The Monoamine Hypothesis, Placebos and Problems of Theory Construction in Psychology, Medicine, and Psychiatry
Paul C. L. Tang
Can there ever be scientific theories in psychology, medicine, or psychiatry? I argue that one approach to answering this question consists of investigating the nature of such purported theories and I focus on the monoamine hypothesis of clinical depression. (1) By a careful examination of a typical experiment involving the action of drugs and placebos for the cure of clinical depression-an experiment founded upon the monoamine hypothesis-I raise a number of methodological and conceptual problems that may lead one to conclude that rigorous scientific theories in these three disciplines may never be forthcoming, or at least that the state of scientific research in these areas is still in very rudimentary shape and in need of much logical and philosophical analysis. In addition, because no such biomedical theories may be forthcoming, I also undercut Adolf Grünbaum's analysis of the concept of a placebo, an analysis that makes the definition of "placebo" relative to a biomedical theory in the first place.
I begin in section 2 by discussing Grünbaum's detailed analysis of what a placebo is. I then use his idea of a generic intentional Placebo and discuss one of many similar experiments concerning the phenomenon of clinical depression, experiments that use the monoamine hypothesis and the notion of a generic intentional placebo. This experiment is described in section 3. In section 4 I analyze the results of this study and in section 5 I offer concluding remarks.
II. Grünbaum's Analysis of Placebo
Exactly what is a placebo? Adolf Grünbaum offers an important analysis
(Grünbaum, 1985). He characterizes the Grünbaum literature on placebos as "a veritable Tower of Babel" (Grünbaum, 1985, p. 9). While the focus of my paper is not on the problem of formulating the proper definition of "placebo", I need to use a working definition of "placebo" that at least comports with the term's use in the biomedical literature. To accomplish this, I shall begin with a reconstruction of Grünbaum's account of placebo (Grünbaum, 1985). He presents the following index of symbols:
Grünbaum's explication of "placebo" highlights the genus placebo first, and then specifies the two species, viz., intentional placebo and inadvertent placebo.
Genus: t is a "generic placebo" with respect to D iff:
Grünbaum's account relativizes the placebo species to both the target disorder, D, and to the therapeutic theory, TT, that articulates the specific treatment modality, t. The TT (in this case, monoamine hypothesis of treatment clinical depression) lists/defines the F (characteristic factors of t) and C (incidental factors of t) for a given t (e.g., drug therapy). Under this conception, a prescribed drug could be a placebo for one treatment while a nonplacebo for another.
III. A Case Study
Let us now look more closely at a typical experiment that results in a placebo response. I take my study from the British Journal of Psychiatry. The article is entitled "A Double-Blind Comparative Trial of Moclobemide v. Imipramine and Placebo in Major Depressive Episodes" by M. Versiani, et. als. (Versiani et al, 1989). Clinical Depression is a particularly fruitful example for the discussion of placebos because of the documented large placebo response rate for patients with this malady (Zavodnick, p. 276). The sense of "placebo" used in these familiar studies is Grünbaum's sense of generic placebo, species intentional, discussed above.
Let us begin with the monoamine hypothesis. Physiologists have understood for some time the effects on patients of Monoamine Oxidase Inhibitors (MAO-I's) and Tricyclic antidepressants (TCA's). The effectiveness of such types of drugs is well reviewed in the biomedical and psychological literature. The drugs developed out of the monoamine hypothesis for depression, which holds that depression is caused by insufficient activity of monoaminergic neurons (Carlson 1988, p. 478-480). Both the MAO-I's and TCA's remedy this situation, although they do so in different ways. MAO-I's, like moclobemide, inhibit the enzyme MAO, an enzyme which destroys the monoamines norepinephrine and serotonin, which are neurotransmitters. The result is an increase in monoamines.
TCA'S, on the other hand, inhibit neurotransmitter re-uptake of monoaminergic neurons, thereby causing a similar increase in the monoamines norepinephrine and serotonin. The theoretical result in either case is a decrease in depression due to the effect of the increase of neurtransmitter monoamines by either the MAO-1 drugs or the TCA drugs.
The study in question was designed to compare the efficacy of MAO-I's and TCA's on the reduction of depression. A placebo group was used but was not the focus, i.e., the placebo was not a viable therapeutic option.
Rather, the placebo acted as the marker by which to judge the effectiveness of moclobemide and imipramine. The reasoning behind this methodology is that the drug must accrue a high rate of improvement over the placebo in order to be considered an effective pharmacological treatment.
In this study, 490 patients were enrolled in a double-blind, prospective, randomized, 6-week multi-center study. Participants were men and non-pregnant and non-lactating women, aged 18 to 65. A requirement for inclusion was meeting the DSM-III (Diagnostic and Statistical Manual of Mental Disorders, third edition) (2) criteria for Major Depressive Episode and to have a minimum baseline score of 17 on the 21-item Hamilton Rating Scale for Depression (HRSD). (3) Patients were randomly assigned to a study center and given either moclobemide (n=164), imipramine (n=164), or placebo (n=162). All patients received capsules of identical appearance and taste.
The efficacy of the different conditions was judged primarily on the HRSD administered at the end of the six weeks. As judged by the HRSD exit scores, both moclobemide and imipramine were more effective than placebos in the reduction of major depression. The criterion for reduction was a >50% 1 improvement on the HRSD. The mean final percentage reduction of total HRSD score was 52% with moclobemide, 56% with imipramine, and 28% with the placebo. The mean final improvement was 61.8% for the moclobemide group, 66.4% for the imipramine group, and 37.3% for the placebo group. Differences in both assessments between
moclobemide and imipramine on the one hand and placebo on the other were statistically significant.
IV. Five Methodological Problems
I begin by raising five important methodological problems about this typical experiment. First, there is not a great difference in improvement rate of the clinically depressed patients between and among the three groups (61.8% for moclobemide, 66.4% for imipramine, and 37.3% for the placebo groups). True, the main purpose of the study was to compare the efficacy of moclobemide and imipramine in relieving clinical depression. However, from a philosophical, theoretical or even methodological perspective, what is missing or consistently overlooked in these studies is an adequate and rigorous account of the interpretation of the data in terms of scientific explanation and, in particular, in terms of causal explanation. The monoamine hypothesis is an attempt to explain-in terms of anatomy, physiology and biochemistry-the cause of clinical depression. But as I shall note in my subsequent questions, a close examination of the hypothesis and the experimental data reveals serious weaknesses about both the hypothesis itself and about the scientific and experimental methodology involved. Second, all these comparative studies leave unanswered the precise, detailed mechanism of the causal nature of the placebo. If the drug imipramine, for example, is considered causally efficacious for the 66.8% who improved, why is not the placebo also considered equally causally efficacious for the 37.3% who also improved?
Third, since the drugs moclobemide and imipramine and the placebo all resulted in a marked rate of patient improvement, could I not propose a third, common cause as the efficient cause for all three trials? Fourth, what can be said about the fairly large percentage of patients who did not improve, whether they took moclobemide, or imipramine or the placebo? How does one explain these results? Fifth, we are all aware that parts of the human body often heal themselves, e.g., when one slightly cuts his or her finger. This phenomenon is called, in medicine, the vis medacatrix naturea. Over time, the body works to restore equilibrium via "recuperative powers." This is a common explanation. (See Neuberger, 1932, cited in Brody, 1985). But can this explanation be used for the clinically depressed patients in the above study who improve and yet have taken only the placebo? How does one justify using vis medacatrix naturea or "the body's natural healing power" as an explanation for the placebo effect, since, at the very least, there is no clear criterion as to why the vis medactrix naturea did not manifest itself in those placebo patients who did not improve, nor for those patients who took either drug and also did not improve. If one uses the vis medacatrix naturea explanation at all, the explanation tends to become circular in the sense that the explanation of why all these patients did not improve is because the vis medacrtix natures was not operative, and for the placebo patients who did improve, the vis medacatrix naturea was operative. But all this does not say much, if anything. Indeed, this type of "explanation" reminds one of a character in one of Molière's plays: When Molière's hero is asked why the sleeping pills worked, he replies that the answer is quite obvious that the pills have dormative powers! And, presumably, if Molière's hero took some other pills that did not cause him to fall asleep, he would say that these latter pills did not have dormative powers! So, too, with the vacuousness of explanations involving the body's "recuperative powers." These five criticisms are closely intertwined in their ramifications for scientific methodology and theory construction, such as in the monoamine hypothesis, which forms the basis of the theory of clinical depression. The five points emphasize at least the need for a deeper analysis as to whether explanations in terms of the monoamine hypothesis are unfalsifiable,bordering on ad hoc, or whether they are ceteris paribus. (4)
Consider the fourth and fifth criticisms noted above concerning the non-cure rates of both moclobemide and imipramine. If these drugs restore the levels of norephinepherine and seratonin, why do they not restore the levels for the marked number of patients who do not improve? At the very least the non-cure rate of these patients might be said to refute the monoamine hypothesis, or at least to cause the hypothesis to be amended, or to cause it to be supplemented, or to cause the investigator to propose and test new hypothesess not based on the monoamine hypothesis. Indeed, what would count as a refutation of the monoamine hypothesis? Typically, however, no answer is given and no further hypothesis proposed. These questions raised here should be the impetus for the construction of new hypotheses for clinical depression in the context of scientific discovery.
A similar conceptual and methodological problem plagues the placebo group. For the significant percentage of patients who did improve we not only do not have an adequate causal explanation in terms of the monoamine hypothesis, but we also do not have an adequate explanation in terms of the vis medacatrix naturea. To repeat, to say that in the placebo cure group the body's recuperative powers took hold, while in the placebo non-cure group the body's recuperative powers did not hold, is just to commit the circular fallacy exemplified by Molière's hero. So, whether considering the two cases of drugs and/or the case of placebo, the monoamine hypothesis on which this study of drug therapy is founded seems unfalsifiable, and therefore unscientific in Karl Popper's sense, since the number of "non-cures" is never taken to falsify the monoamine hypothesis, and no further hypothesis is proposed as to why these patients remain clinically depressed. The cases of non-improved patients for the drug groups are not explained nor, for that matter, is the non-cure rate for the placebo patients explained under the monoamine hypothesis.
These problems must be addressed and solved if the biomedical theory of major depression is even to begin to qualify as scientific. The way the scientific studies are conducted now seems to indicate that the conceptual and methodological problems just raised may not be solvable. For if the experimenter's response to the non-cure patients is that the situation is too complex and that he or she cannot isolate any independent variables to construct further hypotheses about the cause of major depression, then this answer requires careful scrutiny and philosophical reflection as to whether the investigator is not so much "saving the phenomenon" as "saving the theory" (at all cost), i.e., the monoamine hypothesis. If the latter, then they are making their theory (hypothesis) unfalsifiable, bordering on being ad hoc in character. To clarify this criticism further: an analogous argument arises in defenses of neo-Darwinism. If the neo-Darwinist evolutionary theorist sees gaps in the fossil record in a certain line of speciation that would seem to falsify the Darwinian claim of continuous (small, incremental) variation, all the neo-Darwinist has to say is that missing fossils may be in that part of the fossil record that we do not have, or that we may never have, thus allowing him or her to maintain the Darwinian claim of continuous variation at all cost. But this ultimate unfalsifiability of neo-Darwinism is a built-in weakness of the theory. Leaving aside a critique of neo-Darwinism, which would take us too far afield, the same logic applies here. To maintain the monoamine hypothesis in the light of so many cases of non-cures that might claim to falsify the hypothesis, the experimenter might say that the complex situation doesn't allow isolation of any other variable for test. And if he were to say further that, if we were to isolate and investigate the non-cure cases, we would see that the monoamine hypothesis does work, we would have a case of an explanation that comes close to being ad hoc. He would be trying to save the monoamine hypothesis at all cost in the same way that the neo-Darwinian would be trying to save the continuous variation hypothesis at all cost. If I am right so far, I may have an argument for the impossibility of a truly falsifiable scientific biomedical theory of clinical depression and, moreover, no scientific biomedical theory relative to which the notion of placebo can be defined according to Grünbaum's analysis.
Focusing on a typical experiment dealing with clinical clinical experiments in psychology, medicine and psychiatry. These experiments rest upon the monoamine hypothesis of clinical depression. My analysis has centered on:
I have not here attempted much by way of solution, although this is my long-term goal. Too often these experiments are so pragmatically oriented (and this orientation is, of course, perfectly understandable and acceptable) that the theoretical questions concerning causal explanation are overlooked. Such is the case with the causal efficacy of placebos, not to mention the question of what a placebo is, although, as noted above, this latter question has received some solid philosophical analysis by Grünbaum. But what is also very important is whether biomedical theory can ever develop scientifically at all, given that so many of the relevant theoretical hypotheses (such as the monoamine hypothesis) systematically overlook a careful, rigorous examination of the concepts of testability of hypotheses as well as problems of circularity and unfalsifiable versus ceteris paribus explanations. I claim that the questions raised here are just the first steps toward a great deal of much needed analysis from philosophers of science.
(1) I do not mean to imply that the term 'hypothesis' is interchangeable with 'theory'. However, one way to conceive of the relationship of hypothesis to theory is that hypotheses are tested and, when highly confirmed, become laws that are joined together with other laws to form a theory. Under this view, hypothesis is the first stage of theory development.
(2) The DSM-III is a diagnostic manual of mental disorders listing various criteria and diagnostic steps for numerous psychopathologies, depression included. Published by the American Psychiatric Association, the DSM series continues to be the standard diagnostic handbook for psychiatrists and psychologists alike.
(3) The HRSD continues to be the pre-eminent rating scale for depression, being used in the majority of research studies since it was introduced in 1960. Questions take the form: "Are you often tired upon awakening from a night's sleep?" or "Do you have enough energy to eat regularly?"
(4) An ad hoc hypothesis is a unique hypothesis constructed solely to save an explanation and is not testable. Such an hypothesis has no place in science. On the other hand, a ceteris paribus hypothesis claims that an explanation or prediction does not hold because all things are not equal; if all things were equal, the explanation or prediction would hold. Unlike the ad hoc hypothesis, the ceteris paribus statement must be empirically testable. The hypothesis that the orbit of Uranus deviated from the orbit predicted by Newtonian mechanics because of the gravitational force of a planet further out was a ceteris paribus hypothesis; the statement was testable (by simple, direct observation) and, indeed, another planet-Neptune-was discovered. I am claiming that, in light of my discussion of the results of the experiment, biomedical scientists should be treating the monoamine hypothesis as a ceteris paribus hypothesis. If they do not do so, they are de facto treating the hypothesis as something bordering on being very close to being an ad hoc hypothesis, one that is not subject to further test and perhaps falsification. However, I am not saying that the monoamine hypothesis is unequivocally ad hoc, for there does appear to be some empirical grounding for the use of the monoamine hypothesis. What I am claiming is that, at least, further analysis of this whole cluster of problems that I discuss in this paper is needed.
Bakan, David. "The Apprehension of the Placebo Phenomenon." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press., 1985)
Bootzin, Richard R. "The Role of Expectancy in Behavior Change." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Brody, Howard. Placebos and the Philosophy of Medicine: Clinical, Conceptual, and Ethical Issues. (Chicago: The University of Chicago Press, 1980)
_______. "Placebo Effect: An Examination of Grunbaum's Definition." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Carlson, Neil R. Foundations of Physiological Psychology. (Boston: Allyn and Bacon, Inc., 1988)
Evans, Fredrick J. "Expectancy, Therapeutic Instructions, an the Placebo Response." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Grünbaum, Adolf. "Explication and Implication of the Placebo Concept." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Pepper, O. H. P. "A Note on the Placebo."American Journal of Pharmacy Vol. 117 (1945): 409-412.
Popper, Karl R. Conjectures and Refutations. (New York: Harper & Row, 1965)
Roberts, Alan H., Kewman, Donald G., et. als. "The Power of Nonspecific Effects in Healing: Implications for Psychosocial and Biological Treatments." Clinical Psychology Review Vol. 15, No. 5 (1993): 375-391.
Shapiro, Arthur K. "A Contribution to A History of the Placebo Effect." Behaviourial Science Vol. 5 (1960): 109-135
_______. "The Placebo Response." in Modern Perspectives in World Psychiatry, ed. J. G. Howells. (Edinburgh: Oliver and Boyd, 1968)
Versiani, M., et. als. "A Double-blind Comparative Trial of Moclobemide v. Imipramine and Placebo in Major Depressive
Episodes." British Journal of Psychiatry Vol. 155, suppl. 6 (1989): 72-77.
White, L., Tursky, B., and Schwartz, G. E. "Placebo in Perspective." In Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Wilkins, Wallace. "Placebo Controls and Concepts in Chemotherapy and Psychotherapy Research." Placebo: Theory, Research, and Mechanisms, ed. Leonard White, Bernard Tursky, and Gary E. Schwartz. (New York: The Guilford Press, 1985)
Zavodnick, Steven. "Somatic Theories of Depression." Depressive Disorders: Facts, Theories, and Treatment Methods, ed. Benjamin B. Wolman, and George Stricker. (New York: Wiley-Interscience Publication, 1990).