The Psychiatry We Lost
For thirty years, psychiatry bet everything on the biological model of mental illness. The drugs worked, sort of, for some people. What got lost was everything else — and now a generation is paying the price.
In 1990, the National Institute of Mental Health declared the coming decade the "Decade of the Brain." The statement was an announcement of a wager: psychiatry, which had spent its first century oscillating between psychoanalytic interpretation and biological determinism, was going to commit. The brain was a biological organ. Mental illness was a biological malfunction. The solution was biological — molecular, pharmaceutical, tractable to the methods of bench science. Within a generation, the announcement implied, we would understand the neurochemical basis of depression, schizophrenia, anxiety, and addiction, and we would treat them accordingly.
The decade of the brain became the decade of the antidepressant. Fluoxetine, which Eli Lilly had introduced to the market in 1987 as Prozac, was the leading edge of a wave of SSRIs and SNRIs that, by the mid-1990s, had transformed psychiatric practice in the United States and, progressively, throughout the developed world. Peter Kramer's 1993 book Listening to Prozac captured the cultural moment: here, at last, was a molecule that reached into the brain's chemistry and corrected what had gone wrong, with fewer side effects than the tricyclics and MAOIs it was replacing, without the decades-long commitment of psychoanalysis, without the still-stigmatized option of electroconvulsive therapy. Psychiatry had made its bet, and for a brief, lucid moment, the bet seemed to be paying off.
It is now thirty years later, and the reckoning is underway.
The Theory That Wasn't
The chemical imbalance theory of depression — the idea that depression results from insufficient serotonin in the brain's synapses, and that SSRIs work by restoring serotonin to its correct level — was never a scientific finding. It was a hypothesis, advanced in the 1960s and refined in subsequent decades, that was convenient, communicable, and not quite true.
The evidentiary problems with the serotonin hypothesis were documented in the scientific literature long before they reached public consciousness. The inconvenient facts accumulated: healthy volunteers given drugs that temporarily depleted serotonin did not develop depression. Some patients with depression had elevated serotonin metabolite levels, not reduced ones. Tianeptine, an antidepressant that works by reducing synaptic serotonin rather than increasing it, proved effective in clinical trials, which would be impossible if the theory were correct. By the late 1990s, most research psychiatrists had moved to more nuanced neurochemical models, involving multiple neurotransmitter systems, neuroplasticity, and inflammation pathways.
What did not change was the marketing. The pharmaceutical industry had built a profitable and publicly legible story around the chemical imbalance concept, and that story was too useful to complicate. Eli Lilly's advertising for Prozac included graphics showing a synapse with a deficit of serotonin. Pfizer's Zoloft campaign featured an animated oval shape, visibly unhappy, drifting through scenes until serotonin (represented as floating orbs) restored its equilibrium. These images communicated something that wasn't true and couldn't be corrected without undermining the commercial foundation of the entire sector.
The public scientific reckoning arrived in July 2022, when a systematic review led by Joanna Moncrieff at University College London, published in Molecular Psychiatry, synthesized the existing evidence on the serotonin hypothesis and found no consistent support for it. Moncrieff and her colleagues were not saying that SSRIs don't work — the evidence for their efficacy in moderate-to-severe depression, relative to placebo, is real, if modest. They were saying that the mechanism by which they work is not the one that has been publicly communicated for three decades. The review was widely covered and widely misunderstood: a significant fraction of commentary treated it as a claim that antidepressants are useless, which is not what the evidence says. But the correct reading is troubling in its own way. We do not, in 2026, have an adequate scientific account of why antidepressants work for the people they work for. The most plausible current hypothesis involves their effects on neuroplasticity and on the brain's capacity to form new connections — a considerably more complicated story than serotonin levels, and one that has no corresponding pharmaceutical campaign.
The DSM Machine
Alongside the pharmaceutical transformation of psychiatry ran a taxonomic one. The Diagnostic and Statistical Manual of Mental Disorders, in its third edition published in 1980, was a deliberate act of paradigm replacement. Robert Spitzer and his collaborators designed DSM-III to move psychiatry away from psychoanalytic diagnostic categories — which were based on theories about unconscious conflict that were clinically unreliable — toward a descriptive taxonomy based on observable symptoms. You did not need to know why someone was depressed to diagnose them with Major Depressive Disorder; you needed to observe that they met a specified number of symptom criteria over a specified duration.
The DSM-III was, in the context of the 1980s, a genuine improvement in diagnostic reliability. It created a common language for clinical practice and research, and it made psychiatric diagnosis legible to insurance companies — a practical consideration that was not incidental to its adoption. But it also imported a specific philosophy of mental illness that was presented as theoretically neutral and was anything but. By defining disorders through symptom checklists without reference to etiology, the DSM encoded a view of mental illness as a category of discrete biological dysfunctions, each with its own set of necessary and sufficient criteria, waiting to be matched with a targeted pharmaceutical intervention. The analogy was to infectious disease: diagnose the pathogen, prescribe the appropriate drug, monitor for response.
The analogy was wrong, and the DSM expansion that followed DSM-III demonstrated why. Each successive revision — DSM-IIIR, DSM-IV, DSM-5 — expanded the category of diagnosable disorder, lowered the threshold for many diagnoses, and created new ones. DSM-5, published in 2013, removed the bereavement exclusion from Major Depressive Disorder: under DSM-5, a person experiencing grief symptoms after losing a spouse could be diagnosed with MDD as early as two weeks after the death. This change was contested by multiple prominent psychiatrists, including Allen Frances, who had chaired the DSM-IV task force and published a scathing book, Saving Normal, about the pathologization of ordinary human experience. The expansion of diagnosable disorders was not driven primarily by scientific discovery; it was driven by the logic of a diagnostic system that incentivized pharmaceutical treatment and by a pharmaceutical industry that needed new markets.
The result is a situation in which the United States simultaneously overdiagnoses and underserves. Twelve percent of Americans aged twelve and over are currently taking antidepressants, according to CDC data. The treatment gap — the proportion of people with diagnosable mental illness who receive no treatment — remains above forty percent. The people receiving treatment are frequently receiving it in the form of a prescription from a primary care physician who has had a fifteen-minute appointment, without access to psychotherapy, without adequate follow-up. The people not receiving treatment are often unable to access care for structural reasons: cost, availability, stigma, or the sheer inadequacy of a system that has been organized around pharmacological management rather than comprehensive care.
The Whitaker Problem
Robert Whitaker is the most uncomfortable figure in contemporary mental health journalism. His books — Mad in America (2002) and Anatomy of an Epidemic (2010) — argue that the pharmaceutical treatment of mental illness in the United States has, at the population level, made outcomes worse rather than better: that the rise in disability rates associated with mental illness, which increased dramatically in the decades following the introduction of SSRIs and atypical antipsychotics, is causally related to the drugs themselves rather than to any increase in underlying rates of illness.
Whitaker's argument is contested. The most sophisticated critics point out that the causal inference is difficult — disability rates and diagnosis rates rose together, but disentangling the drug effect from the diagnosis effect from the treatment-seeking effect from genuine changes in social conditions (isolation, economic precarity, the digital disruption of social connection) is genuinely hard. The mainstream psychiatric establishment has been largely hostile to Whitaker's work, and the hostility contains both legitimate scientific objections and what reads as institutional defensiveness.
What is harder to dismiss is his documentation of specific failures. The trial data on long-term antipsychotic use, particularly for schizophrenia, shows a pattern that the clinical establishment was slow to acknowledge: many patients do better, by standard outcome measures including relapse rate and quality of life, on reduced doses or after gradual tapering than on the standard long-term maintenance doses that became clinical practice in the 1980s. This is not an argument for no medication; it is an argument that the evidence base for standard dosing practices was weaker than it was presented to be, and that the clinical culture that developed around long-term medication management was more confident than the data warranted.
Whitaker's fundamental argument — that psychiatry organized itself around a biological model that was partly wrong and hasn't fully reckoned with that fact — is not his alone. It is increasingly voiced by researchers within the establishment: by the former NIMH director Thomas Insel, who wrote in his 2022 memoir that after decades of funding neuroscience research, the field had made almost no progress on the question of what actually helps people with mental illness recover; by Irving Kirsch, whose meta-analysis of antidepressant trial data found that the drug-placebo difference, while statistically significant, was often below the threshold of clinical significance for mild-to-moderate depression; by the working groups within APA itself that are arguing for a more dimensional, transdiagnostic approach to classification.
The Psychedelic Exception
In 2023, the FDA granted MDMA-assisted psychotherapy "Breakthrough Therapy" designation for PTSD treatment, before a final approval process that became more complicated due to data irregularities at a principal clinical trial site. In parallel, psilocybin-assisted therapy has generated phase 2 clinical trial results for treatment-resistant depression that are remarkable by the standards of a field accustomed to modest effect sizes: some studies showing sustained remission in a substantial proportion of patients who had not responded to multiple prior treatments.
The psychedelic renaissance is interesting not only for its clinical promise but for what it reveals about the inadequacy of the model it is challenging. MDMA and psilocybin work — to the extent that current evidence suggests they work — not by correcting a neurochemical deficit but by creating a specific kind of psychological experience, within a therapeutic context, that enables something like emotional processing. The mechanism, insofar as it is understood, involves temporary changes in how the brain's default mode network operates, reduced activity in the circuits associated with self-referential rumination, a window of increased neural plasticity that the therapeutic context can use. This is not a drug that fixes a broken brain. It is a compound that temporarily reorganizes how the brain processes experience, in ways that appear to facilitate therapeutic work.
The implication is that the therapeutic context matters — that what happens in the hours and days around the drug experience, the quality of the therapeutic relationship, the preparation and integration, are as important as the pharmacology. This is a return, in a paradoxical form, to something the biological revolution suppressed: the idea that human beings recover from psychological suffering partly through relationships, through meaning-making, through the slow work of understanding themselves in the company of other people.
The Structural Constraint
The tragedy of American psychiatry is not primarily intellectual. The field has the intellectual capacity to correct its errors; the process is underway. The tragedy is structural. The insurance reimbursement system, which took the DSM diagnostic framework and built a billing infrastructure around it, incentivizes fifteen-minute medication management appointments and actively disincentivizes the long-form psychotherapy that the evidence base most strongly supports for conditions including depression, anxiety, PTSD, and personality disorders.
The psychiatry training pipeline has responded to these incentives in the predictable way: the proportion of psychiatrists who provide any psychotherapy — as opposed to diagnosis and medication management — has fallen steadily for decades. A 2010 analysis in the Archives of General Psychiatry found that from 1996 to 2005, the proportion of psychiatric visits involving psychotherapy declined from 44 to 29 percent. Subsequent data suggests the trend has continued. The psychiatrist as psychotherapist has nearly disappeared; the psychiatrist as prescriber has taken their place.
What this means for patients is a system in which the comprehensive care that would address the full complexity of psychological suffering is, for most people, simply not available. The person in acute distress sees a prescriber for twenty minutes and leaves with a prescription. The person who needs to understand why she keeps choosing relationships that harm her, or why she cannot sustain attention, or why she wakes at 3 a.m. every night with a specific flavor of dread — that person needs something that the American healthcare system is organized, structurally and financially, not to provide.
The drugs helped some people. That is true. They did not deliver on the promise of the Decade of the Brain. That is also true. The question now is whether psychiatry has the institutional will to acknowledge the gap between the promise and the delivery, and to rebuild toward something more adequate — more psychologically sophisticated, more relationally serious, more honest about what it knows and doesn't know.
The generation that grew up on SSRI prescriptions from family practitioners, without adequate therapy, without the structural supports that would make their lives actually livable, deserves better than a correction in the academic literature. They deserve a psychiatry that is as interested in the meaning of their suffering as in its neurochemistry — and that understands, finally, that these questions cannot be separated.