The Biopsychosocial Model: What It Promises and Where It Stops

This post contains Amazon affiliate links. If you purchase through them, I earn a small commission at no extra cost to you.

This is a post in the Clinical Foundations Series.

George Engel titled his paper in Science the way someone titles an argument rather than a proposal: “The Need for a New Medical Model: A Challenge for Biomedicine.” Engel was an internist by training who held appointments in both medicine and psychiatry at the University of Rochester, which gave his critique an unusual vantage. He was arguing against the house he lived in.

To understand why the argument landed as it did, it helps to understand what medicine looked like when he made it. The mid-twentieth century had produced extraordinary gains through reductionist science (antibiotics, vaccines, surgical advances), and those gains had cemented a particular way of thinking about disease: find the pathogen, the lesion, the chemical deficit, and correct it. The body was, in this framework, a biological mechanism. When it failed, the failure had a molecular explanation, and the clinician’s job was to locate and address it. This was not an unreasonable position given what medicine had accomplished. It became a problem when it was applied wholesale to conditions that did not fit the model, so that behavioral, psychological, and social dimensions of illness were treated not as factors requiring attention but as noise to be filtered out.

Psychiatry’s struggle with this framework was particularly acute, and the stakes were not merely philosophical. The field faced a legitimacy crisis throughout the mid-twentieth century, caught between those who, like Thomas Szasz, argued that mental illness was a myth and those who argued that it was a genuine disease that simply hadn’t yet been explained at the biological level. The second position won, with consequences that are still playing out. The search for a biological substrate of mental illness became the organizing project of academic psychiatry, and the serotonin hypothesis of depression became its most culturally visible product. Depression, the hypothesis held, was the result of insufficient serotonin activity in the brain. Antidepressants worked by correcting that deficit. The explanation was clean, teachable, and not borne out by the evidence.

Antidepressants work. That is not the contested claim. The evidence for their efficacy in adults with major depression is substantial , and the current political climate makes it worth stating plainly. What is contested is the mechanism, and a comprehensive umbrella review of the research has since found no consistent support for the serotonin explanation . The research does not support the proposition that depression is caused by low serotonin or that antidepressants work by correcting a serotonin deficiency—any more than headaches are caused by an acetaminophen deficiency. They work, at least in part, by some other means that remains incompletely understood. The serotonin hypothesis was not a discovery; it was a story told in the language of molecular medicine because that was the language psychiatry needed to speak to be taken seriously. Because the story was so clean and so widely repeated (to patients, in direct-to-consumer advertising, in medical education), it created the conditions for a specific kind of clinical error: the assumption that a depressed patient’s primary need is a medication that targets serotonin, rather than an assessment of what is actually driving her distress and what range of interventions might address it. Inappropriate prescribing is not a failure of the drugs. It is a failure of the framework that reduced the clinical question to a chemical one.

Engel’s proposed remedy was the biopsychosocial model: a framework that would hold biological, psychological, and social factors in view simultaneously, that would treat the patient’s subjective experience as clinically meaningful rather than as noise, and that would give medicine the conceptual architecture to understand illness as something that happened to a person embedded in a social world rather than to a body in isolation. The ambition was genuine and the critique was correct. His paper shaped medical education and clinical culture in ways that persist nearly fifty years later.

What the biopsychosocial model did not do—what it could not do, as its own sympathetic critics have since demonstrated—was tell clinicians what to do next.

The Reassertion

If the biopsychosocial model’s influence were as broad as its adoption in medical education would suggest, we would expect the decades since 1977 to show a gradual broadening of how psychiatry and medicine conceptualize and treat mental illness. The data do not support that expectation. Studies of research published in major psychiatric journals show that biomedical framing became more prevalent in the decades after the biopsychosocial model became stated orthodoxy, not less . The framework that was meant to broaden the clinical gaze did not prevent the narrowing. At best, it provided a nominally inclusive banner under which reductionist research proceeded largely on its own terms.

The most ambitious recent version of that reductionist impulse was the Research Domain Criteria (RDoC) initiative launched by the National Institute of Mental Health (NIMH) in 2010, a project that proposed reorienting psychiatric classification away from symptom-based diagnosis and toward neuroscience and biomarkers. Thomas Insel, who directed NIMH for thirteen years and championed RDoC, later offered a frank assessment of what that era of research accomplished. The science it generated was real. The improvement in patient outcomes, measured by rates of suicide, hospitalization, disability, and recovery, was not . The aspiration to find fMRI signatures or genetic markers that would guide diagnosis and treatment selection ran into a problem that the biopsychosocial model, whatever its limitations, had always understood: mental illness does not reduce cleanly to biology. The biomarkers that would make psychiatric diagnosis as legible as a chest X-ray have not materialized, and the reasons they haven’t are not merely technical. They are conceptual. Our diagnostic categories were not built for biological validity, and mapping brain circuits onto symptom clusters has not yet yielded the clinical precision the project promised.

The commercial market has moved faster than the evidence. A number of companies now market pharmacogenomic testing to clinicians and, increasingly, directly to patients, promising to identify which psychiatric medications a patient’s genetic profile favors. The proposition is seductive. Instead of the trial-and-error process that characterizes psychotropic prescribing, a cheek swab generates a report that sorts medications into color-coded categories: green for use as directed, yellow for use with caution, red for use with increased caution and more frequent monitoring. Patients bring these reports into primary care visits and hand them to their primary care providers (PCPs) with a reasonable expectation that the science is settled. It is not. The largest blinded randomized controlled trial of the leading commercial product did not reach significance on its primary endpoint of symptom improvement . It did show statistically significant differences on secondary endpoints, with response rates of 26% versus 20% and remission rates of 15% versus 10% favoring guided care, but those findings require careful interpretation. The absolute differences amount to roughly five or six additional patients per hundred reaching response or remission, in a population that had already failed a mean of three and a half medication trials. Treating clinicians were not blinded to study arm, and several co-authors were employees of the test’s manufacturer. Statistical significance and clinical meaningfulness are not the same thing, and whether those margins justify the cost of testing, the access barriers it creates, and the certainty with which results are marketed is a question the trial does not answer. A review conducted for the American Psychiatric Association’s task force on biomarkers concluded that the evidence did not support widespread use of combinatorial pharmacogenetic testing, and noted that much of the literature had been produced by investigators with financial interests in the outcome . Pharmacogenomics may yet earn a place in psychiatric care. For now, the distance between what the science supports and what is being sold is wide, and the premise underneath the marketing, that psychiatric treatment can and should be guided primarily by biological data, is already shaping prescribing decisions in primary care.

Neuroscience is not irrelevant to psychiatric care, and the search for better biological understanding is not misguided. The problem is the biomedical model’s tendency to treat biological explanation as the terminal point of clinical inquiry rather than as one level of a more complex picture, a tendency that keeps reasserting itself despite the evidence that it is insufficient. Engel diagnosed it correctly. The biopsychosocial model he proposed to address it did not cure it.

A Map Without a Route

Borrell-Carrió, Suchman, and Epstein revisited Engel’s model twenty-five years after its introduction in the Annals of Family Medicine, a venue that signals, from the title page, an audience of clinicians rather than theorists. They wrote as defenders of the model, not its critics. They acknowledged Engel’s contribution, worked through the model’s philosophical commitments, and then proposed a set of practical pillars for what they called biopsychosocially oriented clinical practice: self-awareness, active cultivation of trust, empathic curiosity, self-calibration, educated emotions, informed intuition, and communicating evidence as dialogue.

This is a list of what a good clinician looks like. It is not a procedure. The distinction matters, because a framework that can tell you what virtue resembles without telling you what action to take leaves the clinician standing in the exam room with the same practical question she had before she encountered it. A forty-five-year-old man presents with low back pain, an elevated PHQ-9, and a chart note from his last visit flagging alcohol use. You understand, because you absorbed the biopsychosocial model in graduate school, that his pain and his mood and his drinking are likely interrelated, operating across biological, psychological, and social levels at once. You hold that understanding. What do you do next?

The model does not say.

Borrell-Carrió and colleagues conceded as much themselves, if quietly. The clinical practice they were describing, they wrote, does not necessarily follow from the model’s theoretical constructs; it rests instead on the content and emotional climate of the clinician’s relationship with the patient . Read plainly, that is an admission that the model is clinically inert at the point of intervention. It describes a stance. It does not generate a move.

The Permission Slip Problem

The most pointed version of this critique comes from S. Nassir Ghaemi, who objected to the model’s “eclectic freedom”: a clinician can emphasize the biological, the psychological, or the social as preference dictates, an openness Ghaemi regarded as bordering on anarchy . The charge sounds dramatic, but the clinical observation beneath it is familiar to anyone who has watched the model deployed in practice. Because the framework accommodates biological, psychological, and social explanations with equal theoretical legitimacy, it provides no mechanism for adjudicating between them. A clinician who emphasizes the neurobiological substrate of depression is practicing biopsychosocially. So is a clinician who foregrounds adverse childhood experiences. So is a clinician who orders a medication and ends the visit in eight minutes. The model cannot distinguish between these approaches, because it was never designed to. It was designed to ensure that all three domains were held in view. Whether they are actually held, and what the clinician does when they are, was left to individual judgment, training, and luck.

Benning , reviewing the critical literature, observed that the model’s sharpest critics reach their conclusions by different routes, and that it is difficult to avoid agreeing with at least one of them. Either the biopsychosocial model failed to hold back biomedical reductionism, or it licensed an eclecticism loose enough that reductionism could proceed under its name. For the clinician in the exam room, the practical result is the same. The model functions as a permission slip: you may attend to whichever domain you find most tractable and call it integration.

Habit and the Map

In the chapter on habit in The Principles of Psychology, William James observed that habit functions as “the enormous fly-wheel of society, its most precious conservative agent,” the force that keeps behavior within its grooves regardless of what the mind knows or intends. The grooves, once established, persist. They organize action not because they are chosen moment to moment but because they have been worn in by repetition, and because the alternatives require more than recognition to become available. You cannot simply see that a different path exists. You need a procedure for taking it.

This is the gap the biopsychosocial model leaves open. It can produce recognition: recognition that the patient’s biology and psychology and social circumstances are interrelated, that the low back pain and the elevated PHQ-9 and the alcohol use are not independent facts to be addressed in sequence but facets of a single clinical picture. Recognition is not nothing. Engel was right that medicine needed it. But recognition is a map. It shows you where things are. It does not put your foot on the path.

When a framework provides recognition without procedure, the flywheel takes over. Clinicians default to the grooves worn deepest in training, and for most clinicians trained in Western medicine those grooves run biomedical. The biopsychosocial model’s failure, then, is not neutral; it has a direction. A framework permissive enough to accommodate any clinical emphasis will, under the ordinary pressures of practice (time, reimbursement, institutional culture, the cognitive load of a full panel), tend to settle on whichever emphasis asks the least departure from the familiar. The serotonin story, the RDoC initiative, and the pharmacogenomic testing market are three versions of that drift. Seeing the groove does not get you out of it.

What is needed is something closer to what James described in the practical maxims that follow his flywheel passage: a procedure explicit enough to interrupt the groove, to launch a new habit with what he called “as strong and decided an initiative as possible” , and to make the alternative available not as a theoretical possibility but as a concrete next move. The biopsychosocial model cannot provide this. Its ambitions were descriptive and corrective, not procedural. It wanted to expand the frame, and it left the question of what to do inside the expanded frame to others.

What the Model Leaves Behind

That question has not gone unanswered. The gap the biopsychosocial model describes but cannot fill is precisely the space that functional contextualism occupies, a philosophical and clinical framework grounded not in the question of what a problem is but in what it does, and in what intervention would produce a different outcome in this context, with this person, toward this end. It is the same question that ran through The BHC as Mentat. Where the biopsychosocial model asks the clinician to hold biological, psychological, and social factors in view, functional contextualism asks what holding them in view is for, and what action follows from the seeing. The functional question is not a supplement to biopsychosocial thinking but its operationalization. Clinical inertness is the diagnosis; functional contextualism is what the treatment looks like.

For behavioral health consultants (BHCs) working in primary care, this is not an abstract distinction. The biopsychosocial model is the water we swim in, embedded in the language of integrated care, in the rationale for having a BHC in the clinic at all, and in the framework most of our PCP colleagues absorbed in training. Understanding it well means understanding both what it accomplished and where it stops. Engel diagnosed a real problem and proposed a genuine remedy. The remedy expanded the frame. What carries patients through that expanded frame is something the model itself could not provide, and fifty years of evidence suggests we cannot simply wait for the biopsychosocial ideal to assert itself against the pull of the grooves.

That is the subject of the next post in this series.

This post was written by the author. Claude (Anthropic) assisted with research, drafting, and editorial refinement. All clinical content, interpretations, and recommendations reflect the author’s own professional judgment.