Twilight of the Psychopharmacologists
A Review of “Psychopharmacology Reconsidered”
Robert Haim Belmaker and Pesach Lichtenberg, Psychopharmacology Reconsidered: A Concise Guide Exploring the Limits of Diagnosis and Treatment (Springer, 2023).
There is a particular kind of book that can only be written late in one’s career or at a critical juncture in one’s profession. It is the book in which a scientist of established reputation takes stock of the promises made in the field and their own aspirations at the beginning of their work and asks, with as much honesty as can be managed, which of those promises were kept. Psychopharmacology Reconsidered is such a book. Robert Haim Belmaker, former president of the International College of Neuropsychopharmacology, has a fifty-year career spanning laboratory neuroscience and clinical psychiatry. With co-author Pesach Lichtenberg, also an accomplished psychiatrist, he has produced a textbook (part clinical manual, part confessional) of psychopharmacology organized around the collapse of what can be characterized as diagnosis-centered psychopharmacology (not a term they use).
The book covers the major medication classes along with discussions of the biochemical basis of psychopharmacology, DSM diagnosis, clinical trial methodology, and future directions. A chapter by psychiatry resident Alexander Moshe Clayman provides a trainee’s perspective (respectful but not obsequious). The tone throughout is conversational and opinionated, and the volume functions as a guided tour by two clinician-scientists who share their hard-won conclusions of how to think about psychiatric medications.
The introduction is structured as two sequential personal narratives, first by Belmaker and then by Lichtenberg, each tracing a trajectory of progressive disillusionment with what they see as psychopharmacology’s foundational promises.
Belmaker organizes his account across multiple domains of disappointment. On genetics: the confidence of the 1970s that molecular genetics would yield discrete disease genes and rational drug targets has given way to the reality of hundreds of common variants of small effect, mimicking the architecture of traits like height and weight. On lithium: despite decades of mechanistic research in which Belmaker himself was deeply invested, all purported mechanisms of lithium action proved speculative, and no rational lithium-like alternative has ever emerged. Lithium works in many but not all cases of bipolar disorder, and so do carbamazepine, valproate, and second-generation antipsychotics… pharmacologically disparate compounds that network meta-analyses show to be equally effective (see a detailed discussion of lithium by Belmaker in a guest post for Psychiatry at the Margins). On antipsychotics: Snyder’s dopamine hypothesis seemed to validate the disease model of schizophrenia, but dopamine blockers act primarily on positive symptoms of psychosis, work across many diagnoses, and affect normal thought and motivation. On antidepressants: SSRIs were developed on the premise that serotonergic specificity would yield greater efficacy, but they are not more efficacious than imipramine and other TCAs; the diagnosis of depression has expanded so dramatically that placebo-drug differences have collapsed to near zero; the chemical imbalance narrative remains on scientifically shaky grounds. On biological markers: limited progress in translating transdiagnostic biological associations into clinical relevance. On clinical trials: reliability was prioritized over validity; meta-analyses contradicted each other; the commercial trial apparatus became self-serving.
Lichtenberg’s portion narrates a parallel arc. Having entered psychiatry during the ascendancy of biological psychiatry, he describes a growing conviction that the standard neurochemical tools of psychopharmacology were inadequate and the minds of his psychotic patients were “too fascinating and complex to be reduced to the blockade of dopamine receptors.” He invokes Moncrieff’s distinction between a disease-centered model of drug action (the drug corrects an etiological cause, like an antibiotic) and a drug-centered model (the drug produces psychoactive effects that provide symptomatic relief, like a shot of whiskey). Lichtenberg is dissatisfied with the neuro-reductionism prevalent in the field and believes that psychiatric diagnosis is fundamentally personal and behavioral (rather than brain-based) and that the meaning of a patient’s symptoms can only be found in the psychological realm. He comes to see that there is “a role for medication, and occasionally it is crucial,” but ultimately many patients require care that goes beyond medications and brain-based interventions. Lichtenberg’s trajectory culminates in his founding of the first Soteria home in Israel, a supportive, dialogical environment where medication is one component among many.
“If the result of this textbook will be to nudge clinical psychiatry to accept the limitations of psychopharmacological solutions for the complex problems of extreme emotional distress, and to reinvigorate the search for other means of providing succor for our patients, we will have accomplished our purpose.” (page 10)
It is an engaging and curious opening. The book is inspired, in part, by work in critical psychiatry (Whitaker and Moncrieff are both cited). I was in a similar kind of state at the very beginning of my career, circa 2018-2019, and reading this book, I was strongly reminded of my own sense of disillusionment back then. Resultantly, I was interested to see where Belmaker and Lichtenberg would go from there. In my own case, once I had come to terms with critical psychiatry, I sought to understand the new scientific and philosophical landscape of dimensionality, complex systems, pluralism, systems neuroscience, and Mad studies, and how it applies to psychopharmacology. Belmaker and Lichtenberg’s impulse, in many instances, is to retreat to the clinical past, to an older and simpler pharmacological era when imipramine treated melancholia, benzodiazepines were the norm for anxiety, and stimulants were reserved for children with minimal brain dysfunction.
Reading across the full book, a coherent, if not always explicit, framework emerges, which can be articulated as a set of general principles. (Belmaker and Lichtenberg themselves do not present any such list; I offer it here as a form of summary.)
Non-specificity of psychiatric medications. No existing psychiatric medication is specific to any DSM diagnosis. Antipsychotics work across schizophrenia, mania, psychotic depression, severe anxiety, and agitation. Antidepressants work in depression, panic disorder, OCD, and anxiety. Lithium works in bipolar disorder, schizoaffective disorder, augments antidepressant treatment, and helps in episodic aggression. The penicillin analogy (a drug that specifically targets a defined pathological process) is fundamentally misleading for psychotropic drugs. These are compounds that alter basic neurochemical systems involved in mood, arousal, salience, and reward, and their effects cascade (rather unpredictably) across multiple domains.
Drug-centered rather than disease-centered drug action. Drugs produce psychoactive effects that may provide symptomatic relief, rather than correcting biochemical abnormalities. No psychiatric disorder has a confirmed biochemical etiology that the relevant drug class corrects. While Moncrieff is approvingly cited in the first chapter, the disease-centered vs. drug-centered distinction isn’t really examined in the book, and like many other commentators, I think Belmaker and Lichtenberg work with a rather superficial impression of the assumptions guiding Moncrieff’s distinction and its implications.
Diagnostic non-validity undermines pharmacological algorithms. DSM diagnoses are not biologically valid. Many patients presenting for care don’t meet specific DSM criteria. Treatment should be guided by symptoms and syndromes (psychosis, insomnia, panic, melancholia) rather than diagnoses.
Diagnostic expansion dilutes drug efficacy. As diagnostic categories broadened from DSM-III through DSM-5, prevalence increased, but placebo-drug differences in clinical trials have narrowed. The drugs that worked well in narrowly defined populations appear far less effective when applied to the expanded diagnostic pools.
The clinical trial enterprise has become degraded. Head-to-head trials almost never find differences within drug classes; the volunteer pool is contaminated; academic trialists rarely see patients; the FDA allows publication of only positive trials; meta-analyses contradict each other; guidelines bias toward newer drugs; findings based on group means don’t meaningfully allow personalization of treatment.
All psychotropic drugs carry subjective costs that tend to go uncatalogued. Antipsychotics reduce pleasure and motivation; antidepressants may blunt emotional life. Beyond pharmacology, the medical model encourages passivity, and biological framing deepens some forms of stigma. These costs must be weighed against benefits for each patient.
Psychopharmacology is one component within a broader biopsychosocial approach. Medication should often not be the first-line treatment, especially where distress is psychosocially mediated or where the patient’s problem is unresponsive to pharmacology.
The totality of “evidence” should be considered rather than a narrow emphasis on RCTs. The idealized hierarchy from basic science to RCTs to meta-analyses to clinical practice is misleading; they propose instead a four-pillar model where basic science, epidemiology, clinical trials, and individualized clinical interpretation all independently support clinical decision-making in a non-hierarchical and continuously evolving way.
The book’s clinical chapters apply these core principles (somewhat unevenly) across drug classes. Antidepressants illustrate the non-specificity and diagnostic-expansion theses most fully: early imipramine studies in narrowly defined melancholia showed robust effects, but as “major depression” expanded to encompass most human sadness, placebo-drug differences collapsed. B&L believe antidepressants remain effective for melancholic depression with vegetative features, classic panic disorder, and relapse prevention, but are likely no better than placebo for the broader DSM-5 population. Antipsychotics similarly demonstrate non-specificity: their efficacy across mania, psychotic depression, agitation, and anxiety marks them as symptomatic dopamine receptor blockers rather than disease-targeted therapies, and second-generation agents proved no superior to first-generation ones per CATIE. They pose pointed questions about the standard treatment model: placebo responders shouldn’t be exposed to side effects, non-responders shouldn’t be maintained on ineffective drugs, and a schizophrenia diagnosis shouldn’t automatically mandate lifetime medication. Benzodiazepines, surprisingly, receive the most favorable treatment; B&L argue that hesitation around prescribing has led to undertreated anxiety and recommend benzodiazepines over antidepressants for many anxiety conditions, reserving antidepressants for well-defined panic disorder.
Their discussion of stimulants exemplifies both the strengths and the limits of the book. Belmaker and Lichtenberg open that chapter with the tension that the same stimulant drugs prescribed to millions of children for focus have a well-documented history as addictive substances and psychosis-inducing agents in adults. They note that the two literatures have developed in near-total isolation from each other, with therapeutic papers on childhood ADHD rarely citing addiction research and vice versa. Belmaker and Lichtenberg acknowledge that stimulants produce clinically significant reductions in hyperactivity and can meaningfully help children remain in their school settings, but they view both the ADHD diagnosis and treatment with stimulants with broad suspicion. They raise the concern that college students and adults who seek stimulants are often responding to the drugs’ euphoriant properties rather than treating a genuine deficit, and warn that adolescents continuing stimulant treatment may develop dependency patterns indistinguishable from adult amphetamine addiction. The chapter closes with the “unresolved” paradox that dopamine-enhancing stimulants somehow help rather than worsen hyperactivity.
The book was published in 2023, and given the timeline of academic book publication, the draft may have been finalized ~2021-2022. I am highlighting this because we cannot blame B&L for not being aware of research that came out in subsequent years and also because anyone reading the book today needs to be aware that important developments from recent years are missing.
Here are some conceptual, clinical, and scientific points of critique.
The book treats “non-specificity” as a singular phenomenon, but it encompasses several conceptually distinct claims that require different analysis. Pharmacological promiscuity (a drug acts on multiple receptor systems) is a fact about molecular pharmacology. Transdiagnostic efficacy (dopamine blockers work in schizophrenia, mania, and psychotic depression) could mean the drug targets a transdiagnostic dimension, or that diagnostic categories carve nature incorrectly, or both. Mechanistic convergence from pharmacological divergence (lithium, valproate, and olanzapine all help in bipolar disorder) could mean the diagnosis is heterogeneous and each drug works on a different subgroup, or that multiple routes lead to the same downstream effect. B&L slide between these senses in ways that sometimes weaken their argument.
The disease-centered vs. drug-centered dichotomy doesn’t say what most readers think it says. Moncrieff’s distinction is the philosophical scaffold the book nods towards in the beginning and it stays in the background. The distinction is widely thought to assert that either a drug corrects a specific pathological process (like an antibiotic) or it merely produces psychoactive effects that mask symptoms (like alcohol for anxiety). On this casual reading, the framework simply says that psychiatric drugs are symptomatic treatments rather than disease-modifying ones, an interpretation that, for many readers, makes the binary seem reasonable. But this is not what Moncrieff actually claims. As I have discussed elsewhere, her framework makes quite specific commitments that most of its sympathetic readers do not realize they are taking on. Moncrieff classifies symptomatic pain medications like acetaminophen and ibuprofen as disease-centered, because they act on the physiological pathways that are involved in pain. Symptomatic medications targeting fever, cough, edema, blood pressure, etc., all are disease-centered in her usage because they act on physiological processes involved in symptoms even when they do not address primary etiological causes. The “drug-centered model” is reserved for a narrower and more specific claim: that psychiatric drugs work like alcohol or opiates, through psychoactive effects (sedation, cognitive slowing, euphoria, emotional blunting) that suppress or distract from symptoms without acting on the physiological mechanisms that produce them.
This conflation of disease-modifying and symptomatic treatments under “disease-centered” injects intentional confusion into discussions of psychopharmacological mechanisms. It severely restricts which mechanisms one is allowed to invoke for psychiatric drugs: hypotheses involving prediction-error attenuation, neuroplasticity, cognitive flexibility, neurogenesis, or inflammatory pathways are all dismissed as “disease-centered” and therefore disallowed. What remains are intoxication-like mechanisms: sedation, blunting, suppression. The framework also enables a motte-and-bailey: the defensible motte (“the medication doesn’t correct a confirmed dysfunction”) is defended while the contentious bailey (“the medication works only by numbing or masking your symptoms”) is the position actually being asserted. There are better ways to think about this terrain but Belmaker and Lichtenberg unfortunately endorse the disease-centered/drug-centered distinction without examining what it actually entails, and the book inherits its problems.
The philosophy of mind is underdeveloped. Lichtenberg articulates a non-reductive physicalism: mind derives from brain, but subjective experience belongs to persons, and meaning can only be found in the realm of the mind. This is defensible, but its implications for psychopharmacology are never developed systematically. The drugs can work in part through biochemical mechanisms, but the gap between synaptic physiology and phenomenological experience cannot be crossed by a single level of explanation. This points toward something like explanatory pluralism: the view that psychiatric phenomena require multiple irreducible levels of explanation (biochemical, psychological, social, phenomenological). B&L come close to this position in spirit but never develop it philosophically. (See my thoughts on explanatory pluralism and psychopharmacology.)
Here the theoretical developments in levels of biological organization/explanation, complex systems, and embodied cognition/enactivism would have given B&L richer conceptual resources. The enactivist tradition in particular offers a principled account of why the organism–environment relationship cannot be reduced to neurotransmitter levels. And why pharmacological intervention, while genuinely affecting neural dynamics, can never fully substitute for the relational, embodied, and situated dimensions of mental life.
The benzodiazepine position overcorrects. I am sympathetic to the judicious use of benzodiazepines, and I am more open to using benzodiazepines than some of my colleagues, but it’s still a stretch to advocate for benzodiazepines as being the first-line treatments for generalized anxiety today. B&L substantially understate the clinical problems that surround the use of benzodiazepines. Their clinical vignettes depict idealized time-limited use that bears little resemblance to complicated real-world presentations where time-limited use is often aspired to but infrequently achieved.
The melancholia thesis is on shaky grounds. The claim that antidepressants work well for narrowly defined melancholia but poorly for the broader population is clinically intuitive (especially for an older generation of psychiatrists, I’ve noticed) but poorly supported by empirical evidence. The clearest finding from the literature is not that antidepressants are specifically efficacious in melancholia but that tricyclic antidepressants outperform serotonin reuptake inhibitors on core melancholic symptoms… an advantage that, on the available meta-analytic evidence, appears to extend to non-melancholic depression as well and so does not establish melancholia as a pharmacologically privileged target. The Undurraga et al. (2020) meta-analysis found near-identical antidepressant response rates in melancholic (39.4%) and non-melancholic (42.2%) depression, with the lower placebo response in melancholic patients doing much of the work in apparent drug-placebo separations. Subjects responded better to TCAs (50.6%) than SRIs (30.0%). Imai and colleagues’ (2021) individual-patient-data meta-analysis showed that melancholic features were prognostic of overall symptom reduction but did not moderate the drug-placebo difference.
I just don’t think it’s true that melancholia is the population in which antidepressants genuinely work while the rest is diagnostic dilution. (See here for my discussion of antidepressant efficacy and mechanisms).
The stimulant skepticism is selectively sourced. Belmaker and Lichtenberg don’t engage seriously with the substantial evidence base for stimulant efficacy in carefully diagnosed populations. Their clinical vignettes are chosen to illustrate the problem cases rather than the straightforward ones. They apply their diagnostic-expansion-dilutes-efficacy argument to antidepressants with more nuance than they apply it to stimulants, where the tone becomes broadly dismissive. Research on brain circuits has also begun to dissolve the central paradox that stumped B&L: stimulants act on arousal, salience, and reward networks rather than on attention networks proper, helping with sustained engagement on unrewarding tasks rather than amplifying selective attention. This reframes stimulant action in a way that dispenses with the supposition of paradoxical pediatric calming and clarifies why both inattention and motoric restlessness in ADHD are best understood as motivational rather than purely attentional problems.
There is, strangely, no discussion of deprescribing. In fact, the term doesn’t even appear in the book. For a book adjacent to critical psychiatry, I was expecting at least some discussion of it.
The framework doesn’t engage with the new sciences of psychopathology. A significant scientific omission is the absence of any meaningful engagement with dimensional, network, clinical staging, and biotype approaches to psychopathology. These frameworks offer a principled response to many of the problems B&L identify. The HiTOP structure, for instance, provides a framework within which the transdiagnostic efficacy of dopamine blockers (effective across the Thought Disorder spectrum) or SSRIs (effective across the Internalizing spectrum) becomes not a failure of specificity but a positive finding about the spectral structure of psychopathology. B&L’s clinical observations are largely compatible with dimensional models, but they don’t seem aware of or interested in this literature. Computational psychiatry and predictive processing approaches are also absent. The network approach in which mental disorders are constituted by causal interactions among symptoms rather than by underlying latent diseases also provides hypothetical explanations for diagnostic non-specificity. If depression is a self-reinforcing network of interacting symptoms, then a drug that intervenes at any node could disrupt the network without being “specific” to the “disease.”
The relational dimension of prescribing is ignored. B&L never examine the prescribing relationship itself as a site of therapeutic pharmacological action or therapeutic failure. Psychodynamic psychopharmacology is entirely absent. If the prescribing psychiatrist contributes more to outcome variability than the pill (a finding they themselves cite in their placebo chapter), then the meanings patients attach to medications, the transference dynamics that shape adherence and response, and the countertherapeutic uses of prescribing that can chronify illness are relevant concerns to psychopharmacology. They catalogue the failures of the pharmacological treatment without really examining how the relational context in which drugs are prescribed and consumed might account for some of those failures. (See my interview with David Mintz on psychodynamic psychopharmacology.)
What I’m trying to say is that Belmaker and Lichtenberg’s diagnosis of the field’s predicament can be accepted, and yet the inferences can be taken in a quite different direction. Their response is, in essence, a chastened and humble clinical humanism-and-pragmatism: prescribe less, prescribe more narrowly, return to the prescribing sensibility of psychopharmacology pioneers, attend to the patient in front of you, resist the totalizing claims of either pharmaceutical marketing or anti-psychiatric critique. Psychopharmacology Reconsidered does not really say what a positive framework of understanding the transdiagnostic effects of psychiatric medications would or should look like. My own view is that the failures of a diagnosis-centric and disease-centric view should drive us toward richer multi-level mechanisms situated within the new clinical and scientific landscape of dimensions, networks, computations, biotypes, enactivism, and phenomenology.
Psychopharmacology Reconsidered is best understood as belonging to a clinical wisdom tradition, an accumulation of hard-won practical insights from decades of prescribing and scholarly work. Every psychopharmacology textbook is aware of the transdiagnostic effects of psychiatric medications, but few address the relevant tensions head on. B&L are refreshing in their willingness to do so. Reading this book is like talking to and learning from two experienced colleagues.
The book occupies a transitional space: it documents the exhaustion of one pharmacological framework and the necessity of continuing on. It is a book written at dusk, unsure of what the night will bring. It is a book written amidst the ruins of diagnosis-centered psychopharmacology, looking back at the early days of syndromic psychopharmacology, but not yet oriented toward the emerging outlines of a post-DSM, post-critical scientific psychopharmacology.
See also:
