Putting Freud to the Test

Freud thought that he was the founder of a science. In one of his later papers, he wrote that psychoanalysis is “a part of the mental science of psychology.”¹ But his detractors, a number of whom have lately mounted several sensational efforts to discredit his character, have contended that he did nothing of the kind. For some of them, he is at best a gifted writer whose eloquence concealed the defective reasoning behind his theories and who properly received the Goethe Prize for literature instead of the Nobel Prize for medicine.

But the question of whether psychoanalysis is a science is hardly clear. What is “science”? There seem to be so many kinds of science that it is difficult to extract a common characteristic that would qualify all of them as science. And what is “psychoanalysis”? A theory of mind? A therapy? A method of investigating mental disorder? In his new book, Adolf Grünbaum, a philosopher whose most important previous work concerned philosophical questions raised by the physics of space and time, tries to clarify this question and also to articulate what he calls the “logical foundations” of psychoanalysis. By this he means the logical relations between the hypotheses of the theory and the kinds of evidence that Freud and his followers thought could support the theory. Those familiar with Grünbaum’s work will expect a far more fastidious and exhaustive treatment of this subject than is to be found in previous philosophical studies.

Few philosophers have concentrated systematically on psychoanalysis, although there have been some distinguished exceptions, such as Sartre and, lately, Richard Wollheim. Karl Popper sought, some thirty years ago, to illustrate his model of science by showing psychoanalysis to be what he called a “nonscientific” theory. According to Popper, a theory is scientific only if it is “falsifiable.” By this he meant that the theory yields some predictions that are observable and could refute the theory if they failed to occur. Popper complained that Freudian theory is not falsifiable in this sense because anything a human being says or does is compatible with it; it can explain everything, and thus explains nothing. He also claimed that psychoanalysts were not especially concerned to test their theory critically, by which he meant that they did not try to falsify it. In 1962, in a forceful paper read at a symposium organized by Sidney Hook, the distinguished philosopher Ernest Nagel put forth a more complex view that also emphasized the unfalsifiability of many psychoanalytic hypotheses.

Grünbaum stands apart from this critical tradition. He wishes both to acknowledge Freud’s accomplishments as a scientific methodologist and to rebut Popper’s charges. But he also claims that Freud never succeeded in firmly grounding his theory in empirical findings. His book is a strangely organized, difficult work, unmistakably a string of scholarly articles to which vast accretions of evidence and afterthoughts have been added. It is written so much in reaction to the views of other philosophers and interpreters of Freud that Freud seems at times incidental to Grünbaum’s purpose. The book has, for example, a ninetyfour-page “introduction” (in fact, one third of the book) in which the “hermeneutic” or nonscientific approach to psychoanalysis is criticized. This is followed by a chapter on the testability of Freudian theory, which concentrates largely on Popper. Only in the second chapter do we arrive at Grünbaum’s positive contribution to the debate about the scientific status of Freud’s arguments for his theory.

Grünbaum’s views are not excessively complicated, but he presents them in an obscure and frustrating manner. The book is a stupendous philosophical Nymphenburg divided in parts, chapters, and subchapters. He uses words like “scientophobic” and “psychonoxious”; for him a woman with bad skin has a “dermatological deficit” and a footprint in the sand is a “pedal invasion.” The book contains some of the most dreadful and inadvertently funny formulations I have ever seen, as when Grünbaum chastises a philosopher who has “treated us to a procession of logical enormities interlaced with homiletic ipse dixits.”

“Psychoanalysis” has generally stood for a psychology of mental functioning, including disorders of functioning, and for a method of treating such disorders. But it has had other meanings. Freud’s theory is, in fact, a vast collection of propositions of different kinds. Some of them are abstract and speculative, such as his sketch, completed in 1895, of a theory of mind which sees the mind as a set of elements, each charged with “psychic energy,” the system of elements as a whole conforming to various constraints, such as a law of conservation of energy. Freud thought at one time that if these elements could be identified with physical entities in the brain (such as neurons), his theory could provide a physical basis for psychoanalysis. But he let the project languish.

Other components of Freud’s thought are remote from observable evidence because they are general propositions about human nature—such as his claim that “the force behind all human activities is a striving toward the two convergent aims of profit and pleasure.” Such propositions are on a par with Rousseau’s claim that all human beings are basically good, or with the postulation of original sin: instead of being amenable to empirical confirmation or refutation, such statements are vague principles governing inquiry into human beings.

Still other Freudian propositions (such as his claim that repressed homosexual love is causally necessary for the condition of paranoia) are not in principle untestable, although they may be difficult to test, as we shall see. But given the heterogeneous nature of Freud’s thought it is difficult to view it in its entirety as a scientific theory like the standard formulations of statistical mechanics. Accordingly, when the “scientific status” of Freudian theory is discussed by philosophers and others they usually mean a reconstruction of his work that excises some of Freud’s ideas, plays down others, and improves still others. For example, few contemporary writers include in Freud’s “theory” his comments in favor of telepathy or the inheritance of memory, and they often exclude aspects of his thought that he considered of the greatest importance, such as his “death instinct.”

Grünbaum, for his part, generally confines his discussion to Freud’s theory of unconscious motivation and personality. He concentrates on Freud’s view that ideas can be “repressed” or made unconscious by agencies at work within the mind, and on his belief that forbidden wishes and impulses of a sexual character that are unsuccessfully repressed in childhood can return in later life in the disguised form of psychopathological symptoms. He also discusses Freud’s therapeutic method, which tries to undo past repressions and restore conflicts to consciousness by artificially creating a situation in which the patient’s unconscious is permitted to express itself through free association, and in which the patient is supposed to work out a mature solution to his previously repressed infantile conflicts. Drawing on Freud’s hypotheses about motivation and personality, the psychoanalyst interprets the remarks produced by the patient in free association in order to help him arrive at a solution of his conflict.

Popper wrote that Freudian theory is without empirical content because it does not exclude “any particular person’s acting in any particular way, whatever the outward circumstances. Whether a man sacrificed his life to rescue a drowning child (a case of sublimation) or whether he murdered the child by drowning him (a case of repression) could not possibly be predicted or excluded by Freud’s theory.”

Grünbaum contests Popper’s claim by arguing that a number of the hypotheses of Freud’s theory are falsifiable (and thus exclude a variety of events); and he adds that Freud was willing to accept the falsification of some of his ideas. He cites the example of Freud’s hypothesis that repressed homosexual love is causally necessary for the occurrence of paranoia.² In 1915 Freud was consulted by a lawyer whose client, a young woman, had sought his protection from a man with whom she had been having an affair. He suspected she was paranoiac and asked Freud to examine her. Freud found that she showed signs of paranoid delusion, but he also found no evidence of a repressed homosexual attachment.

Grübaum claims that Freud was prepared to surrender his theory of paranoia, and he quotes Freud’s remark that in view of his findings, “either the theory must be given up or else, in view of this departure from our expectations, we must side with the lawyer and assume that this was no paranoic combination but an actual experience which had been correctly interpreted.” Grünbaum alsoclaims that Freud’s theory of paranoia is not empty of “empirical content,” as Popper had implied, for it yields the empirical prediction that with the decline of the taboo on male homosexuality in our society, there should be a decreased incidence of male paranoia. Moreover, Grünbaum says, Freud showed his willingness to give up a refuted idea when he abandoned his early hypothesis that neurosis originated in the children’s actual seductions by adults.

Grünbaum pursues at considerable length the question of whether psychoanalytic theory is “scientific” insofar as it can meet a formal “criterion of demarcation” between “scientific” and nonscientific statements. Formulated in this way, the question is not of evident importance. What is important for the scientific status of a theory is the promise it holds to contribute to systematic and controlled knowledge of the world. Even a vague, untestable, or false theory can have such promise if it suggests an interesting path of research to scientists. Moreover, whether or not a theory is itself “scientific” it can be investigated in a scientific spirit. Even if it had not testable claims of its own, psychoanalysis could still be scientific insofar as psychoanalysts made efforts to clarify their hypotheses and submit them to rational criticism.

Of course if a theory is testable, or has some evidence in its favor, its claim to scientific status may be stronger. But this status can depend on numerous other considerations, such as the theory’s power to organize hitherto unconnected facts, its scope or range, and the strength of competing hypotheses. Philosophers and sociologists do not know enough about science to compile a full list of whatever considerations would contribute to the scientific status of a theory. The effort to construct a general rule or “criterion” of scientific status remains a dubious enterprise. No firm understanding of science has resulted from the construction of such criteria.

Far more important than his controversy with Popper are the chapters in which Grünbaum tries to identify some of the main obstacles that stand in the way of finding empirical support for psychoanalysis. He largely confines his discussion of this question to asking whether the most important part of Freud’s theory—the theory of motivation and personality—can receive support from the kinds of empirical evidence Freud thought could support it. He pays particular attention to Freud’s hypothesis that the origin of neurosis lies in repressed fears, wishes, and impulses, noting that Freud called the theory of repression “the cornerstone on which the whole structure of psycho-analysis rests.” According to Grünbaum, not only does Freud’s theory of dreams (which are held to represent repressed wishes) and slips of the tongue rest on this basic theory, but it is presupposed by the major post-Freudian schools of psychoanalysis, including the “object relations” school derived from the work of Melanie Klein, among others, and the “self psychology” of Heinz Kohut as well. If the repression theory is discredited, then so are other theories.

The repression theory was formulated by Freud in the last decade of the nineteenth century. He administered to patients suffering from symptoms of hysteria a “cathartic” treatment using hypnosis. Each hysterical symptom, he argued, originated in the repression of the memory of a trauma which preceded the appearance of the symptom. By his treatment he sought to lift this repression by bringing to light the traumatic experience. When he did so, he said, the symptom disappeared, so he felt justified in explaining his success by postulating that the uncovered trauma was the cause of the symptom.

Freud later replaced hypnosis by the method of free association, but he retained the view that any therapeutic gain depended on the patient’s insight into the causes of his affliction. And he assumed that any lasting success with his method could be held as good evidence that the interpretation he had put to the patient was true. Freud therefore rested his repression theory on therapeutic success and thought that this success could be ascertained through “clinical observations”—data drawn from within the psychoanalytic treatment session. Clinical observations, he thought, could both support his theory and give the analyst access to deep aspects of the patient’s mind that could not be examined in other ways.

Using clinical observations to test the repression theory and related causal psychoanalytic hypotheses, Grünbaum writes, presented Freud with several difficulties. First, such observations alone could not, logically, support psychoanalytic hypotheses, at least a great many of them. Consider, for example, Freud’s celebrated case of the “Rat Man,” Paul Lorenz, who suffered from compulsive fears that a horrible punishment involving rats would befall both his father (who was dead) and a woman friend. After examining Lorenz, Freud concluded that he was an obsessive neurotic. According to his theory, the specific cause of obsessional neurosis was repressed sexual experiences in early childhood; he therefore assumed that Lorenz must have had such experiences as infantile masturbation. It so happened that Lorenz did not have any direct memory of them. But Grünbaum argues that even if Lorenz had truthfully recalled that he had had such experiences and had repressed them, such memories would not have supported Freud’s hypothesis about the origins of obsessional neurosis. Even if many such neurotics existed, it might also be true that many persons who did not have such experiences also turned out to be obsessional neurotics. As Grünbaum writes:

Suppose it were hypothesized that drinking coffee is causally relevant to overcoming the common cold. Consider, too, the case of a recovered cold sufferer who turns out to have been drinking coffee while still afflicted by the cold. Then such an instance, taken by itself, would hardly qualify as supportive of the hypothesized causal relevance.

To gain satisfactory support for Freud’s hypothesis, we also need to have evidence of persons whose unconscious was not afflicted with repressed infantile sexual activity and who did not subsequently go on to develop obsessional neuroses. Such evidence must inevitably be found outside the psychoanalytic treatment session.

But Freud faced a second, equally serious, difficulty in using clinical observations to support psychoanalytic hypotheses like the repression theory. The treatment of a patient could be called a “success” only if the analyst had correctly identified the actual cause of the patient’s affliction and only if the symptoms of the patient had disappeared because he gained insight into the cause of his condition. But if the patient accepted the analyst’s interpretations, and if his symptoms subsequently went away, is it not possible that they did so because the analyst had unwittingly “talked” the patient into improvement without either identifying the true cause of the patient’s condition or guiding him to genuine self-knowledge?

If this were so, of course, then the apparent “success” of the analysis would not be good evidence for the truth of the analyst’s various interpretations, or for the theory that lay behind these interpretations. It would only confirm the analyst’s skill in inducing the patient to manufacture responses to his questions that validated his interpretations, a skill shared by African witch doctors and Christian Scientists, among others. As Grünbaum expresses the difficulty, such evidence of “success” would be “contaminated” by suggestion. The analysis would be a placebo (though an inadvertent one), like a sugar pill that “cures” a headache because the patient believes that it does. The therapy might “work,” but not because of factors that are deemed to be remedial by psychoanalytic theory.

Freud, who was intensely concerned about the problem of the contamination of clinical evidence by suggestion, called it “the objection that is most often raised against psycho-analysis.” It was raised, indeed, even by his colleague Wilhelm Fliess. Freud sought to minimize the effects of suggestion in psychoanalysis, but at the same time he was forced to admit that a form of suggestion is used in every psychoanalysis. For as an analysis proceeds, and the patient uncovers more and more repressed material, he comes, in Freud’s words, to reexperience “emotional relations which had their origin in his earliest object-attachments during the repressed period of his childhood.” He relives these childhood situations and uses the analyst as the “object,” thus transferring to the analyst the unconscious infantile attitudes he held toward his parents and other “objects.”

Freud wanted the analyst to encourage this “transference” because the patient’s feelings for the analyst, once marshaled, help the analyst to draw out memories from the patient and other material useful to the analysis; in doing so, Freud’s wrote, “the transference is changed from the strongest weapon of the resistance into the best instrument of the analytic treatment.” But Freud insisted that this use of suggestion differed entirely from that used in hypnosis, in which symptoms were “forbidden” to exist. In analysis, he said, suggestion functions to educate the patient in uncovering the underlying cause of his symptoms and in achieving insight.

In 1917 Freud gave what Grünbaum calls “a brilliant effort to come to grips with the full dimensions of the challenge of epistemic contamination by adulterated clinical responses.” He used evidence of therapeutic success not only to confirm his theory (as he had done before) but to counter charges that this confirmation was spurious because of suggestion.

The analyst, Freud wrote, can make to the patient all sorts of suggestions that influence the analysis. Indeed, he has no difficulty in making the patient

a supporter of some particular theory in thus making him share some possible error of his own. In this respect the patient is behaving like anyone else—like a pupil—but this only affects his intelligence, not his illness. After all, his conflicts will only be successfully solved and his resistances overcome if the anticipatory ideas he is given tally with what is real in him. Whatever in the doctor’s conjectures is inaccurate drops out in the course of the analysis.³

According to Grünbaum, Freud assumed in this argument that the patient cannot conquer his illness without insight into its causes, and that he cannot get insight into its causes by hypnosis or any other therapy than psychoanalysis, with its method of unlocking repressed ideas by “free association.” So if an analysis is successful, and if the patient’s symptoms go away and new ones do not appear, then it is good evidence that the doctor’s conjectures did indeed “tally” with what was “real in him” and are therefore true. To be sure, suggestion was used in the analysis, but the analysis was not in its entirety a matter of suggestion, for if the symptoms disappeared—were in “remission,” to use the jargon—then the patient must have had insight into his affliction, an insight arrived at with the help of the analyst. Therefore every successful analysis is evidence of the truth of the analyst’s interpretation—and indirect evidence both of the truth of the theories that have guided the analyst in arriving at these in terpretations and of the validity of the method of free association.

Grünbaum carefully shows why this argument is flawed. He casts doubt on Freud’s premise that a variety of mental disorders and the symptoms accompanying them cannot be conquered unless the patient gains psychoanalytic insight into the causes of his condition. This premise is vulnerable, Grünbaum argues, first, because durable remissions of symptoms may be “spontaneous”—by which he means not the disappearance of symptoms for no apparent reason but remissions caused by events outside the psychoanalytic treatment.

Secondly, the sort of insight given by psychoanalysis is evidently not necessary for curing patients, Grünbaum says, since rival treatments (like behavior therapy) can yield results as impressive as those of psychoanalysis, if not more so. Grünbaum also writes that Freud recognized some of the flaws in his argument and by 1937 had surrendered his claim that psychoanalytic insight is necessary for the cure of psychoneurotic illness. But he complains that although Freud gave up the “tally argument,” he never replaced it with another way of supporting his theory or of purifying clinical evidence for it. So the charge of what Grünbaum calls “epistemological contamination” of clinical observations through the analyst’s suggestions remains. As a result, such observations cannot be legitimately used to confirm psychoanalytic hypotheses. Nor can the method of free association, whose credibility as a method lay in its alleged ability to uncover repressions, be supported unless some substitute for the “tally argument” is found.

Grünbaum claims, therefore, that the central hypotheses of psychoanalytic theory are devoid of evidence of the kind Freud thought could support them. Indeed, he goes further: “The seeming ineradicability of epistemic contamination in the clinical data, adduced as support for the cornerstones of the psychoanalytic edifice, may reasonably be presumed to doom any prospects” for the cogent testing by clinical observations of Freud’s major tenets.

Important as this conclusion may be, it is a restricted one. Grünbaum does not claim that there has been, or can be, no evidence, clinical or otherwise, for Freudian theory. According to Grünbaum’s argument, it remains possible that what Freud described as a “tally” between the analyst’s interpretations and the patient’s psychic “reality” does occur. The problem is whether that “reality” has not itself been determined by the analyst’s suggestions. Grünbaum’s book is for the most part confined to creating serious doubt that one kind of evidence—“clinical observations” from the analytic session itself—which Freud thought could be used to support his theory, is capable of doing so. This conclusion does not rule out the possibility that methods of verification other than Freud’s might confirm psychoanalytic theory.

For example, psychoanalytic hypotheses about the causes of illnesses like paranoia or obsessional neurosis might be partly tested, at least in principle, in the way epidemiologists try to test their hypotheses about the causes of diseases like lung cancer or malaria by studying the incidence of the illness in a population and inquiring whether the conditions under which it occurs match those postulated by psychoanalytic theory. Freud’s hypothesis that the specific cause of paranoia is repressed homosexual love, which is cited by Grünbaum, furnishes an illustration of a psychoanalytic hypothesis that might be studied in this way. It yields, as we saw, what Grünbaum calls a “statistical prediction” that “the decline of the taboo on homosexuality in our society should be accompanied by a decreased incidence of male paranoia.” Such a prediction can be tested in principle, and in doing so one need not appeal to any data drawn from the psychoanalytic interview. Indeed, Grünbaum says that researchers might even now “begin garnering appropriate statistics on the incidence of paranoia with a view to ascertaining in due course whether these epidemiologic data bear out the psychoanalytically expected decline.”

In some cases, researchers might even be able to test the causal claims of psychoanalysis experimentally, by deriving from them hypotheses that can be tested in a laboratory. Such experiments have been attempted, and I shall discuss one of them below. But Grünbaum neither systematically assesses the body of evidence that has been thought by some experimental investigators and others to bear on Freudian theory nor explores ways of testing the theory that might be developed in the future. Of course standard experimental procedures comparing people who are and are not undergoing analysis generally are applicable to the psychoanalytic interview between analyst and patient, but even “clinical observations” might confirm Freudian theory if some way of “purifying” them of the possible effects of suggestion could be found.

Nor is Grünbaum’s conclusion as controversial as he sometimes suggests, for he tells us that an orthodox analyst, Kurt Eissler, has put forth the view that “the future validation and/or disconfirmation of Freudian theory will come very largely from extraclinical findings.” And if findings drawn from outside the psychoanalytic interview eventually prove capable of testing Freudian theory, that should not be so surprising, as Philip Holzman, an analyst who teaches at Harvard, has pointed out in a recent paper.⁴ Psychoanalysis is a speculative theory that does not make many specific predictions and, as Holzman says, many such theories in the history of science have not been testable by the methods from which they arose. The theory of natural selection for example,

is probably not capable of being tested by the same method from which it arose—that of observation of continuities, discontinuities, and diversity among species. To test ideas about how species evolve requires experimentation within a sphere different from the one that led to the hypothesis included in the theory of natural selection. It requires controlled experiments in genetics and in biochemistry, fields that were not even known to Darwin.

At best, Holzman writes, “the psychoanalytic situation is an admirable one for generating ideas about human conduct that would not occur in a psychology laboratory,” but it is not the place to test whether these ideas are true.

Grünbaum writes that the validation of Freudian theory outside the psychoanalytic session is “largely a task for the future.” In fact there have been a great many efforts made during the past half-century to apply scientific methods to test the theoretical and therapeutic claims of psychoanalysis.⁵ But these studies and experiments—which have sought to find empirical evidence for the psychoanalytic theory of dreams, repression, paranoia, personality types, and other phenomena—have been controversial. Some students of such experimental studies of Freudian theory, like the English psychologist Hans Eysenck, have argued that they provide no strong evidence at all for any Freudian hypothesis. Why is this so? For one thing, investigators have not agreed on what standards of experimental design and justification of results should be employed in such tests. Secondly, even if uncontroversial standards of testing are used, they are extraordinarily difficult to apply in testing the psychoanalytic theory or therapy. To construct a controlled test of the therapeutic efficacy of psychoanalysis by classical scientific standards, for example, two groups of persons would have to be matched for both the nature and the intensity of their condition and then randomly distributed into a “treatment” and a “control” group. An analyst must be “randomly” selected. The life of the “control” group must be regulated so that no “psychotherapy” is inadvertently given it. Criteria of “cure” or outcome must be specified in advance.

All of these conditions have been exceptionally hard to obtain. How can it be shown that two persons have the same psychological problem and in the same degree of intensity? How would the problem be defined, and what constitutes “psychotherapy”? Is “cure” to be construed as “insight” or as symptom remission? There are numerous other conditions that must be met if such tests are to have validity, such as arranging for follow-up studies to ascertain the durability of successes or failures of treatment.

But even if the conditions of controlled experimentation could be realized, tests of Freudian claims are often impossible to design because of the loose texture of the claims themselves. S. Fisher and R.P. Greenberg, in their book on the scientific status of psychoanalysis, speak of the complexity of the Freudian theory of how “personality defenses” are organized: “Freud, in describing such defenses, suggests that a wish may variously express itself in direct pursuit of the goal of the wish or in outright denial of the wish or even in the defensive pursuit of goals that are the direct opposite of the wish. How, ask many analysts, can you put this brand of complexity into an experimental design?” Because of this inaccessibility to experiment, many alleged “tests” of Freudian theory are not really tests of the theory at all.

In his careful book, which covers much the same ground as Grünbaum’s and more, the Oxford philosopher B.A. Farrell cites one experiment he found “impressive” because he thinks it went some way toward testing Freud’s theory of repression by finding evidence of it in perception and recognition:

Dixon used an apparatus which allowed him to present stereoscopically two spots of light—one brighter than the other—to the left eye of the subject and stimulus words, subliminally presented, to the right eye. The apparatus enabled the subject to control the brightness of the two spots, and he was instructed to work it continuously so that he could “just see the brighter of the two spots but never the dimmer one.” It was found that when emotionally disturbing words (whore and penis) were presented subliminally to the right eye, the visual threshold went up and the subject had to increase the brightness of the spot shown to the left eye.

If the experiment is “firmly replicated, it suggests very strongly that some internal control machinery is at work of the sort described by the theory of repression.”⁶

But another student of experimental work on psychoanalysis, the philosopher Edward Erwin, claims that Dixon’s experiment fails to support Freud’s theory of repression, not because it is poorly designed, but because it does not really test that theory; it is thus an example of what, to use Farrell’s own distinction, is a test of a claim “suggested” by Freudian theory but not of the theory itself. Thus is because Dixon shows only that there can be “discrimination without awareness which affects recognition thresholds.” But Erwin claims that this could not test the Freudian theory of repression, which postulates an unconscious mind that is performing the repression:

the mere fact that there is discrimination without awareness and that this discrimination affects recognition thresholds does not establish that any subject has defended against threatening material by incorporating it into his unconscious. Repression could be the cause of perceptual defense effects, but there is no firm evidence so far that it is.⁷

Criticisms such as Erwin’s support Grünbaum’s view that the testing of Freudian theory outside the clinical setting is largely a task for the future. But this is because tests of psychoanalytic hypotheses have proved difficult to design and to conduct. The claim that there is “no evidence” for psychoanalytic theory usually means not that positive evidence has been found that decisively runs against the theory, but rather that investigators have been unable to invent good tests of it and therefore no strong evidence exists for or against it.

There is no reason to suppose that better tests might not be invented in the future, although such tests might confirm or disconfirm a number of competing theories, not just Freud’s. Will psychoanalysts be indifferent to the outcomes of such tests, as Popper and others have implied? Again, although some psychoanalysts have ignored or treated capriciously considerations of evidence, we need not suppose that all have behaved in this way, or that they must do so, now or in the future. One of Grünbaum’s contentions, after all, is that Freud did not behave in this way. He was, indeed, scornful of religion and of obscure occult theories—which he spoke of as a “black tide of mud”⁸—for their insensitivity to questions of evidence. Grünbaum says that he has found “ample evidence that Freud’s successive modifications of many of his hypotheses throughout most of his life were hardly empirically unmotivated, capricious, or idiosyncratic.”

If psychoanalysts wish to respond to criticisms like Grünbaum’s, it would seem appropriate that they systematically attempt to clarify and test their hypotheses. Have they failed to do so because of a blind, dogmatic adherence to their views? Some may have done so, but this path of research has been blocked for other reasons as well. For one thing, psychoanalytic theory is still in a process of development, so that there is no one clear research tradition to follow or to test. For example, the supporters of the object relations school have contested Freud’s claim that all behavior and all aspects of personality, including psychopathology, express efforts to satisfy or discharge innate drives like those of sex and aggression. They deny that social relations with others are secondary, a derivative of the expression of these drives.

The analysts of the object relations school have therefore either abandoned or supplemented Freud’s view. The English psychoanalyst John Bowlby, for example, wrote that the child’s tie to his mother is not secondary to basic drives; it does not result from the fact that the mother provides the child a way of gratifying his need for nourishment or contact. Rather, he argued, the need for attachment in human beings is primary; it is an instinctual need independent of such drives as sex and aggression, and all children have an innate propensity to establish such ties of attachment.⁹ The difference between the views of orthodox Freudians and object relation theorists might lead them to hold entirely different hypotheses worthy of being tested.

Even if there were a clear research tradition to test, there is, as Holzman points out, an insufficient number of scientists who wish to devote their time to testing psychoanalytic hypotheses. This is so not only because few medical and other scientists have shown strong interest in Freud’s ideas, but also because, as Holzman expresses it, “in some quarters, the appropriate career goal of a psychoanalyst is one that emphasizes clinical psychoanalysis only, undistracted by the siren call of laboratory or other empirical research, university teaching, or other nontherapeutic but scholarly pursuits.” Holzman suggests that even if psychoanalysis is scientific in its approach to its problems, it will not advance to the stage of testing its ideas and becoming a mature science if its institutions remain exclusively concentrated on psychoanalytic therapy. The question of whether psychoanalysis is likely to become a science is therefore an “institutional” one. The “emphasis on maintaining a cadre of exclusive practitioners or practitioners of an exclusive therapy, produces advocates with vested interests in maintaining the therapy against all change, influence and criticism.” And he places the responsibility for the shortage of scientists working on the testing of psychoanalytic theory “at the doorstep both of the psychoanalytic institutes who focus only on teaching the therapy, and on the universities who shun the psychoanalytic data.”

These institutional questions Grünbaum mostly ignores, but he does discuss another force within psychoanalysis that arrests the clarification and testing of Freudian and other psychoanalytic theory: the recent vogue among some psychoanalysts and philosophers to complain that Freud misunderstood his own enterprise when he claimed scientific status for it. In their view, psychoanalysis should not be regarded as a science, or as a “proto-science,” or even as scientific. It is a “hermeneutic” enterprise which does not discover the causes of mental afflictions but analyzes their “meaning” through a collaboration between patient and analyst whose goal is to create a more satisfactory “story” of the patient’s life.

In his “introduction” Grünbaum attacks the views of such “hermeneutic” interpreters of psychoanalysis as Paul Ricoeur and the late George Klein; especially effective is his discussion of Jürgen Habermas, who holds that psychoanalysis is an instance of what he calls “critical” science, which aims at the emancipation of human beings. Grünbaum shows how Habermas and others trivialize psychoanalysis by removing its claims to scientific status; and he shows as well how most of their objections to seeing it as science—such as their claim that the concern of psychoanalysis with the “meaning” of symptoms or illnesses excludes it from being a causal science—rest not only on logical mistakes and ignorance about psychoanalysis but on misconceptions about the scope of natural science. One could add that the hermeneutic position is a retreat to the line of greatest safety by those psychoanalysts and philosophers who have been unduly impressed by the arguments that psychoanalysis cannot attain scientific status.

For those who believe that the “tally argument” is sound, Grünbaum’s account of its defects should be clarifying and challenging. One frequently feels, however, that his book pursues the wrong questions to ask about psychoanalysis today. Since psychoanalytic theory is not a discrete story like a myth or a novel, but might, like other theories, be further developed, and since new experimental designs might be able to test some of its claims in the future, one wants to ask how much is gained by examining its “logical foundations” in the way Grünbaum does, however useful it may be to know what Freud conceived them to be. Instead of repeatedly ferreting out the methodological flaws in Freud’s way of testing his theory within the psychoanalytic treatment session, Grünbaum might profitably have examined in greater detail, as Farrell tries to do, the full range of efforts that have been made to test the main hypotheses of psychoanalysis outside the clinic, and to specify what obstacles must be surmounted by similar tests in the future if they are to yield strong confirmations or disconfirmations of that theory.

It is true that psychoanalysis was created at the turn of the century and has not yet fulfilled its scientific promise. But many theories in the history of science took a long time to acquire a reputation for being “scientific.” As Farrell remarks, “Harvey’s theory of the circulation of the blood remained very unsatisfactory until the capillaries were discovered. Prout’s suggestion—that the atomic weights of elements could be exhibited as multiples of hydrogen—was only taken up and appreciated later on. Newton’s theory of matter and his corpuscular view of light had to wait for a couple of centuries before they could be handled by science with some confidence.”

It is also true, however, that many theories, after much development and patient attention from scientists, have failed to yield confirmed hypotheses and have either collapsed under the weight of repeated disconfirmations or have been rendered obsolete by new theories. It is possible that psychoanalysis will have this fate. The profession has become increasingly isolated from organic medicine; it has found no Freud; its theoretical development has been stagnant. It is arguable that what seems evidently true in Freud, such as his notion of repression or his emphasis on the unconscious and on the irrational springs of much of human behavior, has long been known and that Freud introduced an unnecessary technical language and a dubious metaphysical backdrop to describe these phenomena. And it is possible that psychoanalytic therapy will be replaced in time by shorter therapies of various kinds and by psychopharmacology derived from new developments in the neurosciences.

Nevertheless, there is something to psychoanalysis, and to the insights of what Grünbaum calls “Freud’s brilliant intellectual imagination,” and our attitude toward psychoanalysis might prudently be one of benevolent skepticism. It would be unwise to judge it finally as a theory at the present time, or to foreclose the future of scientific invention. However imperfectly formed and unsupported some of its hypotheses may be now, they may be improved; and even if they are eventually falsified or otherwise discredited, they may stimulate the imagination of scientists working in other fields to incorporate what seems true in them into more satisfactory and confirmed theories. It is somewhat strained to say, with Farrell, that “the sceptic’s argument against Freud today may turn out to be just as ridiculous as his argument against Newton in the eighteenth century.” But he is not wrong in claiming that many of Freud’s ideas may be “pointers to the truth—signposting the avenues to pursue if we wish to get at a reasonably definitive account of human nature.”