Science in Christian Perspective



A Commentary On Being Sane in Insane Places
Social Criticism and Scientific Responsibility
Department of Psychiatry and Human Behavior
University of California, Irvine and Deputy Director, Training Orange County Department of Mental Health California

From: JASA 26 (September 1974): 110-114.

A recent highly publicized article in Science magazine takes a broad swiping attack on the mental health system. This article presents a technical analysis of the author's research, The analysis suggests that the author's conclusions are unrelated to his research as reported. However, evidence is presented to suggest that the author used his research material as an occasion to present a trenchant social criticism. The covert and unlabelled combination of social criticism in conjunction with and under the label of empirical scientific study is brought under question. This method of publication is potentially dangerous to the intellectual autonomy of science, and undercuts the recognition of moral responsibility for social criticism as a human venture.


An interesting, perplexing, and disturbing event occurred in the spring of 1973 that raises some basic issues in terms of scientific responsibility. A long major article was published in the January 19th issue of Science, on the topic of psychiatric diagnosis in mental hospitals, with the intriguing title, "On being sane in insane places".11 Simply put, the author described a series of experiments in which eight subjects gained admission to twelve psychiatric facilities by requesting admission with a presenting complaint of bizzare hallucinations. In each instance the subject presented himself in the role of a psychotically disturbed patient, was admitted, received a psychotic diagnosis, was observed for several days on a ward, had his ease reviewed by staff, and subsequently was discharged as a case of psychosis in remission.

The author admits that this sequence of events is not particularly surprising, although this admission comes in his subsequent rebuttal to letters to the editor some months later. The author says: "The issue is not that the pseudopatient lied. Of course he did. Nor is it that the psychiatrist believed him. Of course he must believe him. Neither . . . whether admitted (which) was the only humane thing to do."12 So what is at issue, that should stir an article written and presented in stirring polemic fashion? According to the author it is simply: "the diagnostic leap between the single presenting symptom . . . and the diagnosis."

To those outside the field of psychiatry this might seem like a topic hardly scintillating enough to stir controversy. And indeed within the field of psychiatry this issue has been the topic of many sober studies. So we are left wondering why the sensationalistic publication in a general journal such as Science?

Before attempting some speculations and reflections, I should like to present a brief critique of the research methodology and conclusions of this study from the standpoint of empirical research. Although this may be a bit technical for the reader, I wish to illustrate the serious methodological, conceptual, and logical flaws of the author's report.

A Critique of Method

The recent article by D. L. Rosenhan, "On Being Sane in Insane Places", has received widespread notice in the public press. These reports proclaim that psychiatrists are unable to differentiate normal persons from those suffering from severe emotional disorders, thus implying serious question as to scientific procedure and professional competency. The fact that the press may have sensationalized this report does not negate the fact that Rosenhan has published a very ambiguous piece of research that is open to serious scientific criticism.

The exact intent of the Rosenhan research is impossible to divine, for he makes allusion to just about every major issue in the mental health field with a lick and a promise. Yet an adequate rejoinder would have to deal with several major conceptual issues that Rosenhan never clearly delimits as arenas for discourse.

Moreover, his scientific methodology is open to serious question. He does not test his alleged major hypothesis, his methodolgy is irrelevant to the test question, and his data are tangential to his hypothesis. Therefore his conclusions can only be taken as assertions of his opinion, rather than tenable interpretations of his data.

Does Rosenhan present new findings? We must answer, no. To begin at the ending, I have no cavil with the observations and experiences reported by Rosenhan and his colleagues. In fact, I would strongly validate the reality and pervasiveness of many such hospital situations. But he describes nothing new. His observations are already immortalized in the work of his fellow Stanford colleague, Ken Kesey, who authored the best seller, One Flew Over the Cuckoo's Nest, which became a sell-out theatrical production. For those not familiar with the story (based on Kesey's experience as a ward attendant), a basically sane hero attempts to organize the psychotic patients on a hospital ward to oppose the malignant behavior of the treatment staff. Our sane hero is promptly diagnosed as a trouble-making psychotic who is eventually therapized to literal death. Thus the fact that sane persons can be labelled and treated in insane fashions has entered the common knowledge of public domain.

Does his work possibly lead to new solutions? Again we must answer, no. Rosenhan calls for more research. Yet the social psychology of mental institutions has been a major field for over twenty years. We have known what Rosenhan describes in exquisite theoretical and practical detail for over a decade. But he ignores the psychology of institutional change. Knowledge, per se, does not produce change. The ultimate irony of this regards Goffman. His classic research on total institutions was conducted at a famous hospital. When I interviewed staff at that hospital some ten years after Goffman, most knew of his book, most did not know it was a study of their hospital, and nothing had changed in the hospital in ten years post-Goffman.

Rosenhan suggests that if we do not send people to insane places, our impressions of them are likely to be less distorted. Here he assumes that psychiatric hospital units are inevitably bound to a gross distortion process. That is an assumption that can be empirically

Rosenhan has published ("On Being Sane in Insane Places") a very ambiguous piece of research that is open to serious scientific criticism.

tested. There is abundant evidence to indicate that organizational change is possible to redress the distortions Rosenhan describes.13

The Rosenhan alternative is to retain disturbed persons in their community for treatment, because their community is a non-pejorative environment. His support for community treatment programs is certainly consonant with our theoretical and therapeutic concepts of the day, but his rationale is not supported by evidence that the community is non-pejorative. The disturbed or deviant person is not labelled as such after he comes to the psychiatric hospital, but rather before he comes to the hospital.6,14

Furthermore, the whole community treatment process in the community is not devoid of labelling and social role assignment.2,9 It is only more covert, and thereby might even be more noxious. I am not arguing in support of the sad state of affairs in many psychiatric hospitals. But we will not escape the same labelling and dehumanizing processes by merely moving into the community.2 For example, Cumming1 has done a brilliant social role analysis of the total human services system in a community which demonstrates the same phenomena Rosenhan describes in the hospital.

Does he formally test a hypothesis? For the third time we must answer, no. This leads us hack to the central thesis of the Rosenhan piece. The stated hypothesis to be tested is one of clinical judgment. Can the psychiatric clinician distinguish-I cannot finish the sentence. Roseohan never states a clean hypothesis. Nor does he state which variables he intends to deal with in his research on clinical judgment.

Harty4 proposes five classes of variables involved in the assessment of clinical judgment:

    1. The nature of the judgment task; type of judgment required, and response alternative open to the judge.
    2. The nature of the input which provides the judge with his data.
    3. Characteristics of the judge which enter into the process; the types of cognitive operations performed, and               individual traits which affect these operations.
    4. The nature of the context in which judgments are made.
    5. Interactions among these classes of variables.

In terms of the first variable, Rosenhan fails to specify the nature of the judgment task. He uses terms such as sanity (a legal term), mental illness (a term of social convention), and schizophrenia (a technical diagnostic term) as synonyms. Then normality is thrown in-a notoriously ambiguous term-along with the issues of cross-cultural norms.

A careful reading reveals that a judgment is being required that Rosenhan does not specify. The task is this: Will judges (hospital staff) agree with a self-definition of the psychiatric patient role, and thereafter continue to judge the person in a consistent fashion according to the initial labelling definition despite contradictory behavioral and intellectual data?

Thus his study turns out to be research on role behavior, not research on clinical diagnostic decision making. Central to role theory is the process of role assumption and role-assignment. The data reveal that his subjects enter a role-assuming pose and receive congruent role-assignment. They claim to be patients, and in turn they are accepted and treated as patients. These are recurrent distressing data, but not novel findings. The research data do present some provoca
tive questions related to role-theory (not clinical diagnosis). For example, what social characteristics lead to role assignment on the basis of inadequate and/or inappropriate data? Or, what types of data must be introduced into the social transactional world to change role assignments, etc?

Although Rosenhan is dealing with role theory and labelling theory, he fails to relate his research at either a theoretical or applied level to the corpus of relevant research in medical sociology.

Rosenhan fails to provide substantive data on variables 2. and 3. while he does dwell on variable 4. and ignores variable 5.

Nor does he deal with any of the relevant empirical studies that deal with variables 2. and 3. For example, Gauron & Dickinson3 have shown that cognitive closure may lead to a diagnostic label unrelated to substantive data input. On the other hand, when one deals with scientifically competent judges, a high level of validity and reliability can be obtained on clinical psychiatric judgments.5,15

Reasons for diagnostic disagreement have been summarized by Ward, et al.9

A. Inconstancy on the part of the patient: 5% 
B. Inconstancy on the part of the diagnostician: 32.5% 
C. Inadequacies of the nosology: 62.5%

It is clear that diagnostic problems reside primarily in the issues of theoretical constructs, rather than the ability to accurately observe and make clinical deductions.

Rosenhan deals with none of this research, nor any of the conceptual issues involved. If his research were a study of clinical diagnostic judgment, then it would be incumbent upon him to propose some rationale for the fact that his conclusions are totally opposite to the empirical research in the field. But in fact, he has not conducted a study on clinical judgment, and therefore his conclusions in regard to the failure in psychiatric judgment are irrelevant.

This is the central theoretical issue at stake. For the distinction must be made between the scientific capacity to make reliable and valid clinical deductions resulting in a conceptual diagnosis, versus the use of diagnostic labels in the service of social role transactions.

It is possible to construct a research methodology that would address the stated hypothesis. Subjects could he presented to a panel of judges apart from the social role transactions of a treatment setting. Then we would study whether the judges could differentiate between those subjects who claimed to be emotionally disturbed but were not (the pseudo-psychotic, if you will) and those who were in actuality emotionally disturbed. Of course, this purely experimental situation does not address a second theoretical issue, namely, how do social, cultural, contextual, and transactional variables influence the processes of data collection and data evaluation. Here one would perforce deal with a complex matrix of interactive variables, requiring a quasi-experimental multi-matrix methodology.

These issues are by no means academic, for in many clinical settings, most notably in forensic psychiatry, the psychiatric clinician is requested to differentiate between feigned illness and actual illness. From personal clinical experience this is often a vexing problem if one approaches the task with scientific objectivity. For example, the Ganser syndrome allegedly describes a person, usually in a legal setting, who claims mental illness to avoid legal penalties. The long history of controversy about this syndrome illustrates the difficulty involved in assessing a person in that social role. Thus the questions which Rosenhan raises are by no means trivial. One possible argument might be that social role assumption and role ascription are central variables in any assessment of a person. As one thinks he is, so he is. Which immediately leads us into issues of phenomenology and philosophy as Rosenhan hints.

Inhumane institutional practices in part reflect the demands and expectations of society.

As the Rosenhan report exists, it suggests that the problems lie solely with the mental health professions and psychiatric institutions. However, institutions and professional practice exist in reciprocal relation to public attitudes and public demands. Inhumane institutional practices in part reflect the demands and expectations of the society. The rejection and dehumanizing of the psychiatric patient within the institution can be seen as a projection and acting out of the community rejection and dehumanization of the labelled deviant.1,2,8

However, solutions do not come from blaming the public, the institution, nor a profession. For blame demands punishment. And while punishment may appease it will not necessarily produce change. Our humanistic desire for fundamental changes in our response to deviant behavior requires that we not be defensive nor protectionistie about basic problems in our society, institutions, and professions.9 Therefore, the conclusions and recommendations that Rosenhan proposes miss the central issues and end up as seapegoating observations rather than as catalytic clarification. In my opinion, Rosenhan ends up doing what he decries. He labels behavior instead of conducting an accurate assessment.

Implicit Social Criticism

Having stated my critique, I return to the author's assertion that his study was merely a piece of research on the diagnostic leap from single symptom to diagnosis. If the critique I have made has validity, we must conclude that Rosenhan either engaged in some incredibly sloppy research in which his conclusions were unrelated to and unsubstantiated by his research data; or that he ignored a substantial body of research that totally contradicts his method and conclusions; or that he had other purposes in mind, not reflected in his stated research aim.

Inasmuch as Rosenhan is a respected scholar who has published well known material, it seems implausible that he would tolerate uncritical research or ignore substantive research publications relevant to his work. On the other hand, there are several indications that Rosenhan may have had other implicit goals in mind in his publication. First, he presents simplistic attacks on every complex issue in mental health in his original article. Why attack every issue-with no suggestions or discussion, in a research article? Second, he concludes that mental hospitals are bad and should be abolished. Granted the deplorable state of some institutions, why the desire to throw the baby out with the bathwater? Third, he repeatedly attacks psychiatrists for being presumptuous, or at least disdainful of scientific data. This is the facet of psychiatric research data directly relevant to his research. Who is he attacking? These are observations on the implicit tone of his article.

In his subsequent rebuttal to letters, he comes more directly to his implicit concerns. Basically, he is concerned about how we study man. To his mind man can be studied only in terms of objective external tests and measures. What man says and does-man's testimony of himself does not constitute scientific reliable and valid data. Rosenhan says: ". . . (diagnosis) is not independently verifiable beyond what a patient says and does. "12 Thus Rosenhan is back arguing a type of Logical Positivism philosophy. He wants a laboratory operational approach to the study of man.
I suppose this approach to the study of man might fall under the now popular category of behaviorism. One need not quarrel with Rosenhan for taking this position, which is certainly a tenable way to study man, albeit only one view of man. But since this argument is an old one, argued many times in the psychological literature, why should Rosenhan raise this issue in such a covert and tangential manner?

Should social criticism and empirical research be combined? . . It seems most necessary that we do not subvert scientific research and publication as propaganda for a social position.

Let us pursue the matter one step further. He is opposed to the use of psychiatric diagnosis. Ostensibly because it is scientifically inaccurate and based on subjective patient self-reports rather than objective laboratory data. But why are psychiatric diagnoses disturbing to him? In his original article he states: "Psychological suffering exists. But normality and abnormality, sanity and insanity, and the diagnoses that flow from them may be less substantive than many believe them to be." This quote may not seem very clear, and Rosenhan does not exegete his concerns that flow from the use of diagnostic labels, But I should like to suggest some issues currently in hot debate in our society which I believe Rosenhan ultimately wishes to address.

(1) There is the real concern about the potential role of the psychiatrist as an agent for political social control. Recent cases in Russia suggest that political foes have been declared insane and imprisoned in psychiatric hospitals as pseudo-patients. Such allegations have not been fully investigated, but it raises similar concerns in our American society.
 (2) Over the past ten years there has been increasing concern for the civil rights and civil liberties of the patient admitted to a psychiatric facility. A joint task force of the American Bar Association and the American Psychiatric Association met to draft model legal code revisions for admission procedures and civil rights of hospitalized patients. Many states have since adopted versions of this model legislation, although there are continuing inequities in many parts of the country. (3) The liberation movements of the 1960's were reflected in a "radical left" movement in American psychiatry, led by Thomas Szasz in America and Ronald D. Laing in England. In effect they proclaimed the "myth of mental illness". To them and others in the movement, mental illness was the product of social oppression. Thus society was sick and made unrealistic demands for conformity, or labelled those who deviated from traditional social convention as "mentally ill" in order to control them. This position has been joined by certain sociologists such as Thomas Scheff, who argue for a social role theory of mental illness. In brief, these sociologists argue that mental illness is nothing more than a deviant social role created by society. The radical left therefore demands the elimination of psychiatric diagnoses because such diagnoses are means of social manipulation that hurt people.

In this light we can see that Roscnhan's concern for psychiatric diagnosis fits with a certain zeitgeist. He is raising an argument, in line with other social critics, of the possible social misuses and abuses of the mental health system in society. At this point I can now note that Rosenhan is not only a psychologist, but is a law professor. This is potentially significant in that lawyers take a leading role in much of the social criticism I have alluded to. These admittedly loosely connected observations, taken as a whole, suggest that Rosenhan is not concerned with a narrow research question on psychiatric diagnostic method, but rather is assuming the role of a social critic.

Now it should he stated that in my opinion there is considerable reason for concern in each of the three areas of social criticism outlined above. Thus one cannot fault Rosenhan for being a social critic, nor can we fault him for raising issues relevant to any of these social concerns. But if my major thesis stands, namely, that Rosenhan has published a highly visible piece of social criticism, then several issues present themselves.

Scientific Responsibility

I have taken considerable space to present a rather technical analysis of this piece of science publishing to illustrate how social criticism can be embedded in empirical research. I have concluded that in this instance we have a confusing combination of the two. And this type of combination raises serious questions in my mind.

(1) Should social criticism and empirical research be combined? I would answer a qualified yes. Particularly in the social and psychological sciences it is often impossible to separate empirical research from basic social positions. Indeed the separation may not be desirable, for research in relation to social positions is critical to social evaluation. However, I consider it poor science and potentially destructive to science and the larger society to confuse a position of social criticism with the research pertaining thereto.

The dangers are twofold. (a) It may preclude a clear analysis of the empirical data. (h) It may lead to dismissal of the data because of the social position it supports or negates; or conversely the data may lend undue credence to a social position solely on the merits of the present data of the study.

(2) How should social criticism and empirical research be combined? When no distinction is made between the two in a report, then the above dangers are encountered. Those dangers are in a sense logical and technical problems of accurate reading of a report. But an unclear combination also confuses the basic distinction between science and social policy. Science cannot determine human attitudes or define social policy. To my mind social criticism and the ensuing debate over humane and moral directions for social action cannot be resolved by appeal to empirical data alone. More social action is a uniquely human responsibility. 9 15

Therefore, it seems most necessary that we do not subvert scientific research and publication as propaganda for a social position. Conversely, we should not shirk the responsibility to engage in forthright social criticism and social moral dialogue. To confuse the two can lead only to discredit of intellectual autonomy in the scientific enterprise-as in the Lysenko science of the Stalinist era. And just as important, it makes social criticism an objective amoral affair, rather than the moral responsibility of all of us in a human society.

On these counts, then, the Rosenhan publication presents an example of dubious procedure that should he cause for concern for both the scientist and the social critic. In fact I happen to agree with much of the Rosenhan criticism and I am largely in sympathy with his social positions. However I strenuously object to his perhaps unwitting subervision of both science and social criticism. For in this instance we all lose rather than gain.


1Gumming F. Systems of Social Regulation. New York: Atherton, 1968.
2Duff, F. S. & Hollingshead, A. B. Sickness and Society. New York; Harper & Row, 1968.
3Cauron, E. F. & Dickinson, J. K. Diagnostic Decision Making in Psychiatry. Arch. Gen. Psychiat. 14:225, 1966.
4Harty, M. K. Studies in Clinical Judgment. Bull. Meninger
Cl. 35:335, 1971.
5Kendell, R. E., Everett, B., Cooper, J. E., Sartorious, N., & David, M. E. The Reliability of the 'Present State' Examination. Social Psychiat. 3:123, 1968.
6Levinson, D. J., Mesrilield, J., & Berg, K. Becoming a Patient. Arch. Gen. Psychiat. 17:385, 1967.
7Pattison, E. M., Coe, F. & Rhodes, R. 3. Evaluation of Alcoholism Treatment: Comparison of Three Facilities. Arch.
Gen. Psychiat
. 20:478, 1969.
9Pattison,, E. M., Bishop, L. A., & Linsky, A. S. Changes in Public Attitudes on Narcotic Addiction. Amer. J. Psychiat.
125:160, 1968.
10Pattison, E. M. Psychosocial and Religious Aspects of Medical Ethics. In: To Lice and To Die. R. H. Williams (ed.) New York: Springer-Verlag, 1973.
11Rosenhan, D. L. On Being Sane in Insane Places. Science 179:250, 1973.
12Rosenhan, D. L. Response to Letters to the Editor. Science 180:365, 1973.
13Schwartz, M. S. & Schwartz, C. C. Social Approaches to Mental Patient Care. New York: Columbia Univ. Press, 1964.
l4Tischler, G. T. Decision-Making Process in the Emergency Room. Arch. Gen. Psychiat. 14:69, 1966.
l5Ward, C. H., Beck, A. T., Mendelson, M., Mock, J. E., & Erbaugh, J. K. The Psychiatric Nomenclature. Arch. Gen. Psychiat. 7:198, 1962.