When I was a second year psychiatry resident in the 1980s, I once had occasion to go over a CT scan of the brain with the chief resident in radiology. CT technology was still relatively new back then, and, after discussing the findings with the resident, I said, “So, the scan was read as normal?” He glared at me with withering contempt. “It wasn’t read as normal,” he growled, “it is normal!”
He was expressing a common but mistaken understanding of science and medicine; namely, that objective and certain knowledge is possible, apart from acts of interpretation. Some critics of psychiatry make the same mistake when they claim to know what the research literature on schizophrenia or depression “really shows.” In fact, the literature itself “shows” nothing, absent its often contested interpretation.
No, I’m not advancing the post-modernist notion that “there is no truth,” or that all “narratives” are of equal value. I am arguing that genuine science produces only tentative and provisional knowledge, always subject to revision according to new evidence–and new interpretations.
Speaking of new interpretations, here’s a quick quiz. Which medical specialty decides what is or isn’t pathological, by taking a vote on the question? Which medical specialty is frequently roiled by controversy as to what is, or is not, “normal”? If you answered, “Psychiatry”, you would be only partly right—you could easily have answered, “Oncology” and been equally correct.
This point was recently driven home when an international panel of medical experts decided that “…a type of tumor that was classified as a cancer is not a cancer at all.”  As reported in JAMA Oncology, the re-classified tumor is a lump in the thyroid that is completely surrounded by a fibrous capsule . The nuclei of the tumor cells look like those of cancer cells, but the tumor is completely contained and treatment is unnecessary.
What was once classified as a papillary thyroid carcinoma is now called a “noninvasive follicular thyroid neoplasm with papillary-like nuclear features”—a NIFTP. Poof—no more thyroid cancer! Interestingly, while the majority of the panel was composed of pathologists, one member was a psychiatrist “who knew the impact a cancer diagnosis could have”  on vulnerable patients—an instance in which on ostensibly “objective” and “scientific” decision was actually influenced by overtly subjective considerations.
To be sure, the “vote” that decided the matter was not capricious or arbitrary; it was based on careful follow-up of several hundred cases of supposed thyroid cancer. The finding was clear: none of the patients whose tumors remained encapsulated had any evidence of cancer after 10 years.
The panel therefore shifted from using “nuclear features” to “the presence of invasion” as the criterion for cancer. This change may well prevent unnecessary and potentially harmful treatment—but the decision itself represents an exercise in interpretation and judgment.
Now, on one level, it is clearly not the case that psychiatry and oncology are comparable fields—the “sensory data” each discipline deals with are quite different. Psychiatrists, for the most part, process words, gestures and behaviors.
Oncologists, to oversimplify, generally process lumps, labs and slides. Yet these inter-specialty differences may conceal important similarities. Thus, psychiatrists also process observable physical data (“signs”), such as weight loss and psychomotor slowing or agitation; and, for their part, oncologists must deal with the often devastating emotional reactions of their patients.
Furthermore, within their respective provinces of perception, both psychiatrists and oncologists engage in acts of interpretation, based on ever-changing empirical data.
For example, the DSM-5 made the rather radical decision to eliminate all the traditional subtypes of schizophrenia—paranoid, disorganized, catatonic, etc. For many of us, this came as something of a surprise–but the decision was based on considerable empirical evidence.
Specifically, as Dr. Rajiv Tandon has noted, “…these subtypes have limited diagnostic stability, low reliability, poor validity, and little clinical utility.”.
Now, one can argue—and many have—that numerous DSM-5 diagnoses lack validity, based on the most rigorous validity criteria developed by Robins & Guze . But the process of investigation and interpretation that goes on in psychiatry is fundamentally the same as in general medicine—even though the objects of investigation may be quite different.
Scientism, Not Science
As to the objection that psychiatric diagnoses are not “scientific” because they are not grounded in clear-cut biological findings, this represents a misunderstanding of the scientific process. Indeed, of the original five Robins & Guze (1970) criteria for validity, only one was specifically “biological.”
The five phases proposed to achieve valid classification of mental disorders included clinical description, laboratory study, exclusion of other disorders, follow-up studies and family studies [4,5]. As Dr. Bernard Carroll has pointed out, “The existence of a medical disorder is not predicated on having a laboratory test for it.” (personal communication, 4/11/16). (What, after all, is the imaging study or laboratory test for validating migraine headaches or atypical facial pain?). Dr. Carroll elaborates this crucial point:
“…biomarkers are not an automatic gold standard of evidence for diagnostic validity. Indeed, thoughtless diagnostic testing can cause mischief through unnecessary further testing and treatments. Laboratory measures are the servants of clinical science, not the other way around, because most diagnostic tests are probabilistic rather than pathognomonic—so clinical judgment enters into their selection and their interpretation.” 
Indeed, the insistence on lab tests or other biological criteria for the “scientific” diagnosis of disease is not science, but scientism—“an exaggerated trust in the efficacy of the methods of natural science applied to all areas of investigation (as in philosophy, the social sciences, and the humanities)” . Psychiatry, after all, is not a “natural science” in the sense that anatomy or biochemistry is. Rather, as Dr. Jose de Leon has stated,
“…psychiatry is…a hybrid scientific discipline that should combine the methods of the natural sciences (defined as the empirical sciences which study the natural world) and the social sciences. These sciences provide, respectively, an explanation of illness that follows the medical model and an understanding of psychiatric abnormalities that are variations of human living.” 
Furthermore, a purely “biological” explanation of a psychiatric problem—even if correct—does not necessarily mean that a biological approach to treatment will be the most effective.
For example, suppose we were able to identify a specific “neurocircuit” that was causally related to post-traumatic stress disorder. It would not follow that a biological intervention, such as a medication, would be superior to, say, some form of trauma-specific psychotherapy.
Moreover, the whole notion that we should separate “biological” from “psychological” treatments** can be challenged on both philosophical and clinical grounds, as Dr. Glen Gabbard has forcefully argued. 
All that said, we should not succumb to the myth that psychiatry has no “biology” at all. While “office-ready” biomarkers of psychiatric illness are probably years away, there is growing evidence that schizophrenia, melancholic major depression, obsessive-compulsive disorder and bipolar disorder (among others) are associated with specific biological abnormalities; e.g., “…the occurrence of ocular motor dysfunction in schizophrenia patients and their first-degree biological relatives is remarkably consistent…” . Meanwhile, on the clinical level, a recent meta-analysis by Leucht et al found that psychiatric medication is roughly as effective as medications used in general medicine