therapies. In Handbook of Psychotherapy and Behavior Change, (Bergin & Garfield, eds.). New York: Wiley.
Wall, P. D. (1993). Pain and the placebo response. In Experimental and theoretical studies of consciousness. Chichester: Wiley.