Turning down the volume

Seventy-odd years of antipsychotics: the surgeon's cocktail that emptied the asylums, the ads and the movie that shaped how we see them, how each generation actually works (read through their receptor blooms), what the big trials found, what they're used for now, and the first genuinely new mechanisms in decades.

By Reid Robison, MD. 4,837 words, with a timeline, a dopamine occupancy dial, the antipsychotic blooms side by side, the trials as pictures, a circuit you can switch on and off, and the pipeline scoreboard as of October 2026.

I started my career running clinical trials, and a surprising number of them in those early years tested an atypical antipsychotic added to an antidepressant for depression that hadn't fully responded. That's a strange sentence if you think of these drugs only as "medicines for schizophrenia," and it's a good place to start, because the class has always been stranger, more useful and more complicated than its name. This post is the whole arc: where they came from, what they replaced, how they show up in movies and old ads, how each generation works at the receptor level, what the best evidence says about outcomes and side effects, and what's coming next. It's meant for prescribers and for anyone who has ever opened a pharmacy bag and wondered what they were just handed.

A note on language before we start. Psychosis is a set of experiences, and schizophrenia is one illness among several where it happens. People living with these conditions are people first, and much of the history below is the history of how badly that was forgotten. I'll try to tell it honestly without making anyone a punchline.

1Before the pill

In 1955, state and county mental hospitals in the United States held about 559,000 people.¹ Many stayed for years, some for life. The treatments available to them in the 1930s and 40s were the "somatic therapies," and it is worth understanding them, both because they were desperate attempts to help people with no better options and because they explain why an oral pill felt like a miracle.

In 1933 the Viennese physician Manfred Sakel began inducing deliberate comas with insulin, on the theory that hypoglycemic shock could reset the psychotic brain; insulin coma units spread through Europe and the United States for two decades.⁶ In 1934 Ladislas Meduna induced seizures with camphor and then metrazol, reasoning, wrongly, that epilepsy and schizophrenia could not coexist. In 1938 Ugo Cerletti and Lucio Bini in Rome replaced the drug with electricity, and electroconvulsive therapy was born; it survives today, much changed, as one of the most effective treatments in psychiatry for severe depression and catatonia.⁷ And in 1935 the Portuguese neurologist Egas Moniz performed the first prefrontal leucotomies. He won the Nobel Prize for it in 1949, and in the United States Walter Freeman's "ice-pick" transorbital lobotomy was performed on thousands of people before the practice collapsed.⁷ Insulin coma and lobotomy are now regarded as among the darkest chapters in medicine. What replaced them was a molecule nobody had designed for the mind at all.

2An accident in the operating room

In December 1950 the chemist Paul Charpentier at the French company Rhône-Poulenc added a chlorine atom to an antihistamine, promethazine, and sent the new compound, RP 4560, to the pharmacologist Simone Courvoisier.¹ About six months later a navy surgeon named Henri Laborit began using it in his "lytic cocktail," a mixture meant to blunt the body's stress response to surgery. His February 1952 report described something odd: patients given the drug outside of surgery became calm and "disinterested," awake but strangely unbothered.¹ That word caught the attention of psychiatrists.

Later in 1952 Jean Delay and Pierre Deniker at the Sainte-Anne hospital in Paris gave chlorpromazine alone, in repeated doses, to agitated and psychotic patients, and reported that it calmed psychosis itself, not just agitation.¹⁸ Rhône-Poulenc marketed it in France as Largactil in December 1952. The FDA approved it in 1954, for psychiatric use and for nausea and vomiting, and Smith, Kline and French sold it in the United States as Thorazine. Within eight months, by the company's count, it had been given to more than two million people.¹

The hospital census began to fall the same year. From 559,000 in 1955 it dropped to about 452,000 a decade later, and kept falling for decades.¹ Chlorpromazine was one cause among several: the new federal funding rules of the 1960s, the community mental health movement, and the slow realization that warehousing people was cruel all pushed in the same direction. The honest coda is that many people left hospitals for communities that never built the support they needed, and today too many people with serious mental illness live in jails, prisons or on the street. A medicine opened the door. Society did not finish the job.

Haloperidol followed in 1958, synthesized by Paul Janssen in Belgium, far more potent per milligram and less sedating, and for the next thirty years "the typicals," the first-generation antipsychotics, were the treatment.⁸

3Seventy years in one line

Here's the whole story at a glance. Tap any point, or step through with the buttons.

4The ads and the movie

Two kinds of pictures shaped how the public came to see these drugs, and both are worth a careful look.

The first is the advertising. Mid-century drug ads in psychiatric journals were aimed at doctors and hospital administrators, and they said a great deal about who was being treated and why. The psychiatrist and historian Jonathan Metzl traced how the imagery changed. A 1955 Thorazine ad showed calm wards of white women. By the 1970s, ads in the Archives of General Psychiatry looked very different: a 1974 Haldol ad showed an angry Black man raising a fist on a city street under the line "Assaultive and belligerent?"⁹ Metzl ties that shift to the era's anxieties about race and protest, and to a documented pattern that persists in research: Black men in the United States have been more likely to be diagnosed with schizophrenia and to receive higher antipsychotic doses than white patients with similar symptoms.⁹ The ads are worth seeing not as curiosities but as a warning about how easily a medicine for suffering can be sold as a medicine for control.

The second is One Flew Over the Cuckoo's Nest. Ken Kesey's 1962 novel became Miloš Forman's 1975 film, shot largely on empty wards of the Oregon State Hospital in Salem, where the superintendent, Dean Brooks, played the hospital administrator and 89 patients were hired as production assistants and extras. It won all five major Academy Awards.¹⁰ The film is mostly about power: Nurse Ratched's daily "medication time," the electroshock used as punishment, and the lobotomy at the end. It is a great film, and it left millions of people with a picture of psychiatric treatment as something done to people rather than with them. Psychiatry earned some of that, and the field changed in part because of it. But the picture lasts longer than the practices it showed, and I still meet patients who are afraid of medicines, and of ECT, because of a movie made half a century ago.

5Why they work: dopamine and the window

For two decades nobody knew how chlorpromazine worked. The answer came in steps. In the early 1960s the Swedish pharmacologist Arvid Carlsson showed that antipsychotics changed dopamine metabolism, work that contributed to his Nobel Prize in 2000. Then, in 1975 and 1976, two groups, Philip Seeman's in Toronto and Ian Creese, David Burt and Solomon Snyder's at Johns Hopkins, measured how tightly dozens of antipsychotics bound to what we now call the D2 dopamine receptor. The fit was striking: the more tightly a drug bound D2, the lower the dose that worked in patients.¹¹¹² That correlation became the dopamine hypothesis of schizophrenia, and it still holds for every antipsychotic approved before 2024.

Brain imaging later put numbers on it. In first-episode patients treated with haloperidol, Shitij Kapur and colleagues found that the chance of clinical response rose sharply once about 65 percent of D2 receptors were occupied, prolactin elevation became likely above about 72 percent, and movement side effects above about 78 percent.¹³ That leaves a narrow therapeutic window, and it explains why "more is better" was never true of these drugs. Try it.

Second-generation drugs complicated the picture without overturning it. Many add strong blockade of the serotonin 5-HT2A receptor, which may soften movement side effects at a given D2 occupancy. Clozapine and quetiapine let go of D2 quickly, which may matter too. And the partial agonists, starting with aripiprazole in 2002, don't block D2 so much as turn it down to a lower setting, which changes the side-effect profile again. The cleanest way to see all of this is to look at the drugs themselves.

6Reading the antipsychotic blooms

On this site, every psychiatric drug has a receptor bloom: one petal for each receptor or transporter it touches, colored by neurotransmitter system, with length showing how strongly it binds. Solid petals turn a receptor on, hollow ones block it, and half-filled petals are partial agonists. (If blooms are new to you, here's the field guide.) Antipsychotics make the clearest case for why the blooms are useful, because their side effects are written right on the petals: a long purple histamine petal predicts sedation and weight gain, a brown muscarinic petal predicts dry mouth and constipation, an orange alpha-1 petal predicts dizziness on standing.

Look at chlorpromazine and haloperidol side by side and you can see why the two drugs feel so different. Chlorpromazine is a bushy flower of histamine, muscarinic and alpha-1 petals around a moderate dopamine petal: sedating, low blood pressure, dry mouth. Haloperidol is almost a single magenta spike: little sedation, lots of stiffness and restlessness. Olanzapine and clozapine bring the bush back, which is much of why they cause weight gain. Aripiprazole's half-filled dopamine petal is the partial agonist, which is why it raises prolactin less and causes akathisia, an inner restlessness, more often than sedation. And Cobenfy has no dopamine petal at all.

7Clozapine, the exception

Clozapine was synthesized in the late 1950s as part of a series of tricyclic compounds, initially dismissed as ineffective, and then shown in the 1960s to treat psychosis without the disabling movement side effects of the typicals.⁸ It reached European markets in the early 1970s. Then, in 1975, The Lancet reported 18 patients in Finland with severe blood disorders on clozapine, 16 with agranulocytosis, a collapse of the white blood cells that fight infection, and 9 deaths. The drug was withdrawn across much of Europe.⁸

It came back because of one trial. John Kane and colleagues took 268 people with schizophrenia who had not responded to at least three antipsychotics, confirmed nonresponse with a six-week trial of haloperidol, and randomized them to clozapine or chlorpromazine. After six weeks, 30 percent of the clozapine group had improved by prespecified criteria, against 4 percent on chlorpromazine.¹⁴ The FDA approved clozapine in 1989 with mandatory blood monitoring.

Clozapine remains the only drug with clear evidence for treatment-resistant schizophrenia. In the 2019 network meta-analysis of 32 antipsychotics, it had the largest effect on overall symptoms (a standardized mean difference of -0.89 against placebo).³ In a Finnish national cohort of about 67,000 people with schizophrenia followed from 1996 to 2006, clozapine users had the lowest mortality of any antipsychotic group.¹⁵ And yet it is badly underused, largely because of the monitoring burden. In June 2025 the FDA eliminated the clozapine REMS program, concluding that the registry was no longer necessary and could impede access, though it still recommends regular blood count monitoring.¹⁶ If you or someone you love has tried two antipsychotics for schizophrenia or a related psychotic illness without enough benefit, clozapine is a question worth asking about by name. (That rule of thumb is about psychosis, not depression or other conditions where these drugs are also used.)

8What the big trials found

The second-generation drugs arrived in the 1990s on a wave of optimism, and for a while they were marketed as better at everything. The evidence since has been more humbling, and more useful.

They work. In a 2012 meta-analysis of 65 randomized trials in people whose schizophrenia had stabilized, those who continued an antipsychotic relapsed within a year 27 percent of the time, against 64 percent of those switched to placebo: three people need to stay on medication to prevent one relapse. Hospital readmission fell from 26 to 10 percent, and long-acting injections did somewhat better than pills.⁴

They're hard to stay on. The NIMH-funded CATIE trial randomized 1,493 people with chronic schizophrenia at 57 US sites to olanzapine, perphenazine (a mid-potency first-generation drug), quetiapine, risperidone or ziprasidone. By 18 months, 74 percent had stopped their assigned medication, because it wasn't working well enough, because of side effects, or for other reasons. Olanzapine lasted longest, with 64 percent stopping, at the cost of more weight gain; perphenazine, the old drug, held up about as well as most of the new ones.⁵

They differ more in side effects than efficacy. The largest network meta-analysis to date, 402 trials with more than 53,000 participants and 32 drugs, found that apart from clozapine, differences in efficacy between drugs were modest, while differences in side effects were large: weight change ranged from essentially none for ziprasidone to more than 3 kilograms over a short trial for zotepine; prolactin from a large drop with clozapine to a rise of about 49 ng/mL with paliperidone.³ The practical lesson is that choosing an antipsychotic is mostly choosing a side-effect profile that fits a particular person's body and life.

9The costs: movement, metabolism, and age

Movement. Blocking dopamine in the motor circuits of the striatum can cause stiffness, tremor and restlessness in the short term, and tardive dyskinesia, involuntary movements of the mouth, face or limbs, over the long term. In a 2017 meta-analysis of 41 studies, the prevalence of tardive dyskinesia among people on antipsychotics was 25.3 percent overall, 30 percent with first-generation drugs and 20.7 percent with second-generation drugs; among people who had only ever taken second-generation drugs, it was 7.2 percent.¹⁷ Pooled annual incidence runs about 3.9 percent on second-generation drugs and 5.5 percent on first-generation ones.¹⁷ Since 2017, two VMAT2 inhibitors, valbenazine and deutetrabenazine, have been approved to treat tardive dyskinesia, which changed it from a permanent sentence to a treatable condition for many people.

Metabolism. Weight gain, rising blood sugar and lipids are the signature costs of several second-generation drugs, especially olanzapine and clozapine. They matter because people with schizophrenia already die, on average, many years earlier than the general population, mostly of heart disease. Monitoring weight, glucose and lipids, choosing lower-risk drugs when possible, and pairing treatment with metformin or other strategies when weight climbs are standard of care. One newer option, olanzapine combined with samidorphan, was approved in 2021 to blunt olanzapine's weight gain.

Age. In April 2005 the FDA reviewed 17 placebo-controlled trials, about 10 weeks long, of second-generation antipsychotics in 5,106 older adults with dementia-related behavior problems. Death rates were about 4.5 percent on the drugs and 2.6 percent on placebo, roughly a 1.6- to 1.7-fold increase, mostly from heart failure, sudden death or pneumonia. Every antipsychotic now carries a boxed warning about it.¹⁸ That doesn't mean they're never used in dementia; it means they're used sparingly, briefly, and only when the distress or danger outweighs the risk.

10What they're used for now

The name "antipsychotic" undersells the class. In the United States today these drugs are approved for:

• Schizophrenia and schizoaffective disorder, the original indication, both for acute episodes and to prevent relapse, including long-acting injections given every few weeks to months.
• Bipolar disorder: acute mania, maintenance, and, for a few drugs (quetiapine, lurasidone, cariprazine, lumateperone and the olanzapine-fluoxetine combination), bipolar depression.
• Major depression, added to an antidepressant: aripiprazole (2007), quetiapine extended release (2009), brexpiprazole (2015), cariprazine (December 2022) and lumateperone (November 2025), plus the olanzapine-fluoxetine combination for treatment-resistant depression.¹⁹²⁰
• Irritability in autism (risperidone and aripiprazole), Tourette's syndrome, and psychosis in Parkinson's disease (pimavanserin, approved in 2016, which has no dopamine petal at all).

The depression indication is the one I know best from the inside, and it's worth being precise about. A 2009 meta-analysis pooled 16 placebo-controlled trials with 3,480 patients whose depression had not fully responded to an antidepressant. Adding an atypical antipsychotic raised response from 29.9 to 44.2 percent and remission from 17.2 to 30.7 percent: one extra person helped for every nine treated. But 9.1 percent stopped because of side effects, against 2.3 percent on placebo, one extra dropout for every seventeen treated.²¹ That's a real benefit with a real price, and it's why I talk through weight, restlessness and sedation before adding one, and why I check in early.

Plenty of use is off-label, and some of it worries me, especially low-dose quetiapine prescribed for sleep, where the metabolic and movement risks are easy to underestimate for a drug taken nightly for years.

11Thinking in circuits

The dopamine hypothesis has grown up. The modern version, articulated by Oliver Howes and Shitij Kapur, places the problem upstream: PET studies show that people with psychosis tend to synthesize and release more dopamine in the striatum, especially in its associative part, so the excess is presynaptic, and D2 blockers act downstream of it.²² Work in animal models by Daniel Lodge and Anthony Grace pushed further back still, to an overactive hippocampus that drives dopamine neurons in the midbrain to fire too readily.²³ Seen this way, a D2 blocker is like turning down a speaker that's being fed too loud a signal. The newer drugs try to turn down the signal itself.

That is where the muscarinic drugs come in. Xanomeline began at Eli Lilly in the 1990s as a memory drug for Alzheimer's disease; in a 1997 trial it unexpectedly reduced hallucinations and delusions, but its stomach side effects kept it off the market.²⁴ Karuna Therapeutics paired it with trospium, an older bladder drug that blocks muscarinic receptors in the body but barely enters the brain. Xanomeline stimulates M1 and M4 receptors in the brain, and M4 receptors in the striatum and midbrain act as a brake on dopamine release, while trospium soaks up the side effects elsewhere. Explore the circuit.

12The pipeline, October 2026

The muscarinic approval broke a thirty-year streak of dopamine-only drugs, and a wave of candidates followed. The honest scorecard is mixed: several promising mechanisms have failed in larger trials, often with high placebo responses, which is itself a lesson about how hard schizophrenia trials have become. Here's where things stand. Company toplines are noted as such; they are not peer-reviewed until published.^{22526202728293031323334}

A few threads are worth watching. The muscarinic idea is clearly real in acute schizophrenia, but adding Cobenfy to an existing antipsychotic missed its goal in the ARISE trial, and AbbVie's M4 modulator emraclidine failed both of its phase 2 trials, so how best to use the mechanism is still being worked out.²⁵³¹ Treating the cognitive problems of schizophrenia, which matter enormously for work and independence, remains unsolved after the failure of iclepertin.³³ And some of the most practical progress is quieter: long-acting injections that last longer and hurt less, and older molecules re-engineered, like LB-102, a modified version of amisulpride, a drug widely used in Europe but never approved in the United States.²⁹

13How they hold up

Line the evidence up and a fair summary looks like this. Antipsychotics work: they shorten psychotic episodes, they prevent relapse, and in the largest real-world cohorts people who take them live longer than people with schizophrenia who don't. They don't cure, and for a minority of people they don't help enough, which is where clozapine still has no equal. They cost something, in movement, metabolism and, in older adults with dementia, survival, and those costs differ so much between drugs that the choice of drug should be a shared, individual decision. The second generation was not the across-the-board leap it was sold as; it was a trade of one set of side effects for another. The third, the partial agonists, made that trade more flexible. And the first truly new mechanism in decades is now in pharmacies, with more trying to follow.

If you're a patient: ask what a medicine is for, which side effects to watch, and what the plan is if it doesn't work well enough. If you're a prescriber: treat the blooms as a conversation, not a lookup table. The person in front of you is the one who will notice the petals.

References