Dorsal raphe nucleus

In brief

What it is, and what is actually known about it in humans

The raphe nuclei are a chain of cell groups strung along the midline of the brainstem over its entire length, from the midbrain down to the junction with the spinal cord ¹. Raphe is the Greek word for a seam, and in cross-section that is what the midline of the brainstem looks like. The dorsal raphe nucleus is the largest of them. It lies in and beneath the periaqueductal gray — the sleeve of gray matter wrapped around the cerebral aqueduct, the narrow channel carrying cerebrospinal fluid through the midbrain — and extends downward into the upper pons ^1,2.

Human measurements are scarce, because they need whole brains cut in series soon after death. The reference study took four adult brains, mean age 56 years and mean postmortem delay three hours, cut the midbrain and pontine central gray into 50-micrometer sections — a micrometer is a thousandth of a millimeter — and stained them by the Nissl method, which marks every neuron rather than only the serotonergic ones ². It counted 235,000 ± 15,000 neurons, about 1,200 ± 200 per section, in a volume of 71.3 ± 4.5 cubic millimeters, giving a mean density of 3,300 ± 200 neurons per cubic millimeter ². It divided the nucleus into five subnuclei — interfascicular, ventral, ventrolateral, dorsal and caudal — and found four neuron shapes, round, ovoid, fusiform and triangular, characteristic of particular subnuclei; the dorsal subnucleus was the largest and held the most neurons but had the lowest packing density, and the ventrolateral the highest ².

How many of those 235,000 neurons make serotonin is a separate question, and the human answer is not settled. A companion study by the same group used an antibody against the serotonin-synthesizing enzyme and estimated about 125,000 such neurons in the human pontine tegmentum — 64,400 in the median raphe nucleus, 30,700 in the oral pontine reticular nucleus and 29,000 in the supralemniscal group known as B9 ²⁴. Those are the neighboring pontine groups, not the dorsal raphe, and the paper offers no dorsal raphe count. It also records a caution worth carrying through this whole page: humans and other primates have far more pontine serotonin neurons than rats or cats, and the shape and spatial arrangement of those neurons differ significantly between humans and other primates ²⁴.

Almost everything else on this page comes from rodents, zebrafish or cats. That is not a complaint; it is the evidence situation. Where a finding is human, this page says so.

The raphe nuclei as a family, and the projection map

The raphe family divides cleanly in two, and the division is the single most useful fact about it. The rostral complex — the caudal linear, dorsal raphe and median raphe nuclei, in the midbrain and rostral pons — projects primarily to the forebrain. The caudal complex — raphe magnus, raphe obscurus, raphe pallidus and parts of the adjacent lateral reticular formation, in the caudal pons and medulla oblongata — projects to brainstem nuclei and, by three parallel pathways, to the dorsal, intermediate and ventral columns of the spinal cord ¹. Both clusters develop from separate primordia and keep the division into adulthood ¹.

This matters clinically more than it sounds. When a paper reports serotonergic abnormalities in the brainstem of infants who died of sudden infant death syndrome, those findings are in the medullary — caudal — raphe, concerned with breathing, blood pressure and arousal, not in the dorsal raphe ²⁹. When a paper reports an antidepressant acting on raphe firing, it means the rostral group. The two are the same transmitter in different jobs.

Within the forebrain-projecting group, the median and dorsal raphe nuclei send parallel and overlapping projections to many of the same structures, but with axons of distinct structural and functional character ¹. Their serotonergic axons arborize over large territories containing functionally diverse targets, and here the anatomy becomes unusual: while some of these terminals form classical chemical synapses, many do not ¹. Transmitter released without a synapse diffuses through the extracellular space to reach receptors at a distance, which is what the term volume transmission means. A system built this way modulates; it does not instruct.

The same review records something easy to overlook in a page about adults: serotonin shapes brain development, so a disturbance of serotonergic transmission in early life can leave lasting structural and functional changes ¹. That is one reason the question of whether a drug acting on this system during pregnancy or childhood is harmless is not a question the adult literature can answer.

The serotonin it makes, and where the antidepressants act

Serotonin is made from tryptophan, an amino acid that the body cannot synthesize and must take from food. The rate-limiting step is tryptophan hydroxylase, and in 2003 a second isoform of that enzyme was reported, which is the basis of the standard division of labor by which the brain and the gut make their serotonin with different enzymes ³⁰.

The neurons of the dorsal raphe hold serotonin receptors on their own cell bodies and dendrites. In chloral-hydrate-anesthetized rats, intravenous 5-HT1A-selective compounds — ipsapirone and LY165163 — inhibited dorsal raphe firing dose-dependently, while 5-HT1B-selective compounds were weak or irregular; applied directly to the cells by microiontophoresis, the 5-HT1A compounds produced dose-response relationships indistinguishable from serotonin itself, and in brain slices they hyperpolarized the membrane and lowered input resistance, with maximal effects exceeding serotonin's ⁵. That is the somatodendritic autoreceptor: a brake the cell holds on itself.

The brake is probably why selective serotonin reuptake inhibitors take weeks to work. Blocking reuptake raises serotonin everywhere at once, including around the cell bodies, where it engages the autoreceptor and slows the neuron — so the net delivery to the forebrain changes much less than the pharmacology suggests, until the autoreceptor desensitizes. The cleanest test of this is genetic. Mice engineered to carry higher or lower autoreceptor levels in the raphe, with the same receptors elsewhere untouched, showed strong effects on raphe firing rate but no change in basal forebrain serotonin or in conflict-anxiety measures; the high-autoreceptor mice had a blunted physiological response to acute stress, more behavioral despair and no behavioral response to antidepressant, and reducing autoreceptor levels before treatment converted non-responders into responders ⁶. This is mouse evidence, built deliberately to model a human risk allele, and it is causal within that model. The premise deserves a caution, because the delay itself is less solid than the explanation built on it. A meta-analysis of 28 randomized placebo-controlled trials in unipolar depression, 5,872 participants, reporting outcomes at two or more time points in the first four weeks, found that a model of early response fit the data better than a model of delayed onset: treatment with a selective serotonin reuptake inhibitor rather than placebo was associated with improvement by the end of the first week, with an increased chance of a 50 percent reduction in Hamilton Depression Rating Scale score by one week (relative risk 1.64, 95 percent confidence interval 1.2 to 2.25), and improvement continued at a decreasing rate for at least six weeks ³¹. The autoreceptor account explains a delay whose size is disputed.

The hypothesis was taken to the clinic once, and the result is instructive rather than decisive. In 1994 a letter reported that pindolol, a beta-blocker that also occupies the 5-HT1A autoreceptor, appeared to speed the response to serotonin reuptake inhibitors in depressed patients ³². In a double-blind randomized trial of 30 outpatients with major depressive disorder given citalopram with pindolol 5 mg three times daily or citalopram with placebo for six weeks, the group-by-time interaction favored pindolol (p = .01), the median time to first response was 22 days against 30 (p = .03), and the odds of sustaining remission were higher (odds ratio 5.00, 95 percent confidence interval 1.19 to 20.99, p = .03); a meta-analysis reported alongside it favored pindolol at two weeks (relative risk 1.68, 95 percent confidence interval 1.18 to 2.39, p = .004) and marginally at four to six weeks (relative risk 1.11, 1.02 to 1.20, p = .02) ³³. Thirty patients is a small trial and pindolol is not a selective autoreceptor blocker, so this is support for the mechanism rather than proof of it — and it remains the closest thing to a human test of the brake.

On the drug side of the same synapse, positron emission tomography can measure how much of the serotonin transporter — the pump that recycles the transmitter — a given dose occupies. In 12 healthy subjects and 12 patients with major depressive disorder, at least four weeks of high-dose venlafaxine, sertraline or citalopram produced mean striatal transporter occupancy of about 85 percent, significantly above the roughly 80 percent seen at minimum therapeutic dose, with about 15 percent of transporter still unoccupied ⁷. That measurement is in the striatum, not in the raphe, and the baseline is a group average rather than each patient's own pre-treatment scan.

The oldest antidepressants act on the same cells by a different route. In rats, pargyline, tranylcypromine, phenelzine and iproniazid — all monoamine oxidase inhibitors, which block the enzyme that breaks serotonin down — depressed raphe unit firing, and the depression was impaired by prior treatment with an inhibitor of serotonin synthesis, showing the slowing was serotonin-dependent feedback rather than a direct drug effect on the membrane ⁹.

And on the harm side: in mice, dorsal raphe serotonin arriving at the bed nucleus of the stria terminalis engages an inhibitory microcircuit of corticotropin-releasing factor neurons, silences that region's anxiolytic outputs to the ventral tegmental area and lateral hypothalamus, and enhances fear and anxiety — and the same circuit mediated the aversive effect of an acute dose of a selective serotonin reuptake inhibitor, with a corticotropin-releasing-factor type 1 receptor antagonist sufficient to prevent it ²². That is a mechanistic account, in mice, of the early worsening some anxious patients report in the first days of treatment.

The reversal of 2018: it is many systems, not one

Until recently almost every functional study treated the serotonin neurons of this nucleus as a single population — the assumption behind the idea that a drug could raise or lower serotonin tone and that this would mean something simple ¹⁹. Between 2015 and 2026 that assumption came apart, and the way it came apart is the most interesting thing about the nucleus.

The decisive paper used viral-genetic methods in mice to label serotonin neurons by where they send their axons ¹⁹. Subcortical-projecting and cortical-projecting serotonin neurons occupied distinct cell-body territories inside the nucleus and differed in whether they also expressed a vesicular glutamate transporter, the machinery for packaging glutamate — meaning some of these serotonin cells also release the brain's main excitatory transmitter. Amygdala-projecting and frontal-cortex-projecting serotonin neurons had largely complementary whole-brain collateralization patterns, received biased sets of inputs, and responded in opposite directions to the same aversive stimulus. Driving or silencing them gave opposite behaviors: amygdala-projecting neurons promoted anxiety-like behavior, frontal-cortex-projecting neurons promoted active coping under challenge ¹⁹. One nucleus, two systems pulling against each other.

The transcriptional work arrived alongside it. Single-cell RNA sequencing of 39,411 mouse dorsal raphe cells, combined with in situ hybridization and tracing, found at least 18 distinct neuron subtypes, of which 5 were serotonergic, with distinct molecular and anatomical properties — including a serotonergic subtype preferentially innervating the basal ganglia ³. A complementary atlas using subdomain-targeted sequencing and intersectional genetics described up to fourteen subtypes within the predominantly serotonergic lineage alone, many with biased cell-body distributions across the nucleus, and showed that the most molecularly distinct of them has its own efferent projections and electrophysiological properties ²⁰. Subtype counts of this kind depend on clustering choices and are best read as upper bounds, which both papers say. The same laboratory then joined the two approaches. Plate-based single-cell RNA sequencing of serotonin neurons across the dorsal and median raphe nuclei in mice yielded eleven transcriptomically distinct clusters, mapped by in situ hybridization to the principal dorsal raphe, the caudal dorsal raphe or the median raphe and differing in neuropeptides, receptors, ion channels and transcription factors; intersectional viral-genetic tools then showed that dorsal raphe serotonin neurons co-expressing vesicular glutamate transporter 3 preferentially innervate the cortex while those co-expressing thyrotropin-releasing hormone innervate subcortical targets, the hypothalamus in particular, and reconstruction of 50 individual neurons found diverse and segregated projection patterns at the single-cell level ³⁴. That is the bridge between the transcriptional and the anatomical reversals: molecular identity predicts where the axon goes.

An earlier study had already pointed the same way with a different method: retrograde tracing combined with slice electrophysiology and single-cell polymerase chain reaction in mice found that serotonin neurons projecting to medial prefrontal cortex, basolateral amygdala and dorsal hippocampus form largely non-overlapping populations, that amygdala-projecting cells have characteristic excitability and membrane properties, that the peptide galanin is restricted to the prefrontal-projecting cells, and that an atypical low-excitability subtype preferentially expresses vesicular glutamate transporter type 3 ²⁶.

Meanwhile the non-serotonergic majority of the nucleus came into view. Reviews now describe four groups — serotonergic, GABAergic, glutamatergic and dopaminergic — with GABAergic neurons providing potent local inhibition of the serotonergic ones and able to promote or constrain arousal depending on their projection target, glutamatergic neurons resisting anesthetic suppression, and dopaminergic neurons encoding salience and modulating pain ⁴. The dopaminergic group is now the best characterized of these across species, including in people. Immunohistochemistry in mouse, macaque and human brainstem, mapped against the ponto-mesencephalic junction as a common landmark, found tyrosine-hydroxylase-positive neurons intermingled with serotonergic neurons in all three species and co-expressing aromatic L-amino acid decarboxylase, so that they can make dopamine, while fewer than 20 percent expressed the dopamine transporter, implying little capacity for reuptake and a greater reliance on new synthesis; co-expression of calbindin and vasoactive intestinal peptide varied markedly between species ³⁵. In mice the population is causal for mood-related behavior: chronic unpredictable stress lowered their firing and synaptic drive and induced depression-related behavior, chemogenetic inhibition reproduced the phenotype, activation reversed the anhedonia and despair measures, and the effect was carried by a projection to the dorsal bed nucleus of the stria terminalis ³⁶. The species-conserved anatomy is human; the causal behavior is mouse.

A 2026 mouse study found that the opioid peptide enkephalin in this nucleus sits almost entirely outside the serotonin cells, distributed instead on glutamatergic and GABAergic neurons; knocking it down across the nucleus increased inflammation-induced mechanical sensitivity and odor avoidance and reduced sucrose preference and social engagement, but knocking it down in either subtype alone did not reproduce those effects ²⁵. A 2026 mouse model of pain with anxiety traced the phenotype to glutamatergic, not serotonergic, neurons here, acting through 5-HT2C receptors ²³. And a 2026 preprint — not peer reviewed, and to be treated as preliminary — reports that chronic stress drives gene expression changes in the GABAergic neurons of this nucleus and that inhibiting those neurons rescues the stress-induced shift from goal-directed to habitual responding ³⁷.

The practical consequence is that “raising serotonin” is not a description of a treatment. Which projection, which cell type and which firing pattern a drug or a stimulus engages determines the sign of the effect, and no current drug addresses any of those selectively.

The whole history, including every reversal

The dorsal raphe became visible in 1964, when histochemical fluorescence applied to rat brainstem demonstrated monoamines inside the cell bodies of brainstem neurons for the first time ³⁸. The lettered cell-group nomenclature used for the raphe ever since is conventionally traced to that work, though the 1964 report does not itself say so.

What followed was two decades in which the nucleus acquired, lost and partly regained each of its functions. Serotonin was proposed as the chemistry of sleep after lesions and synthesis blockade in cats produced insomnia ³⁹. Lysergic acid diethylamide was found to silence these neurons in rats at doses at or below the threshold for visible behavioral change, which produced the hypothesis that psychedelic experience is what happens when the raphe stops firing ⁸. Monoamine oxidase inhibitors were shown to depress the same cells, in a way that depended on serotonin synthesis ⁹. Then, in 1976, unit recording in cats showed dorsal raphe firing falling during sleep — the opposite of what a sleep-generating nucleus should do ¹². The sleep hypothesis in its original form did not survive it.

The hypothesis that section 10 of this page argues about was formed in the same years and from the same drugs. A historical review marking the sixtieth anniversary records that the catecholamine hypothesis of affective disorders was proposed in 1965 by Schildkraut and by Bunney and Davis, concerning a deficiency of noradrenaline in depression and its excess in mania, and that the serotonin hypothesis of depression was put forward in 1969 by Coppen in Britain and by Lapin and Oxenkrug in the Soviet Union; the clinical introduction of fluoxetine in the United States at the turn of the 1980s and 1990s then made the selective serotonin reuptake inhibitors the most prescribed antidepressants, and serotonin a popular hormone of happiness ⁴⁰. The hypothesis was built backward from drugs that worked, not forward from a measurement in patients, which is the fact that makes the 2022 argument possible.

The 1977 demonstration that stimulating the dorsal raphe directly inhibits amygdala neurons through a serotonergic pathway, abolished by destroying that pathway or depleting serotonin and restored by giving the serotonin precursor, established that this nucleus talks to the limbic system, and that intravenous lysergic acid diethylamide blocks that conversation ⁴¹. In 1987 the 5-HT1A somatodendritic autoreceptor was identified ⁵; in 1990 the human nucleus was mapped and counted ²; in 1992 the field's standard review of the structure and function of the brain serotonin system appeared ⁴².

The modern era has been a sequence of narrowings. In 1995 a positron emission tomography study of spontaneous migraine attacks found brainstem activation that persisted after sumatriptan had abolished the headache, and the phrase “migraine generator” entered the literature ¹⁴. In 2016 the group that had used that phrase revisited it and concluded the driver was more likely a change in hypothalamo-brainstem connectivity ¹⁵. In 2018 the serotonin neurons of the nucleus were shown to be several opposing systems rather than one ¹⁹. In 2019, raphe stimulation was shown to promote sleep after all — provided it was tonic rather than bursting ¹³. In 2022 an umbrella review concluded there was no consistent evidence that depression is caused by lowered serotonin, and in 2023 a 35-author rebuttal argued the conclusion was overstated ^16,17. None of these is a tidy end.

Sleep and arousal, at its evidence class

This is the function with the longest record and the most reversals. The original position, built on cats in which blocking serotonin synthesis produced insomnia, was that serotonin is the chemistry of sleep ³⁹. Unit recording killed it: dorsal raphe neurons fire fastest in active waking and slow as the animal falls asleep, reaching their lowest rates in rapid eye movement sleep ¹². For roughly four decades after that, the standard teaching was the opposite of the original claim — the serotonergic raphe is a waking system.

The scientist who had proposed the original hypothesis set out the contradiction himself, and his resolution anticipated the modern one. Reviewing the field in 1999, Jouvet recorded that destroying the raphe serotonin neurons or blocking serotonin synthesis with p-chlorophenylalanine produced a severe insomnia reversible by restoring synthesis, that this stood in direct contradiction to the demonstration that serotonin cell firing and serotonin release are highest in waking and lowest in sleep, and proposed that serotonin released during waking initiates a cascade of genomic events in hypnogenic neurons of the preoptic area, so that waking serotonin produces a homeostatic regulation of slow-wave sleep rather than sleep itself ⁴⁸. That is a hypothesis about a delayed, indirect action; the 2019 work reconciled the same two observations by a different route, through firing pattern ^13,48.

The 2019 study that reopened it worked in two species. In zebrafish, genetically ablating serotonin production by the raphe reduced sleep, reduced sleep depth and reduced the homeostatic rebound that normally follows sleep deprivation; pharmacological inhibition or ablation reduced sleep, and optogenetic stimulation increased it. In mice, ablation increased wakefulness and impaired the homeostatic response, while tonic optogenetic stimulation at a rate similar to the neurons' own baseline activity induced sleep — and burst stimulation, mimicking the pattern seen during arousing stimuli, induced wakefulness ¹³. Firing pattern, not firing rate, reconciles the 1969 and 1976 results. Evidence class: causal, in two animal species, with no human equivalent.

Anesthesia has become the other arousal literature for this nucleus, and it is almost entirely rodent. Reviews describe serotonergic neurons here as a key driver of emergence from general anesthesia, with activation speeding recovery and inhibition deepening unconsciousness, and GABAergic neurons providing the local inhibition that several anesthetics exploit ⁴. A 2026 mouse study mapped a dopaminergic projection from the ventral tegmental area onto these serotonergic neurons: both were suppressed under propofol, chemogenetic activation of either hastened emergence while inhibition prolonged anesthesia, and optogenetic activation of the dopaminergic terminals produced behavioral and electroencephalographic arousal that was abolished by inhibiting the serotonergic neurons ⁴⁹. Evidence class: causal in mice, for one anesthetic, with the clinical suggestions in the review literature not yet trial-based.

Mood, impulsivity and pain, each at its evidence class

Reward and punishment. In mice with serotonergic neurons identified by optogenetic tagging — the strongest available identification — a large fraction modulated their tonic firing rates over minutes across blocks of reward versus punishment trials, most were phasically excited by punishments, and a subset was phasically excited by reward-predicting cues; dopaminergic neurons showed no equivalent block-level change ²⁸. Evidence class: direct recording from identified cells in behaving animals, which is as good as this question gets outside humans.

Impulsivity, or rather its opposite. Optogenetic activation of these neurons in mice reduced the premature abandonment of waiting for a cue or a food reward and, when the reward was omitted, prolonged how long the animal waited; control experiments argued against general motor inhibition and against the stimulation being rewarding in itself ⁴⁴. The effect is conditional: it appeared at 75 percent reward probability but not at 50 or 25 percent, and grew when the timing of reward was unpredictable, which the authors model as serotonin raising the prior probability or subjective confidence of reward ⁵⁰. And it is carried by particular projections — stimulating serotonin terminals in orbitofrontal cortex promoted waiting nearly as well as stimulating the nucleus itself, nucleus accumbens stimulation did not, and medial prefrontal stimulation worked only when timing was uncertain ²⁷. Evidence class: causal and well replicated in mice. The human work is thinner but it is not absent, and it runs in the predicted direction. In a double-blind study, lowering serotonin synthesis in healthy volunteers by dietary tryptophan depletion increased premature responding on a human four-choice serial reaction time task built as a direct analogue of the rodent five-choice task — the same signature the rodent manipulation produces — and the increase correlated with the motor impulsivity subscale of the Barratt Impulsiveness Scale, while a delay-discounting questionnaire was unaffected ⁵¹. In a separate double-blind study using depletion and loading, low serotonin steepened the discounting of delayed rewards on a dynamic choice task without changing learning speed or choice variability ⁵². The two human results dissociate, as the animal work predicts they should, and both are small studies of healthy volunteers performing laboratory tasks, not patients with an impulse-control disorder.

Mood and coping. In rats, direct activation of medial prefrontal principal neurons had no detectable effect on active response to a behavioral challenge, but activating the subset of those neurons that project to the dorsal raphe produced a profound, rapid and reversible shift toward the active state — the wiring carried the effect, not the region ²¹. In mice, the anxiety-promoting and coping-promoting halves of the nucleus are separate projection systems ¹⁹, and serotonin arriving at the extended amygdala enhances fear through a corticotropin-releasing factor microcircuit ²². In larval zebrafish, early-life chronic unpredictable stress prevented the serotonergic system from habituating to repeated stress by shifting the balance of excitatory and inhibitory responses, most markedly in serotonergic cells co-expressing GABAergic markers, and ablating the neurons changed startle habituation and locomotion but not anxiety-like behavior — a dissociation that argues against a single anxiety function ⁵³. Evidence class: causal in animals, with behavioral readouts that are assays, not diagnoses.

Human mood. The direct human evidence for this nucleus doing anything in mood is thin. The strongest recent result is not about mood at all: functional imaging during a foraging-inspired task found that human dorsal raphe activity, and no other neuromodulatory nucleus examined, signaled changes in behavioral policy as the average value of the environment changed, while dorsal anterior cingulate and anterior insular cortex tracked relative value ⁴⁷. That is a correlational human finding about decision-making, and it is close to the animal reward and patience work without being the same thing.

The one human measurement aimed squarely at the brainstem raphe in depression is an ultrasound finding, and it is interesting and weak at once. Transcranial sonography — imaging the brainstem through the temporal bone and grading the echogenicity, the brightness, of the raphe region — has repeatedly shown reduced raphe echogenicity in depressed patients; a meta-analysis of 12 studies and 809 patients, using DSM-IV criteria and QUADAS-2 quality assessment, gave a pooled sensitivity of 0.66 (95 percent confidence interval 0.61 to 0.71) and a specificity of 0.84 (0.80 to 0.87), with a diagnostic odds ratio of 11.45 (5.57 to 23.02) and an area under the summary receiver operating characteristic curve of 0.86 (0.83 to 0.89) ⁵⁴. What the echogenicity signal corresponds to in tissue is not established, the reading is operator-dependent, and a sensitivity of 0.66 means a third of depressed patients look normal, so this is a group-level association with the raphe region and not a diagnostic test anyone should use as one.

Pain. The dorsal raphe's own contribution is less established than that of the caudal raphe nuclei, which project to the spinal cord and form part of the descending pain-modulating system ¹. The nucleus does have a pain history, and it ends in a qualification. In rats with bipolar electrodes implanted in the dorsal raphe or the ventral periaqueductal gray, electrical stimulation produced analgesia on the tail-flick test, and the long-acting opioid antagonist naltrexone reduced that analgesia by a mean of 79 percent given intravenously at 3.7 mg/kg and by 26, 52, 81 and 83 percent at 0.3, 1.0, 3.0 and 10 mg/kg intraperitoneally, which the authors read as showing that opiate mechanisms, under those conditions only opiate mechanisms, carry the effect ⁵⁵. The human counterpart came from electrodes in the periventricular and periaqueductal gray of six patients with intractable pain: stimulation relieved pain without changing the acute pain threshold, repetitive stimulation produced tolerance to both the stimulation and to narcotics, and naloxone reversed the relief in five of the six ⁵⁶. So stimulating in and around this region relieves pain, but through an opioid-dependent mechanism in a structure the dorsal raphe sits inside, which is a reason to keep this nucleus's own serotonergic contribution to pain modest in the claims made here.

What has been shown recently is more specific and mostly non-serotonergic: in a mouse inflammatory pain model with anxiety, the glutamatergic neurons of this nucleus were overactive, chemogenetic activation reproduced the comorbid phenotype in naive animals and inhibition reversed it, and 5-HT2C receptor levels rose, with agonism worsening and antagonism attenuating the state ²³. Knocking down the opioid peptide enkephalin across the nucleus in mice increased inflammation-induced mechanical sensitivity ²⁵. And in a carbon-dioxide model of panic, blocking an acid-sensing ion channel in the lateral wings of the dorsal raphe reduced escape behavior without changing locomotion, an effect abolished by a 5-HT1A antagonist in the dorsal periaqueductal gray — evidence that this nucleus participates in a defensive circuit rather than in pain as such ⁵⁷. Evidence class: causal, mouse, single models each.

Three further studies place the non-serotonergic populations in pain circuits. In a mouse paw-incision model, tracing identified an inhibitory ascending pathway from the dorsal raphe to the paraventricular thalamic nucleus; inhibiting it produced nociceptive sensitivity in normal mice, activating it relieved incision pain, and a GABA-A, not GABA-B, receptor antagonist injected into the thalamic target blocked the effect ⁵⁸. In a mouse inflammatory model, optogenetically activating neurotensin-expressing neurons of the interpeduncular nucleus, or their projections to the dorsal raphe, induced both pain-like and depressive-like behavior in naive animals, while inhibiting them reversed both in inflamed animals; the projection co-released GABA and neurotensin, and a GABA-A antagonist in the dorsal raphe prevented the behavioral effects, while blocking neurotensin receptors there induced pain without affecting the depressive-like behavior ⁵⁹. And in a nerve-injury model, dorsal raphe dopaminergic neurons were less active and memory was impaired; chemogenetic or optogenetic activation of those neurons, or of their terminals in the bed nucleus of the stria terminalis, restored memory performance ⁶⁰. Evidence class: causal in mice, one model each, and in male animals wherever sex was reported.

Social behavior should be mentioned because it is where the newest work is. A 2026 study, primarily in male mice, visualized oxytocin projections from the hypothalamic paraventricular nucleus onto both the serotonergic and the dopaminergic neurons of the dorsal raphe, found that blocking oxytocin receptors here reduced social interaction without affecting locomotion, and found that oxytocin inhibits the serotonergic neurons while activating the dopaminergic ones in the same nucleus ⁶¹. Evidence class: preliminary, one species, one sex.

The migraine generator: what the 1995 study showed, and what became of it

In 1995, positron emission tomography was used to measure regional cerebral blood flow, as an index of neuronal activity, during spontaneous migraine attacks. Blood flow rose in cingulate, auditory and visual association cortices and in the brain stem; after sumatriptan had produced complete relief from headache and from sound and light sensitivity, only the brain stem activation persisted ¹⁴. The authors read this as an imbalance between brainstem nuclei regulating pain suppression and vascular control, and the phrase that stuck to it was “migraine generator”.

The vascular half of that reading has not aged well either. Magnetic resonance angiography during spontaneous unilateral attacks of migraine without aura in 19 women found no statistically significant dilatation of the extracranial arteries on the painful side compared with attack-free days — external carotid mean difference 1.2 percent, superficial temporal 3.6 percent, middle meningeal 1.7 percent, cervical internal carotid 2.3 percent, none of them significant — while the intracranial arteries dilated modestly (middle cerebral 13.0 percent, p = 0.001; cerebral internal carotid 11.5 percent, p = 0.0004; cavernous internal carotid 11.4 percent, p = 0.001), and the authors concluded that research should move to the peripheral and central pain pathways rather than to simple arterial dilatation ⁶². Nineteen patients, all of them women, scanned during attacks is a small study, but it is the direct test, and it means the 1995 interpretation rested on a vascular model that was later weakened.

Two things should be said at once. First, the finding was real and was replicated in various forms: a single patient with both cluster headache and migraine showed dorsal rostral brainstem activation with his migraine and no hypothalamic activation ⁶³; a study of 24 migraineurs and 8 controls with chemically triggered attacks found significant activation in the dorsal lateral pons, ipsilateral to the side of the headache in the lateralized groups, and again persisting after sumatriptan ⁶⁴. Second, the activated region was never shown to be the dorsal raphe. The dorsal raphe is a midline structure. The replicated activation is in the dorsal lateral pons and is lateralized to the side of the pain ⁶⁴. A lateralized pontine focus is anatomically not the same thing as a midline raphe nucleus, and this page is the place to say so plainly: the “serotonergic migraine generator” is an inference laid over an imaging blob, not a localization.

The attention then moved forward in the brain. A study of eight patients scanned in the painless premonitory phase of triggered attacks found activations in the posterolateral hypothalamus, midbrain tegmental area, periaqueductal gray, dorsal pons and several cortical regions — which fits the premonitory symptoms patients describe, such as fatigue and yawning, better than a pontine pain switch does ⁶⁵.

The reversal came from the group that had used the phrase. One patient was imaged every morning for 30 days, capturing three complete untreated attacks. Hypothalamic response to trigeminal nociceptive stimulation changed during the 24 hours before pain onset, and the hypothalamus showed altered functional coupling with the spinal trigeminal nuclei and with the dorsal rostral pons — named in the paper as the region of the migraine generator — both on the preictal day and during pain. The authors' conclusion was explicit: although the brainstem is closely linked to migraine biology, the real driver of attacks might be the functional change in hypothalamo-brainstem connectivity ¹⁵. A follow-up with nine episodic migraineurs, 282 scanned days and 15 spontaneous attacks found preictal strengthening of connectivity between the right nucleus accumbens and the dorsal rostral pons, and ictal strengthening between the dorsal pons and the hypothalamic area ⁶⁶.

So: the brainstem activation of 1995 stands as an observation. The word “generator” does not, and was withdrawn in substance by the people who popularized it. Whether the dorsal raphe specifically is involved in migraine remains unestablished by imaging. What is established is that serotonin matters in migraine for a different and better-evidenced reason, which is the subject of the next section.

Nausea, the triptans and cyclic vomiting syndrome

The serotonin connection to migraine that has survived is pharmacological. The triptans are agonists at 5-HT1B and 5-HT1D receptors: the 5-HT1B receptors on meningeal arteries, where they constrict dilated cranial extracerebral vessels, and the 5-HT1D receptors on trigeminovascular nerve endings, where they inhibit the release of transmitters driving neurogenic inflammation ^67,68. There is also a central site. In anesthetized cats, with the superior sagittal sinus stimulated electrically and trigeminocervical neurons recorded in the dorsal C2 spinal cord, the baseline probability of cell firing was 0.61 ± 0.1 at a fastest-peak latency of 11.1 ± 0.4 milliseconds; serotonin infused at 15 micrograms per kilogram per minute for five minutes inhibited the evoked firing from 15 to 20 minutes after the infusion began and for about 20 minutes thereafter, and a 5-HT1B/1D antagonist abolished the inhibition entirely ⁴³. That is serotonin damping second-order trigeminal traffic in the brainstem — the kind of action a descending raphe projection could in principle perform, though the experiment used infused serotonin and did not test the raphe.

Cyclic vomiting syndrome enters this page through that door. It is a disorder of stereotyped, self-limited episodes of intense vomiting separated by well intervals, with a periodicity that is often circadian or circannual, and it is classified among the episodic syndromes associated with migraine ⁶⁹. Functional magnetic resonance imaging in both cyclic vomiting syndrome and migraine has shown diminished connectivity of the insula with the sensorimotor network, which is suggestive of shared pathophysiology and no more than that ⁶⁹.

The treatment overlap is the strongest practical argument for the link. The 2019 guideline of the American Neurogastroenterology and Motility Society and the Cyclic Vomiting Syndrome Association strongly recommends a tricyclic antidepressant such as amitriptyline as first-line prophylaxis for moderate-to-severe adult disease, with topiramate or aprepitant as alternates, and conditionally recommends serotonin antagonists such as ondansetron and triptans such as sumatriptan to abort acute attacks; it also recommends screening for comorbid migraine, anxiety, depression, autonomic dysfunction, sleep disorders and substance use ⁷⁰. In children, a systematic review that screened 799 studies published between December 1954 and February 2018 and included 54 found ondansetron best supported as rescue therapy and sumatriptan and aprepitant as candidate abortive therapies, while stating that the quantitative evidence rests on six case series and that controlled trials are still needed ⁷¹. Injectable or intranasal sumatriptan is often effective in children, and amitriptyline, cyproheptadine and propranolol are the common prophylactics ⁶⁹.

Those recommendations were superseded in 2025. GRADE-based guidelines from the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition, developed by a multidisciplinary panel with a patient representative and registered as PROSPERO CRD42022310108, issued 16 recommendations and made exactly one of them strong: use anti-migraine agents to abort attacks in children with a personal or family history of migraine ⁷². Everything else is conditional — oral and intravenous 5-HT3 and neurokinin-1 receptor antagonists and early presentation for abortive care, and for prevention trigger avoidance, supplements, biobehavioral and neuromodulation approaches, beta-blockers, neurokinin-1 and 5-HT2A antagonists and tricyclic antidepressants, with anticonvulsants reserved for refractory disease. A companion 2025 guideline on diagnosis recommends screening serum and urine testing and an upper gastrointestinal radiographic series for all children with suggestive symptoms, finds the other screening tests low-yield, and treats cannabinoid hyperemesis syndrome as a related but separate entity ⁷³. The strong recommendation is the part that matters for this page: the one thing the panel was confident about is that the drug which works is the migraine drug.

A distinction worth holding. The serotonin receptor through which the antiemetics work is not a receptor of this nucleus's forebrain projection at all. The 5-HT3 receptor is a ligand-gated ion channel — the only serotonin receptor that is an ion channel rather than a G-protein-coupled receptor — and the setron drugs bind it; the structure of that binding has been solved by cryo-electron microscopy for palonosetron and modeled for granisetron, dolasetron, ondansetron and cilansetron ⁷⁴. Its clinically relevant locations for nausea are peripheral and in the brainstem chemoreceptive territory, and ondansetron's main safety concern — prolongation of the QT interval, the stretch of the electrocardiogram during which the ventricles reset electrically after each beat, which when it lengthens leaves the heart vulnerable to a dangerous rhythm, with the risk of torsades de pointes — is a cardiac channel effect that has nothing to do with the raphe ⁷⁵. In short: serotonin is central to both migraine and vomiting, but through different receptors in different places, and the dorsal raphe is not an established site where nausea is generated.

It is, however, an established site where nausea can be suppressed, through the brake described earlier on this page. In cats, the 5-HT1A agonist 8-OH-DPAT given before an emetic challenge suppressed vomiting caused by motion, by xylazine and by cisplatin — three stimuli with different entry points — which the authors read as an action at a convergent structure at or near the vomiting center ⁷⁶. The localization was later pinned to this nucleus: in shrews and rats, cannabidiol suppressed nicotine-, lithium- and cisplatin-induced vomiting and lithium-induced conditioned gaping, both effects were blocked by 5-HT1A antagonists, a 5-HT1A antagonist placed in the dorsal raphe reversed the effect of systemic cannabidiol, and cannabidiol injected into the dorsal raphe itself suppressed nausea-like behavior ⁷⁷. And emetic challenges of several kinds induce Fos expression here: in the least shrew, cisplatin and selective 5-HT3, dopamine D2/D3 and neurokinin-1 agonists all raised Fos immunoreactivity in the dorsal raphe as well as in the dorsal vagal complex ⁷⁸. Evidence class: causal animal pharmacology in cats, shrews and rats, with no human equivalent — and a suppression site is not a generator.

Psychedelics: what this nucleus does, and what it does not

The first pharmacological fact about psychedelics and the brainstem was discovered here. In 1968, units in serotonin-rich midbrain areas of rats stopped firing reversibly after parenteral lysergic acid diethylamide, at a dose at or below the threshold for visible behavioral effects ⁸. In 1977 the same laboratory showed that intravenous lysergic acid diethylamide blocked the inhibition that raphe stimulation normally imposes on amygdala neurons ⁴¹. For some years the natural reading was that psychedelic experience is what the forebrain does when the raphe goes quiet.

The reading was tested and found wanting by the end of the 1970s, in cats recorded while awake and moving. Lysergic acid diethylamide lowered raphe unit activity dose-dependently and raised certain behaviors dose-dependently, with the peaks temporally correlated — but low doses produced only small falls in firing alongside significant behavioral change, the behavioral changes outlasted the depression of firing, and raphe neurons remained at least as responsive to the drug in tolerant animals as in non-tolerant ones ⁷⁹. A fuller study two years later added the decisive dissociation by drug class: psilocin lowered raphe firing dose-dependently, but the phenethylamine hallucinogens DOM and mescaline produced large behavioral changes with no overall effect on raphe neurons, some units rising, some falling and some unchanged ⁸⁰. If the two chemical families act alike in behavior while only one of them silences the raphe, silencing cannot be the mechanism; the authors wrote in 1981 that the serotonin hypothesis of hallucinogenic drug action should be re-evaluated, and by 1999 the field's own synthesis placed the common site at the 5-HT2A receptor on cortical pyramidal cells instead ⁸¹.

The decisive human experiment came in 1998: in healthy volunteers, the psychotomimetic effects of psilocybin were blocked dose-dependently by ketanserin, a 5-HT2A receptor antagonist, and by risperidone, and were increased rather than reduced by the dopamine antagonist haloperidol ¹⁰. The effect therefore runs through 5-HT2A receptor activation, independently of dopamine. Those receptors are concentrated in the cortex, and that distribution has been mapped in living people: a high-resolution atlas built from positron emission tomography and magnetic resonance imaging in 210 healthy individuals, with regional binding measures validated against postmortem autoradiography and converted into protein densities in picomoles per milliliter, places 5-HT2A among the cortically dominant targets of the four receptors and the transporter it covers ⁸². The pharmacology has been measured there too: in eight healthy volunteers given single oral doses of psilocybin from 3 to 30 mg, cerebral 5-HT2A occupancy rose with dose to as much as 72 percent, plasma psilocin levels and occupancy fit a single-site binding model, and subjective intensity tracked both ⁸³. The classical psychedelics are agonists at a receptor on cortical neurons; they are not, in any primary sense, raphe drugs.

What the raphe does in this picture is real but secondary. Lysergic acid diethylamide and other classical psychedelics also engage 5-HT1A receptors, which on raphe neurons are the inhibitory autoreceptor described earlier, which is why these drugs slow raphe firing in rats ^5,8. The slowing is a consequence of the pharmacology, and a change in the serotonergic input the cortex receives, but it is not the mechanism by which the drug changes experience; the human blockade experiment places that at 5-HT2A ¹⁰. Since 2024 that secondary receptor has become a target in its own right. Five cryogenic electron microscopy structures of 5-HT1A bound to 5-methoxy-N,N-dimethyltryptamine, the psychedelic found in the toxin of Colorado River toads, together with systematic medicinal chemistry, receptor mutagenesis and mouse behavior, defined what makes a 5-methoxytryptamine potent and selective at 5-HT1A rather than at 5-HT2A; a 5-HT1A-selective analogue produced no hallucinogenic-like effects in mice while retaining anxiolytic-like and antidepressant-like activity in socially defeated animals ⁸⁴. If that separation holds in people, the receptor this nucleus uses as its own brake becomes a route to the benefit without the experience — which is a hypothesis with mouse behavior and protein structures behind it and no human trial.

The 2023 work sharpened the point in an unexpected direction. Activation of 5-HT2A receptors is necessary for psychedelic-induced cortical plasticity, but not every agonist at that receptor produces it — serotonin itself does not. Molecular and genetic work in neurons and mice showed that it is the intracellular pool of 5-HT2A receptors, rather than the surface pool, that mediates the plasticity-promoting effect, which the authors offer as the explanation, and which raises the possibility that serotonin is not the endogenous ligand for that intracellular pool at all ¹¹. If that holds, then the brain's own serotonin — the serotonin this nucleus supplies — is doing something categorically different at the same receptor from what a psychedelic does.

A rival account, published the same year, moves the target off the serotonin receptor entirely. Lysergic acid diethylamide and psilocin were shown to bind TrkB — the receptor for brain-derived neurotrophic factor, a growth factor that supports synaptic plasticity — with affinities about a thousandfold higher than those reported for conventional antidepressants, at a site in the receptor's transmembrane domain that overlaps only partly with the antidepressant site; in mice, the effects on neurotrophic signaling, plasticity and antidepressant-like behavior depended on TrkB binding and on endogenous brain-derived neurotrophic factor signaling and were independent of 5-HT2A activation, while the head-twitch response depended on 5-HT2A and not on TrkB ⁸⁵. Both accounts rest on in vitro and mouse work, they disagree about which receptor carries the plasticity, and neither has been tested in a person. The honest statement is that the plasticity mechanism is contested ^11,85.

On clinical effect, the honest statement in 2026 is that the evidence is promising and weak. A living systematic review current to July 2025 pooled 30 double-blind randomized trials with 1480 participants, 83.3 percent of them judged at high risk of bias, and found for major depressive disorder a standardized mean difference of -0.62 (95 percent confidence interval -0.97 to -0.28) for psilocybin, ayahuasca or lysergic acid diethylamide (LSD), with GRADE certainty rated very low; moderate certainty was reached only for MDMA — 3,4-methylenedioxymethamphetamine, the drug sold illicitly as ecstasy — in post-traumatic stress disorder against placebo ⁸⁶. A 2025 meta-analysis compared the control arms of 17 trials and found that patients receiving control treatment in psilocybin trials improved far less (standardized mean change 0.50) than control patients in trials of selective serotonin reuptake inhibitors (1.00) or esketamine (1.12), with control response rates 14 and 23 percentage points higher respectively, which the authors read as a sign that psilocybin's advantage is overestimated because participants can tell which arm they are in ⁸⁷. Neither of these is an argument that psychedelics do not work. Both are arguments that the size of the effect is not yet known.

Two 2026 trials sharpen that. In a double-blind trial at a single Stockholm clinic, 35 adults with moderate-to-severe recurrent major depressive disorder received a single 25 mg dose of psilocybin or an active placebo, niacin 100 mg, with five psychotherapeutic support sessions; the psilocybin group improved more on the Montgomery-Asberg Depression Rating Scale at day 8 (mean difference -7.27, 95 percent confidence interval -12.89 to -1.65, p = .01) and again at days 15 and 42, but the groups no longer differed at day 365 (-3.68, -9.30 to 1.94, p = .20), and two participants given psilocybin had persistent severe anxiety that needed medical attention ⁸⁸. In treatment-resistant depression the result was negative on its primary endpoint: in a two-center, triple-blinded phase 2b trial of 144 adults withdrawn from antidepressants, response at week 6 was 17.0 percent after psilocybin 25 mg, 12.5 percent after psilocybin 5 mg and 10.6 percent after nicotinamide 100 mg, and the first hierarchical comparison was not significant (adjusted odds ratio 1.73, 95 percent confidence interval 0.53 to 6.23, p = .19); the 25 mg dose was linked to more suicidal ideation on dosing days (4 percent against 1 to 2 percent) and to two serious adverse reactions, one of them a case of hallucinogen persisting perception disorder ⁸⁹. The authors of that trial call it inconclusive, which is the fair word.

The MDMA literature behind that one moderate rating carries a mark that belongs on the record. In November 2024 the journal Psychopharmacology retracted three papers from the sponsor's program: the pooled analysis of six phase 2 randomized trials that set the design and rationale for the phase 3 trials ⁹⁰, the long-term follow-up pooled analysis of the same six trials ⁹¹, and an analysis of how discontinuing reuptake inhibitors affected treatment response ⁹². The authors were affiliated with the Multidisciplinary Association for Psychedelic Studies and with its public benefit corporation, the commercial sponsor of the program, and nothing on this page about MDMA should be read without that.

The serotonin hypothesis of depression, as it stands in 2026

In 2022 a registered umbrella review set out to test one specific claim: that depression is caused by lowered serotonin concentration or activity ¹⁶. It gathered systematic reviews, meta-analyses and large data-set analyses across the six literatures that bear on it, and reported them area by area. Two meta-analyses of the serotonin metabolite 5-HIAA in body fluids showed no association with depression, the largest with 1,002 participants. A meta-analysis of cohort studies of plasma serotonin showed no relationship with depression but did show that lowered serotonin was associated with antidepressant use, in 1,869 people. Serotonin receptor binding (largest study 561 people) and serotonin transporter binding (largest 1,845) gave weak and inconsistent evidence of reduced binding in some regions — which, the review notes, would be consistent with increased synaptic serotonin if that were the original abnormality — and the effects of prior antidepressant use were not reliably excluded. A meta-analysis of tryptophan depletion found no effect in most healthy volunteers (566), with weak evidence of an effect in those with a family history of depression (75); no systematic review of depletion studies had been done since 2007. And the two largest and highest-quality genetic studies, an association study of 115,257 people and a collaborative meta-analysis of 43,165, found no association between the serotonin transporter gene and depression and no interaction between genotype, stress and depression ¹⁶.

Those nulls are the end of a particular story, and it is worth telling. In 2003, a prospective birth-cohort study reported that a functional polymorphism in the promoter region of the serotonin transporter gene moderated the effect of stressful life events: people with one or two copies of the short allele showed more depressive symptoms, more diagnosable depression and more suicidality in relation to stressful events than those homozygous for the long allele ⁹³. It became one of the most cited findings in psychiatric genetics. In 2019, a preregistered analysis of the 18 candidate genes studied ten or more times, in population and case-control samples ranging from 62,138 to 443,264 people, across several definitions of depression and several environmental moderators including childhood abuse and socioeconomic adversity, found no clear evidence for any candidate gene main effect and none for any gene-by-environment moderation; as a set, the depression candidate genes were no more associated with depression than non-candidate genes, and the authors concluded that the earlier literature was largely false positives ⁹⁴. The serotonin transporter gene was the central candidate in that literature.

The conclusion was that the main areas of serotonin research provide no consistent evidence of an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity or concentrations; and that some evidence was consistent with long-term antidepressant use reducing serotonin concentration ¹⁶.

The response was immediate and substantial. A reply signed by 35 authors argued the conclusion was overstated, on grounds of methodological weakness in the review process, selective reporting of data, over-simplification, and errors in the interpretation of neuropsychopharmacological findings, working in detail through tryptophan depletion and serotonergic molecular imaging as the two most relevant literatures ¹⁷. A second reply appeared in the same journal from a separate group ⁴⁵. A 2025 analysis of the exchange identified post-hoc protocol amendments, selective reporting and down-weighting of serotonergic findings as weakening the umbrella review's conclusiveness, and argued for serotonin as a modulatory factor in vulnerable people rather than a sole cause ⁹⁵.

Conflicts of interest run in both directions here and should be stated. Among the signatories of the principal rebuttal, the declared affiliations include one author affiliated with the pharmaceutical company H. Lundbeck A/S and one with Intra-Cellular Therapies ¹⁷. The authors of the umbrella review hold long-standing and publicly argued positions critical of antidepressant prescribing. Neither fact settles anything about the data; both belong on the record.

Since the exchange, one human measurement has come in on the other side. Positron emission tomography with a receptor tracer displaced by the brain's own serotonin, combined with an oral amphetamine challenge, compared 17 antidepressant-free patients in a major depressive episode with 20 healthy controls. Frontal binding fell significantly in the controls (1.04 ± 0.31 to 0.87 ± 0.24, p < .001) but not in the patients (0.97 ± 0.25 to 0.92 ± 0.22, not significant), and the difference between groups was significant (15 ± 14 percent versus 6.5 ± 20 percent, p = .041) ⁴⁶. This is the first direct assessment of serotonin release capacity in depressed people and it found reduced release. It is also small, overwhelmingly male, and age-mismatched by 12 years on average, and a displacement signal is an inference rather than a measurement of serotonin.

Where this leaves a reader in 2026, without either side's slogan. The specific claim that depression is caused by a deficiency of serotonin is not supported by the accumulated biomarker, depletion and genetic literatures, and the umbrella review is right that it was never well supported; the paper continues to circulate in mainstream psychiatric education, having been reprinted in an American Psychiatric Association journal in 2026 ¹⁸. The broader claim that the serotonin system is involved in depression and in the action of antidepressant drugs is a different claim, and is supported by the receptor and release work, by the autoreceptor genetics in animals, and by the trial evidence that these drugs change symptoms in some patients. A systematic review and network meta-analysis of 522 double-blind randomized trials comprising 116,477 participants found all 21 antidepressants studied more effective than placebo for the acute treatment of adult major depressive disorder, with odds ratios from 2.13 (95 percent credible interval 1.89 to 2.41) for amitriptyline down to 1.37 (1.16 to 1.63) for reboxetine; the authors rated 9 percent of the trials at high risk of bias, 73 percent moderate, and the certainty of the evidence moderate to very low ⁹⁶. Modest average effects of uncertain certainty are what is actually on the table ^6,7,46,96. Neither claim implies the other. A drug can work through a system that is not the cause of the illness — that is true of most of medicine.

Where the nucleus shows up in disease

Alzheimer's disease. The dorsal raphe is among the first structures in the brain to accumulate the abnormal phosphorylated tau protein that defines Alzheimer pathology. In 118 brains from a Brazilian brain bank, 38 staged as Braak stage 0 and 80 at stage 1 or above, neurofibrillary changes were present in the supratrochlear subnucleus of the dorsal raphe in every case at stage 1 or above and in more than a fifth of the stage 0 cases — that is, before the transentorhinal cortex, the structure the Braak staging system treats as the first cortical site ⁹⁷. The authors read this as support for transneuronal spread from the raphe to its interconnected cortical targets. This is autopsy data, cross-sectional: the sequence across individuals is inferred, not watched in any one person.

A 2025 study added quantitative human detail by pairing high-resolution postmortem magnetic resonance imaging with validated tau histopathology in the same 11 samples, from controls through the earliest Braak stages. Locus coeruleus volume was 30.0 percent smaller between Braak stage 0 and stage II (p = 0.0381), tau in that nucleus rose 37.6 percent between stages I and II (p < 0.0001), and in the caudal dorsal raphe the authors found a steep rostrocaudal gradient of tau severity, 58.2 percent lower caudally than rostrally (p < 0.0001), which they read as selective regional vulnerability within each nucleus ⁹⁸. Eleven samples is a small series, and what it says is that the relevant unit of vulnerability is a subregion rather than a whole nucleus.

A 2026 study put a mechanism under that observation. In htau mice and in post mortem human tissue, the centromedial part of the dorsal raphe proved to be composed predominantly of neurons that are both serotonergic and glutamatergic. In the mouse these dual-phenotype neurons differentially express ion-channel genes and are more excitable, which the authors propose promotes tau accumulation; in human Alzheimer brain at Braak stage II, the potassium-channel gene KCNA4 was selectively reduced in those neurons, which were more vulnerable to phosphorylated tau than the serotonin-only neurons beside them ⁹⁹. Causal evidence in mice, correlational in humans, and a coherent hypothesis connecting the two.

Sudden infant death syndrome. The best-known serotonergic pathology in the brainstem is not in this nucleus. In frozen medullae from infants who died of sudden infant death syndrome, cases had a higher serotonin neuron count than controls (148.04 ± 51.96 versus 72.56 ± 52.36 cells, p < .001), a higher neuron density, lower 5-HT1A receptor binding density in all nine nuclei examined (p ≤ .01), and a lower ratio of transporter binding to neuron count (0.70 ± 0.33 versus 1.93 ± 1.25 fmol/mg, p = .001); male cases had lower 5-HT1A binding in the raphe obscurus than female cases (16.2 ± 2.0 versus 29.6 ± 16.5 fmol/mg, p = .04) ²⁹. Those are the caudal, medullary raphe nuclei, which regulate breathing, blood pressure and arousal — the family relatives of the dorsal raphe, not the nucleus itself. The control groups numbered as few as 6 to 10 per analysis.

Panic and the carbon dioxide response. In mice, blocking the acid-sensing ion channel ASIC1a — a channel opened by a fall in pH, and therefore by rising carbon dioxide — in the lateral wings of the dorsal raphe reduced escape behavior during exposure to 20 percent carbon dioxide, without affecting baseline movement, and the effect was abolished by a 5-HT1A antagonist injected into the dorsal periaqueductal gray ⁵⁷. The lateral wings are one of the subdivisions of this nucleus, and the result places them in the defensive circuit that responds to suffocation signals. A rodent panic analogue.

Anesthesia. Beyond arousal, the clinical observation reported in the review literature is that patients with serotonergic disorders such as depression and anxiety often respond atypically to anesthetics, which is a reason to plan perioperative care individually; the review states this as a clinical implication of circuit-level findings rather than as a trial result ⁴.

What changed, and what is still unknown

What changed, first. The belief that serotonin from these cells generates sleep was overturned in 1976 when their firing was found to fall as animals fell asleep ¹², and then partly restored in 2019 when tonic stimulation at the neurons' own baseline rate was shown to induce sleep while burst stimulation induced waking ¹³. The belief that psychedelic experience is caused by silencing the raphe was displaced in 1998, when blocking the cortical 5-HT2A receptor in human volunteers blocked the effects of psilocybin ¹⁰. The belief that the brainstem holds a migraine generator was revised in 2016 by the group that named it ¹⁵. The belief that depression is caused by low serotonin was challenged in 2022 and has not been restored, although the narrower claim that the serotonin system participates in depression and in drug response survives ^16,17,46. And the belief that this nucleus is one functional unit was ended in 2018 and after ^3,19,20,26.

What is still unknown, and the honest list is long. How many human dorsal raphe neurons make serotonin is not established; the available human counts of serotonin-synthesizing neurons are for the adjacent pontine groups ^2,24. A human atlas of this nucleus at the resolution of the mouse work does not yet exist, but the gap has narrowed. A single-nucleus RNA sequencing census of the whole adult human brain sampled more than three million nuclei from about 100 dissections across forebrain, midbrain and hindbrain in three postmortem donors, and resolved 461 clusters and 3,313 subclusters, with the highest neuronal diversity found in the midbrain and hindbrain ¹⁰⁰. Three donors and whole-brain dissections are not the same thing as a targeted atlas of this one nucleus, so whether the five or fourteen mouse serotonergic subtypes correspond to anything in people is still open ^3,20,100. No human experiment has selectively activated or silenced any part of this nucleus, so every causal statement on this page about behavior is an animal statement ^13,19,21,44. The available human functional evidence is a small number of imaging studies, the best of them correlational ^46,47.

Several specific questions are open in a way that matters clinically. Whether the delay before antidepressants work is genuinely explained by autoreceptor desensitization has been tested in people, and the answer is half an answer. In 19 medication-free patients with major depressive disorder scanned by positron emission tomography before and after 5 to 9 weeks of a selective serotonin reuptake inhibitor (mean 47 ± 8 days), raphe 5-HT1A autoreceptor binding fell by 18 percent (F = 5.12, p = .036) — but the size of the fall was unrelated to improvement in depression (F = 1.27, p = .276) ¹⁰¹. The human promoter variant the mouse work was built to model is also real: the C(-1019)G 5-HT1A promoter polymorphism was enriched about twofold in depressed patients and fourfold in completed suicides in separate cohorts, and the G allele abolishes repression of the receptor gene by the transcription factor NUDR, which would raise autoreceptor levels ¹⁰². So the receptor changes as predicted and the genetics points the predicted way; what has not been shown in people is that either one accounts for the clinical response ^5,6,101,102. Whether the early anxiety some patients feel on starting a selective serotonin reuptake inhibitor is the circuit described in mice is plausible and untested in humans ²². Whether targeting a projection rather than a transmitter could give an antidepressant without the opposing effects is the obvious implication of the subsystem work and has no drug behind it yet ^19,27. Whether the glutamatergic, GABAergic, dopaminergic and enkephalinergic populations of this nucleus are therapeutic targets in their own right is being asked only now, mostly in mice ^4,23,25,37.

Two cautions to carry away. The first is the species problem, stated in the human anatomy literature itself: humans and other primates have far more pontine serotonin neurons than rats and cats, arranged differently, so the quantitative rodent work does not transfer without argument ²⁴. The second is the one the field has learned three separate times on this page — with sleep, with migraine and with depression. A structure that lights up, slows down or correlates with a state is not thereby the cause of it, and the interval between a confident name and its public revision has run, in each of these cases, to about two decades.

The nucleus that emerges from all of this is not the serotonin dial of the textbooks. It is a small, crowded, chemically mixed piece of midline brainstem that sends several opposed messages to several different places at once, using a transmitter that often does not even cross a synapse, and that is addressed by four of the most-prescribed drug classes in medicine almost by accident — because they alter the transmitter rather than the system ^1,7,9,19. Two stars on this site carry the parts of this argument that are not about anatomy. The serotonin star follows the transmitter itself — where it is made, what fraction of it is in the gut, and what measuring it in blood or cerebrospinal fluid can and cannot mean. The monoamine-hypothesis star follows the theory, from the 1965 catecholamine proposal and the 1969 serotonin proposal through the 2022 umbrella review and the replies, and is the place to take the argument of section 10 further.

The tests, one by one

• Postmortem cytoarchitecture and immunohistochemistrySerial sectioning of the brainstem soon after death, stained either for all neurons or with an antibody against the serotonin-synthesizing enzyme, then counted. This is how the human nucleus is known at all: five subnuclei, 235,000 ± 15,000 neurons, four characteristic cell shapes ^2,24.To establish normal human anatomy, and to find disease: neurofibrillary tau pathology, cell counts, receptor bindingGood for structure; the four-brain sample and three-hour postmortem delay in the reference study are the practical ceiling on precision ²It cannot say which neurons make serotonin unless an enzyme antibody is used, and it gives no information about activity; the available human serotonin-neuron counts are for the adjacent pontine groups, not the dorsal raphe itself ²⁴
• Functional magnetic resonance imaging of the brainstemImaging a structure a few millimeters across on the moving midline of the brainstem. A 2026 study using a foraging task found that activity here, and in no other neuromodulatory nucleus examined, signaled changes in behavioral policy as the value of the environment changed ⁴⁷.To ask what the human nucleus does during behavior, which animal work can only suggestGood enough to separate this nucleus from its neighbors in a group analysis ⁴⁷The signal is blood flow, not firing, and cannot distinguish serotonergic from the GABAergic, glutamatergic and dopaminergic neurons in the same few millimeters ⁴
• Positron emission tomography of the serotonin transporterA radiotracer that binds the pump which recycles serotonin, used to measure how much of it a drug occupies. At high therapeutic doses of venlafaxine, sertraline or citalopram, mean striatal occupancy was about 85 percent, leaving roughly 15 percent unoccupied ⁷.To set antidepressant doses against a measurable target, and to compare patients with controlsReproducible enough that an 80 percent occupancy threshold at minimum therapeutic dose became a standard finding ⁷Measured in the striatum rather than in the raphe; transporter binding in depression has shown weak and inconsistent differences, with prior antidepressant exposure poorly controlled ¹⁶
• Positron emission tomography with an amphetamine challengeA receptor tracer displaced by the brain's own serotonin, so that the fall in binding after a releasing drug estimates release capacity. Frontal binding fell by 15 ± 14 percent in healthy controls but only 6.5 ± 20 percent in patients in a depressive episode (p = .041) ⁴⁶.To test, in living people, whether serotonin release is reduced in depressionThe first direct human measurement of release capacity in this illness, and it found a difference ⁴⁶Seventeen patients and 20 controls, differing in mean age by 12 years and overwhelmingly male; a displacement signal is an inference, not a measurement of serotonin ⁴⁶
• Single-cell and single-nucleus RNA sequencingReading the genes expressed in thousands of individual cells to sort them into types. In mouse dorsal raphe this found at least 18 neuron subtypes including 5 serotonergic ones in 39,411 cells, and separately up to fourteen subtypes within the serotonergic lineage ^3,20.To define cell types, and to find which type changes in a disease or after stressNow the standard method, and the source of the field's main reversal since 2018 ^3,19,20Subtype counts depend on clustering choices and are reported as upper bounds; almost all of this work is in mice, and a comparable human atlas of this nucleus is not yet available ^3,20

Where it connects

Sources

102 sources, numbered as they are cited. Every one was checked against PubMed or its publisher before it was cited here; the note under each says what it shows and what it does not.

1. 1
   Hornung JP. The human raphe nuclei and the serotonergic system.J Chem Neuroanat · 2003 · 26(4):331-43doi:10.1016/j.jchemneu.2003.10.002 · PMID 14729135

   The standard human review of the raphe family. Establishes the rostral complex (caudal linear, dorsal raphe, median raphe) projecting primarily to the forebrain and the caudal complex (raphe magnus, obscurus, pallidus and parts of the adjacent lateral reticular formation) projecting to brainstem nuclei and, by three parallel projections, to the dorsal, intermediate and ventral columns of the spinal cord; median and dorsal raphe give parallel and overlapping forebrain projections with structurally distinct axons; many serotonergic terminals form no classical synapse, hence volume transmission; serotonin modulates brain development. A review, not new measurements.

2. 2
   Baker KG, Halliday GM, Tork I. Cytoarchitecture of the human dorsal raphe nucleus.J Comp Neurol · 1990 · 301(2):147-61doi:10.1002/cne.903010202 · PMID 2262589

   The human anatomical baseline. Serial 50-micrometer Nissl-stained sections through the midbrain and pontine central gray of four adult humans (mean age 56 years, mean postmortem delay 3 hours). Five subnuclei (interfascicular, ventral, ventrolateral, dorsal, caudal); 235,000 +/- 15,000 neurons (about 1,200 +/- 200 per section) in 71.3 +/- 4.5 cubic millimeters, mean density 3,300 +/- 200 neurons per cubic millimeter; dorsal subnucleus largest and most populous but least dense, ventrolateral densest; four neuron shapes (round, ovoid, fusiform, triangular). A Nissl stain counts all neurons, not only serotonergic ones, and n = 4.

3. 3
   Huang KW, Ochandarena NE, Philson AC, et al. Molecular and anatomical organization of the dorsal raphe nucleus.eLife · 2019 · 8:e46464doi:10.7554/eLife.46464 · PMID 31411560

   Single-cell RNA sequencing of 39,411 mouse dorsal raphe cells with in situ hybridization, tracing and spatial correlation: at least 18 distinct neuron subtypes, of which 5 are serotonergic, with distinct molecular and anatomical properties, including a serotonergic subtype that preferentially innervates the basal ganglia. Subtype counts depend on clustering choices; mouse.

4. 4
   Wang A, Zhou H, Wang X, Liao Y, Li C, Wu K. The dorsal raphe nucleus: a critical hub in the neurocircuitry of general anesthesia.Curr Neuropharmacol · 2026doi:10.2174/011570159X490032260630163835 · PMID 42725511

   Review of the four neuronal groups in the nucleus - serotonergic, GABAergic, glutamatergic and dopaminergic - and their roles in anesthetic induction and emergence: serotonergic neurons promote cortical arousal and drive emergence, GABAergic neurons inhibit them locally and can promote or constrain arousal depending on projection target, glutamatergic neurons resist anesthetic suppression, dopaminergic neurons encode salience and modulate pain, and serotonergic neurons produce opposing sleep-wake effects depending on firing pattern. A narrative review of mostly rodent work; the clinical recommendations in it are not trial-based.

5. 5
   Sprouse JS, Aghajanian GK. Electrophysiological responses of serotoninergic dorsal raphe neurons to 5-HT1A and 5-HT1B agonists.Synapse · 1987 · 1(1):3-9doi:10.1002/syn.890010103 · PMID 3505364

   Chloral-hydrate-anesthetized rats. Intravenous 5-HT1A-selective compounds (ipsapirone, LY165163) inhibited dorsal raphe single-unit firing dose-dependently; 5-HT1B-selective compounds (mCPP, TFMPP) were weak or irregular. Microiontophoretic dose-response curves for the 5-HT1A compounds were indistinguishable from serotonin itself; in slices they hyperpolarized dorsal raphe membranes and decreased input resistance, with maximal effects exceeding serotonin's. Identifies 5-HT1A as the candidate somatodendritic autoreceptor. Rat electrophysiology; a candidate, not a proof of the human case.

6. 6
   Richardson-Jones JW, Craige CP, Guiard BP, et al. 5-HT1A autoreceptor levels determine vulnerability to stress and response to antidepressants.Neuron · 2010 · 65(1):40-52doi:10.1016/j.neuron.2009.12.003 · PMID 20152112

   Mice engineered with higher (1A-High) or lower (1A-Low) 5-HT1A autoreceptor levels in raphe nuclei without altering 5-HT1A receptors elsewhere. Autoreceptor level robustly affected raphe firing rates but not basal forebrain serotonin levels or conflict-anxiety measures; 1A-High mice showed a blunted physiological response to acute stress, increased behavioral despair and no behavioral response to antidepressant; lowering autoreceptor levels before treatment converted non-responders into responders. A mouse model of a human risk allele, not a human experiment.

7. 7
   Voineskos AN, Wilson AA, Boovariwala A, et al. Serotonin transporter occupancy of high-dose selective serotonin reuptake inhibitors during major depressive disorder measured with [11C]DASB positron emission tomography.Psychopharmacology (Berl) · 2007 · 193(4):539-45doi:10.1007/s00213-007-0806-z · PMID 17497139

   Twelve healthy subjects and 12 with major depressive disorder. After at least 4 weeks of high-dose venlafaxine, sertraline or citalopram, mean striatal serotonin transporter occupancy was about 85 percent, significantly greater than the 80 percent seen at minimum therapeutic dose; roughly 15 percent of transporter remained unoccupied. Occupancy was measured in the striatum, not in the raphe, and the baseline was the average of the healthy group rather than each patient's own scan; n = 24 in total.

8. 8
   Aghajanian GK, Foote WE, Sheard MH. Lysergic acid diethylamide: sensitive neuronal units in the midbrain raphe.Science · 1968 · 161(3842):706-8doi:10.1126/science.161.3842.706 · PMID 4874578

   Rats. Units in serotonin-rich midbrain areas stopped firing, reversibly, after parenteral d-lysergic acid diethylamide at a dose at or below the threshold for gross behavioral effects. The observation that founded the raphe-silencing theory of psychedelic action. A correlation between a drug dose and a firing rate in anesthetized rats; it does not establish that silencing causes the human experience.

9. 9
   Aghajanian GK, Graham AW, Sheard MH. Serotonin-containing neurons in brain: depression of firing by monoamine oxidase inhibitors.Science · 1970 · 169(3950):1100-2doi:10.1126/science.169.3950.1100 · PMID 5449322

   Rats. Pargyline, tranylcypromine, phenelzine and iproniazid all depressed the firing rate of single serotonin-containing midbrain raphe neurons, and the effect was impaired by prior p-chlorophenylalanine, an inhibitor of serotonin synthesis - evidence that the slowing was serotonin-dependent feedback onto the cells themselves. The mechanistic bridge from monoamine oxidase inhibitors to this nucleus; rat electrophysiology.

10. 10
    Vollenweider FX, Vollenweider-Scherpenhuyzen MF, Bäbler A, Vogel H, Hell D. Psilocybin induces schizophrenia-like psychosis in humans via a serotonin-2 agonist action.Neuroreport · 1998 · 9(17):3897-902doi:10.1097/00001756-199812010-00024 · PMID 9875725

    Healthy human volunteers. The psychotomimetic effects of psilocybin were blocked dose-dependently by the 5-HT2A antagonist ketanserin or by risperidone and increased by the dopamine antagonist haloperidol - the first human evidence that the psychedelic effect runs through 5-HT2A receptor activation independently of dopamine. The receptors in question are predominantly cortical; the study says nothing about the raphe.

11. 11
    Vargas MV, Dunlap LE, Dong C, et al. Psychedelics promote neuroplasticity through the activation of intracellular 5-HT2A receptors.Science · 2023 · 379(6633):700-706doi:10.1126/science.adf0435 · PMID 36795823

    Molecular and genetic work in neurons and mice showing that intracellular, rather than surface, 5-HT2A receptors mediate the plasticity-promoting properties of psychedelics, which the authors offer as the explanation for why serotonin itself does not engage the same mechanism, and which raises the possibility that serotonin is not the endogenous ligand for the intracellular pool. In vitro and mouse; the link from dendritic spines to clinical antidepressant effect is a hypothesis, not a result.

12. 12
    McGinty DJ, Harper RM. Dorsal raphe neurons: depression of firing during sleep in cats.Brain Res · 1976 · 101(3):569-75doi:10.1016/0006-8993(76)90480-7 · PMID 1244990

    The reversal, in its title. Unit recording in cats showed dorsal raphe firing falling, not rising, as the animal went to sleep - the opposite of what a sleep-generating nucleus should do. The PubMed record carries no abstract, so the title, species, year and claim are what it fixes.

13. 13
    Oikonomou G, Altermatt M, Zhang RW, et al. The serotonergic raphe promote sleep in zebrafish and mice.Neuron · 2019 · 103(4):686-701.e8doi:10.1016/j.neuron.2019.05.038 · PMID 31248729

    The re-reversal. In zebrafish, genetic ablation of raphe serotonin production reduced sleep, sleep depth and the homeostatic response to sleep deprivation; pharmacological inhibition or ablation reduced sleep and optogenetic stimulation increased it. In mice, ablation increased wakefulness and impaired the homeostatic response, while tonic optogenetic stimulation at a rate similar to baseline activity induced sleep and burst stimulation induced wakefulness. The firing-pattern dependence is the reconciling observation; both species are animal models, and no equivalent human experiment exists.

14. 14
    Weiller C, May A, Limmroth V, Jüptner M, Kaube H, Schayck RV, Coenen HH, Diener HC. Brain stem activation in spontaneous human migraine attacks.Nat Med · 1995 · 1(7):658-60doi:10.1038/nm0795-658 · PMID 7585147

    The origin of the migraine generator claim. Positron emission tomography of regional cerebral blood flow during spontaneous migraine attacks found increased flow in cingulate, auditory and visual association cortices and in the brain stem, and only the brain stem activation persisted after sumatriptan had abolished headache and phono- and photophobia. The abstract names the brain stem, not the dorsal raphe, and gives no sample size; persistence after treatment is consistent with a driver, with an unrelated ongoing process, or with a drug effect, and does not distinguish them.

15. 15
    Schulte LH, May A. The migraine generator revisited: continuous scanning of the migraine cycle over 30 days and three spontaneous attacks.Brain · 2016 · 139(Pt 7):1987-93doi:10.1093/brain/aww097 · PMID 27190019

    The reversal, by the group that coined the term. One patient imaged every morning for 30 days across three complete untreated attacks. Hypothalamic response to trigeminal nociceptive stimulation changed in the 24 hours before pain onset, and the hypothalamus showed altered functional coupling with the spinal trigeminal nuclei and with the dorsal rostral pons - the region called the migraine generator - preictally and during pain. The authors conclude the driver may be hypothalamo-brainstem connectivity rather than a brainstem generator. A single patient, densely sampled: strong within-subject design, no between-subject generalization.

16. 16
    Moncrieff J, Cooper RE, Stockmann T, Amendola S, Hengartner MP, Horowitz MA. The serotonin theory of depression: a systematic umbrella review of the evidence.Mol Psychiatry · 2022 · 28(8):3243-3256doi:10.1038/s41380-022-01661-0 · PMID 35854107

    Umbrella review of 17 studies, registered as PROSPERO CRD42020207203, searched to December 2020. Two meta-analyses of the serotonin metabolite 5-HIAA showed no association with depression (largest n = 1002); a meta-analysis of cohort studies of plasma serotonin showed no relationship but did show lowered serotonin with antidepressant use (n = 1869); serotonin receptor binding (largest n = 561) and transporter binding (largest n = 1845) showed weak and inconsistent evidence of reduced binding, with prior antidepressant use not reliably excluded; tryptophan depletion showed no effect in most healthy volunteers (n = 566) with weak evidence in those with a family history (n = 75); a genetic association study (n = 115,257) and a collaborative meta-analysis (n = 43,165) showed no transporter-gene association or gene-by-stress interaction. The conclusion is about a specific claim - that depression is caused by lowered serotonin - and the receptor subtype examined is indexed as 5-HT1A though subscripts are stripped from the abstract text.

17. 17
    Jauhar S, Arnone D, Baldwin DS, et al. A leaky umbrella has little value: evidence clearly indicates the serotonin system is implicated in depression.Mol Psychiatry · 2023 · 28(8):3149-3152doi:10.1038/s41380-023-02095-y · PMID 37322065

    The principal published rebuttal, signed by 35 authors. It argues the umbrella review's conclusion is overstated because of methodological weaknesses in the review process, selective reporting, over-simplification and errors in interpreting neuropsychopharmacological findings, and it works through tryptophan depletion and serotonergic molecular imaging as the two most relevant literatures. Commercial interests are declared in the affiliations of several signatories, including one author affiliated with H. Lundbeck A/S and one with Intra-Cellular Therapies; readers should weigh that, as they should weigh that the original review's authors have long-standing public positions against antidepressant prescribing.

18. 18
    Moncrieff J, Cooper RE, Stockmann T, Amendola S, Hengartner MP, Horowitz MA. The serotonin theory of depression: a systematic umbrella review of the evidence.Focus (Am Psychiatr Publ) · 2026 · 24(2):236-251doi:10.1176/appi.focus.26024112 · PMID 42538881

    The 2022 umbrella review reprinted in the American Psychiatric Association's continuing-education journal in 2026, with a note that it appeared originally in Molecular Psychiatry 2023;28:3243-3256. Cited as evidence of the paper's standing and continued circulation in mainstream psychiatric education as of 2026, not as new data.

19. 19
    Ren J, Friedmann D, Xiong J, et al. Anatomically defined and functionally distinct dorsal raphe serotonin sub-systems.Cell · 2018 · 175(2):472-487.e20doi:10.1016/j.cell.2018.07.043 · PMID 30146164

    The paper that broke the nucleus into parts. Viral-genetic tracing in mice: subcortical- and cortical-projecting serotonin neurons occupy distinct cell-body territories within the dorsal raphe and differ in co-expression of a vesicular glutamate transporter; amygdala- and frontal-cortex-projecting serotonin neurons have largely complementary whole-brain collateralization, receive biased presynaptic input and respond oppositely to aversive stimuli. Gain- and loss-of-function suggest amygdala-projecting neurons promote anxiety-like behavior and frontal-cortex-projecting neurons promote active coping. Mouse; behavioral readouts are rodent assays, not depression or anxiety as diagnosed in people.

20. 20
    Okaty BW, Sturrock N, Escobedo Lozoya Y, et al. A single-cell transcriptomic and anatomic atlas of mouse dorsal raphe Pet1 neurons.eLife · 2020 · 9:e55523doi:10.7554/eLife.55523 · PMID 32568072

    High-throughput and subdomain-targeted single-cell transcriptomics with intersectional genetics in mouse: up to fourteen subtypes of dorsal raphe neurons of the predominantly serotonergic Pet1 lineage, many with biased cell-body distributions across the nucleus, and the most molecularly distinct subtype shown to have unique efferent projections and electrophysiological properties. Mouse; the subtype number is an upper bound from clustering, and the abstract text as indexed has gene symbols stripped.

21. 21
    Warden MR, Selimbeyoglu A, Mirzabekov JJ, et al. A prefrontal cortex-brainstem neuronal projection that controls response to behavioural challenge.Nature · 2012 · 492(7429):428-32doi:10.1038/nature11617 · PMID 23160494

    Rats. Direct optogenetic activation of medial prefrontal principal neurons had no detectable causal effect on active response to behavioral challenge, but selective activation of the subset of those neurons projecting to the dorsal raphe produced a profound, rapid and reversible shift toward the active behavioral state. The wiring, not the region, carried the effect. A rodent forced-challenge assay, not a model of human mood.

22. 22
    Marcinkiewcz CA, Mazzone CM, D'Agostino G, et al. Serotonin engages an anxiety and fear-promoting circuit in the extended amygdala.Nature · 2016 · 537(7618):97-101doi:10.1038/nature19318 · PMID 27556938

    Mice. Dorsal raphe serotonin enhanced fear and anxiety and activated a subpopulation of corticotropin-releasing factor neurons in the bed nucleus of the stria terminalis; that inhibitory microcircuit silenced anxiolytic outputs to ventral tegmental area and lateral hypothalamus, and the same circuit mediated aversive behavior after acute selective serotonin reuptake inhibitor exposure, with a corticotropin-releasing-factor type 1 receptor antagonist sufficient to prevent the acute enhancement of aversive learning. The serotonin receptor subtype named in the paper is lost to subscript stripping in the indexed abstract and is therefore not stated here. A mechanistic account in mice of a clinical observation in people.

23. 23
    Zhang Q, Xie L, Zhu C, et al. 5-HT2C receptors in the glutamatergic neurons of dorsal raphe nucleus orchestrate the comorbidity of pain and anxiety in mice.BMC Med · 2026 · 24(1):84doi:10.1186/s12916-026-04620-6 · PMID 41530758

    Complete Freund's adjuvant model of pain with anxiety in mice. Glutamatergic - not serotonergic - dorsal raphe neurons were more active in comorbid animals; chemogenetic activation reproduced the phenotype in naive mice and inhibition reversed it; 5-HT2C receptor protein was upregulated in the nucleus, and agonism or overexpression worsened while antagonism or knockdown attenuated the phenotype. Mouse; an inflammatory pain model, not clinical chronic pain.

24. 24
    Baker KG, Halliday GM, Halasz P, Hornung JP, Geffen LB, Cotton RG, Törk I. Cytoarchitecture of serotonin-synthesizing neurons in the pontine tegmentum of the human brain.Synapse · 1991 · 7(4):301-20doi:10.1002/syn.890070407 · PMID 2042112

    Immunohistochemistry against the serotonin-synthesizing enzyme (PH8 antibody) with morphometry in adult human pontine tegmentum: about 125,000 immunoreactive neurons, of which 64,400 in the median raphe nucleus, 30,700 in the oral pontine reticular nucleus and 29,000 in the supralemniscal B9 group. Notes that humans and non-human primates have far more pontine serotonin neurons than rats or cats, with different morphology and spatial arrangement. These are the pontine groups adjacent to the dorsal raphe, not the dorsal raphe itself; the paper gives no dorsal raphe serotonin-neuron count.

25. 25
    Braden K, Trinagel A, Acevedo E, et al. Dorsal raphe nucleus enkephalin peptide modulates behavioral preference.Neuropsychopharmacology · 2026doi:10.1038/s41386-026-02456-5 · PMID 42225850

    CRISPR-Cas9 knockdown of preproenkephalin in the dorsal raphe of Penk-Cre mice increased inflammation-induced mechanical sensitivity and odor avoidance and reduced sucrose preference and engagement with a novel social stimulus. Twelve-gene in situ hybridization showed the enkephalin cells are largely separate from the serotonin cells and sit instead on glutamatergic and GABAergic neurons, and knocking the peptide down in either of those subtypes alone did not reproduce the effect. Mouse; direct evidence that an opioid peptide population in this nucleus acts outside the serotonin system.

26. 26
    Fernandez SP, Cauli B, Cabezas C, Muzerelle A, Poncer JC, Gaspar P. Multiscale single-cell analysis reveals unique phenotypes of raphe 5-HT neurons projecting to the forebrain.Brain Struct Funct · 2016 · 221(8):4007-4025doi:10.1007/s00429-015-1142-4 · PMID 26608830

    Retrograde tracing with slice electrophysiology and single-cell RT-PCR in Pet1-EGFP mice. Serotonin neurons projecting to medial prefrontal cortex, basolateral amygdala and dorsal hippocampus form largely non-overlapping populations; amygdala-projecting cells have characteristic excitability and membrane properties; the peptide galanin was restricted to the prefrontal-projecting population; an atypical low-excitability subtype with long firing delays preferentially expressed vesicular glutamate transporter type 3. Mouse; small single-cell samples.

27. 27
    Miyazaki K, Miyazaki KW, Sivori G, Yamanaka A, Tanaka KF, Doya K. Serotonergic projections to the orbitofrontal and medial prefrontal cortices differentially modulate waiting for future rewards.Sci Adv · 2020 · 6(48):eabc7246doi:10.1126/sciadv.abc7246 · PMID 33246957

    Terminal-level optogenetics in mice: stimulating serotonin terminals in orbitofrontal cortex promoted waiting nearly as strongly as stimulating the dorsal raphe itself; stimulation in nucleus accumbens did not; stimulation in medial prefrontal cortex promoted waiting only when reward timing was uncertain. Evidence that the behavioral effect is carried by particular projections rather than by the nucleus as a whole. Mouse.

28. 28
    Cohen JY, Amoroso MW, Uchida N. Serotonergic neurons signal reward and punishment on multiple timescales.eLife · 2015 · 4:e06346doi:10.7554/eLife.06346 · PMID 25714923

    Mice, with serotonergic neurons identified by optogenetic tagging (channelrhodopsin-2 and responses to light). Three findings: a large fraction modulated tonic firing rates over minutes across reward versus punishment blocks; most were phasically excited by punishments; a subset was phasically excited by reward-predicting cues. Dopaminergic neurons showed no firing-rate change across blocks. Tagged recordings are the strongest form of this evidence, but the task is a rodent block design.

29. 29
    Paterson DS, Trachtenberg FL, Thompson EG, et al. Multiple serotonergic brainstem abnormalities in sudden infant death syndrome.JAMA · 2006 · 296(17):2124-32doi:10.1001/jama.296.17.2124 · PMID 17077377

    Frozen medullae from infants dying of sudden infant death syndrome and controls. Cases had a higher serotonin neuron count (148.04 +/- 51.96 versus 72.56 +/- 52.36 cells, p < .001) and density, lower 5-HT1A receptor binding density in all nine nuclei examined (p <= .01), and a lower ratio of transporter binding to neuron count (0.70 +/- 0.33 versus 1.93 +/- 1.25 fmol/mg, p = .001); male cases had lower 5-HT1A binding in the raphe obscurus than female cases (16.2 +/- 2.0 versus 29.6 +/- 16.5 fmol/mg, p = .04). Crucially for this page, the pathology is in the medullary - caudal - raphe, not the dorsal raphe; control numbers were as low as 6 to 10 per analysis.

30. 30
    Walther DJ, Peter JU, Bashammakh S, Hörtnagl H, Voits M, Fink H, Bader M. Synthesis of serotonin by a second tryptophan hydroxylase isoform.Science · 2003 · 299(5603):76doi:10.1126/science.1078197 · PMID 12511643

    The one-page report of a second isoform of tryptophan hydroxylase, the rate-limiting enzyme of serotonin synthesis. The PubMed record carries no abstract; the title and the indexing (tryptophan hydroxylase, isoenzymes, brain, duodenum, mouse and rat) are what fix the claim. The brain/gut division of labor between the two isoforms is the standard reading of this work, not a sentence quoted from it.

31. 31
    Taylor MJ, Freemantle N, Geddes JR, Bhagwagar Z. Early onset of selective serotonin reuptake inhibitor antidepressant action: systematic review and meta-analysis.Arch Gen Psychiatry · 2006 · 63(11):1217-23doi:10.1001/archpsyc.63.11.1217 · PMID 17088502

    Fifty trials identified from more than 500 citations; the primary analysis pooled 28 randomized controlled trials of selective serotonin reuptake inhibitors versus placebo in adults with unipolar depression (n = 5872), excluding trials limited to people over 65 or to specific comorbidities. A model of early treatment response fit the data better than alternatives; relative risk of a 50 percent reduction in Hamilton score by one week was 1.64 (95 percent CI 1.2-2.25), and improvement continued at a decreasing rate for at least 6 weeks. Group-level pooled rating-scale data extracted by a single reviewer, so this bounds the delay rather than abolishing it.

32. 32
    Artigas F, Perez V, Alvarez E. Pindolol induces a rapid improvement of depressed patients treated with serotonin reuptake inhibitors.Arch Gen Psychiatry · 1994 · 51(3):248-51doi:10.1001/archpsyc.1994.03950030084009 · PMID 8122960

    The original clinical application of the autoreceptor hypothesis. The PubMed record is indexed as a letter and case reports with no abstract, so what it fixes is the date, the authors, the drugs (pindolol with paroxetine) and the claim of rapid improvement. Cited for the origin of the strategy, not for an effect size.

33. 33
    Portella MJ, de Diego-Adelino J, Ballesteros J, et al. Can we really accelerate and enhance the selective serotonin reuptake inhibitor antidepressant effect? A randomized clinical trial and a meta-analysis of pindolol in nonresistant depression.J Clin Psychiatry · 2011 · 72(7):962-9doi:10.4088/JCP.09m05827blu · PMID 21034693

    Thirty outpatients with DSM-IV major depressive disorder randomized double-blind to citalopram plus pindolol 5 mg three times daily or citalopram plus placebo for 6 weeks (NCT00931775). Significant group-by-time interaction (p = .01); trend for sustained response (odds ratio 2.09, 95 percent CI 0.914-4.780, p = .08); increased likelihood of sustaining remission (odds ratio 5.00, 1.191-20.989, p = .03); median survival time until first response 22 days versus 30 (p = .03). An accompanying meta-analysis of pindolol augmentation in nonresistant depression favored pindolol at 2 weeks (relative risk 1.68, 1.18-2.39, p = .004) and at 4-6 weeks (1.11, 1.02-1.20, p = .02). A 30-patient trial by investigators who had argued for the mechanism; pindolol also has beta-adrenergic actions.

34. 34
    Ren J, Isakova A, Friedmann D, et al. Single-cell transcriptomes and whole-brain projections of serotonin neurons in the mouse dorsal and median raphe nuclei.eLife · 2019 · 8:e49424doi:10.7554/eLife.49424 · PMID 31647409

    Plate-based single-cell RNA sequencing of mouse dorsal and median raphe serotonin neurons giving eleven transcriptomically distinct clusters, mapped by systematic in situ hybridization to the principal dorsal raphe, the caudal dorsal raphe or the median raphe, and differing in neuropeptides, receptors, ion channels and transcription factors. Novel intersectional viral-genetic tools gave access to subpopulations; whole-brain axonal mapping showed that dorsal raphe serotonin neurons co-expressing vesicular glutamate transporter 3 preferentially innervate the cortex while those co-expressing thyrotropin-releasing hormone innervate subcortical regions, the hypothalamus in particular; reconstruction of 50 individual dorsal raphe serotonin neurons showed diverse and segregated single-cell projection patterns. Mouse; cluster numbers depend on clustering choices, as with the other atlases.

35. 35
    Garcia-Verdugo M, Rodriguez-Martin P, Gustincich S, Lanciego JL, Murer MG, Jercog P, Moratalla R. From rodents to humans: conserved codistribution of dopaminergic with serotonergic neurons in the dorsal raphe nucleus and their molecular characterization.Neurobiol Dis · 2025 · 217:107173doi:10.1016/j.nbd.2025.107173 · PMID 41183746

    Comparative high-resolution immunohistochemistry of dorsal raphe dopaminergic neurons and their codistribution with serotonergic neurons in mouse, macaque and human brains, with positions referenced to the ponto-mesencephalic junction. Tyrosine-hydroxylase-positive neurons co-expressed aromatic L-amino acid decarboxylase in all three species, confirming a capacity for dopamine synthesis, while the dopamine transporter was detected in less than 20 percent of them, which the authors read as a reduced capacity for reuptake; calbindin and vasoactive intestinal peptide co-expression varied markedly between species. Postmortem anatomy and marker expression, not function.

36. 36
    Wang W, Wang D, Zhao D, et al. Dorsal raphe dopaminergic neurons target CaMKII neurons in dorsal bed nucleus of the stria terminalis for mediating depression-related behaviors.Transl Psychiatry · 2024 · 14(1):408doi:10.1038/s41398-024-03093-6 · PMID 39358336

    Chronic unpredictable stress in male and female mice induced depression-related behavior and reduced the spontaneous firing rates and the excitatory and inhibitory postsynaptic currents of dorsal raphe dopaminergic neurons. Chemogenetic inhibition of those neurons produced depressive phenotypes; activation reversed the stress-induced anhedonia and despair measures; the effect ran through a projection to calcium/calmodulin-dependent protein kinase II-expressing neurons of the dorsal bed nucleus of the stria terminalis, with changes in dopamine D2 receptor signaling and N-methyl-D-aspartate-receptor-mediated currents there. Mouse; rodent behavioral assays, not diagnosed depression.

37. 37
    Chalençon L, Rahn RM, Chottekalapanda RU, et al. Cellular and molecular coding of habit-driven behavior in the dorsal raphe nucleus.bioRxiv · 2026 · preprint 2026.10.01.755903doi:10.64898/2026.10.01.755903 · PMID 42856539

    A preprint, not peer reviewed, and flagged as such here. Two-photon calcium imaging in awake mice found dorsal raphe neurons integrating effort and reward at the single-neuron level; chronic stress selectively enhanced reward-related responses in habit-driven animals, deficits in reward-related activity predicted later habit formation, single-nucleus profiling found stress-driven gene expression changes in GABAergic neurons, and optogenetic inhibition of those GABAergic neurons rescued the stress effect on habit formation. Treat as preliminary.

38. 38
    Dahlstroem A, Fuxe K. Evidence for the existence of monoamine-containing neurons in the central nervous system. I. Demonstration of monoamines in the cell bodies of brain stem neurons.Acta Physiol Scand Suppl · 1964 · 232:1-55PMID 14229500

    The founding map. The PubMed record carries no abstract and no DOI; what it fixes is the date, the authors, the species (rat), the method (histochemical fluorescence, with monoamine oxidase inhibitors and reserpine among the indexed terms) and the claim that monoamines are present in the cell bodies of brainstem neurons. The lettered cell-group nomenclature used for the raphe since is conventionally traced to this monograph, but that convention is not stated in the record itself.

39. 39
    Jouvet M. Biogenic amines and the states of sleep.Science · 1969 · 163(3862):32-41doi:10.1126/science.163.3862.32 · PMID 4303225

    The review that put serotonin at the center of sleep. The PubMed record carries no abstract; the indexed terms (cats, 5-hydroxytryptophan, antimetabolites, pons, sleep, REM sleep, insomnia) fix the subject and the experimental tradition - depleting serotonin and watching sleep fail. Cited here for the position it established, not for a specific number.

40. 40
    Rybakowski J. Sixty years of neurotransmitter concepts of mood disorders.Psychiatr Pol · 2025 · 59(5):707-720doi:10.12740/PP/205845 · PMID 41493223

    Historical review marking sixty years since the neurotransmitter hypotheses of affective disorders. Records the catecholamine hypothesis as proposed in 1965 by Joseph Schildkraut and by William Bunney and John Davis (noradrenaline deficiency in depression, excess in mania, later a similar role for dopamine) and the serotonin hypothesis of depression as presented in 1969 by Alec Coppen and by Izaslaw Lapin and Grigorij Oxenkrug; traces the introduction of fluoxetine in the United States at the turn of the 1980s and 1990s and serotonin's later cultural career, and notes that ketamine's arrival pointed to glutamatergic transmission. A narrative historical article, cited for dates and attributions rather than for data.

41. 41
    Wang RY, Aghajanian GK. Inhibition of neurons in the amygdala by dorsal raphe stimulation: mediation through a direct serotonergic pathway.Brain Res · 1977 · 120(1):85-102doi:10.1016/0006-8993(77)90499-1 · PMID 137766

    Rats. Electrical stimulation of the dorsal raphe inhibited spontaneous firing in amygdala neurons at latencies compatible with unmyelinated serotonergic fibers; destroying the serotonin projection with 5,7-dihydroxytryptamine or depleting serotonin with p-chlorophenylalanine abolished the inhibition, and 5-hydroxytryptophan restored it. Intravenous lysergic acid diethylamide blocked the inhibition produced by submaximal raphe stimulation. The classical demonstration that this nucleus speaks directly to the amygdala; rat.

42. 42
    Jacobs BL, Azmitia EC. Structure and function of the brain serotonin system.Physiol Rev · 1992 · 72(1):165-229doi:10.1152/physrev.1992.72.1.165 · PMID 1731370

    The long-standing synthesis of the field, cited here as the landmark review it is. The PubMed record carries no abstract, so no specific finding is attributed to it in this page beyond its existence, date and scope.

43. 43
    Goadsby PJ, Hoskin KL. Serotonin inhibits trigeminal nucleus activity evoked by craniovascular stimulation through a 5HT1B/1D receptor: a central action in migraine?Ann Neurol · 1998 · 43(6):711-8doi:10.1002/ana.410430605 · PMID 9629840

    Anesthetized cats with the superior sagittal sinus stimulated electrically and neurons of the trigeminocervical complex recorded in the dorsal C2 spinal cord. Baseline probability of cell firing was 0.61 +/- 0.1 at a fastest-peak latency of 11.1 +/- 0.4 milliseconds; serotonin infused at 15 micrograms per kilogram per minute for 5 minutes inhibited evoked trigeminal firing from 15-20 minutes after the start of infusion for about 20 minutes, and the 5-HT1B/1D antagonist GR127935 (100 micrograms per kilogram intravenously) abolished the inhibition. Establishes a central, second-order site at which serotonin damps trigeminal traffic. Cat; infused serotonin, not endogenous raphe output.

44. 44
    Miyazaki KW, Miyazaki K, Tanaka KF, Yamanaka A, Takahashi A, Tabuchi S, Doya K. Optogenetic activation of dorsal raphe serotonin neurons enhances patience for future rewards.Curr Biol · 2014 · 24(17):2033-40doi:10.1016/j.cub.2014.07.041 · PMID 25155504

    Mice expressing channelrhodopsin selectively in serotonin neurons. Activation reduced premature cessation of waiting for conditioned cues and food rewards and, in reward-omission trials, prolonged waiting; control experiments argued against general motor inhibition or a reinforcing effect of the stimulation itself. Causal in mice for waiting behavior; says nothing directly about human impulsivity.

45. 45
    Bartova L, Lanzenberger R, Rujescu D, Kasper S. Reply to: "The serotonin theory of depression: a systematic umbrella review of the evidence".Mol Psychiatry · 2023 · 28(8):3153-3154doi:10.1038/s41380-023-02093-0 · PMID 37322062

    A second published reply, from the Medical University of Vienna. The PubMed record carries no abstract, so it is cited here only as evidence that the umbrella review drew more than one formal response in the same journal.

46. 46
    Erritzoe D, Godlewska BR, Rizzo G, et al. Brain serotonin release is reduced in patients with depression: a [11C]Cimbi-36 positron emission tomography study with a d-amphetamine challenge.Biol Psychiatry · 2023 · 93(12):1089-1098doi:10.1016/j.biopsych.2022.10.012 · PMID 36635177

    Seventeen antidepressant-free patients in a major depressive episode (3 female, 14 male, mean age 44 +/- 13 years, Hamilton score 21 +/- 4) and 20 healthy controls (3 female, 17 male, mean age 32 +/- 9 years) scanned before and 3 hours after 0.5 mg/kg oral d-amphetamine. Frontal binding potential fell significantly in controls (1.04 +/- 0.31 to 0.87 +/- 0.24, p < .001) but not in patients (0.97 +/- 0.25 to 0.92 +/- 0.22, not significant), and the change differed between groups (15 +/- 14 percent versus 6.5 +/- 20 percent, p = .041). The first direct human assessment of serotonin release capacity in depression, and it points the other way from the umbrella review - but the groups differ in age by 12 years, both are small and mostly male, and the measure is an indirect displacement signal.

47. 47
    Priestley L, Mahmoodi A, Reith WD, Khalighinejad N, Rushworth MFS. Activity in human dorsal raphe nucleus signals changes in behavioural policy.Nat Commun · 2026 · 17(1):1665doi:10.1038/s41467-026-68349-9 · PMID 41690932

    One of the few human functional imaging studies of this nucleus. In a foraging-inspired task, participants changed their policy to pursue or reject reward opportunities as a function of the average value of opportunities in the environment, and activity in the dorsal raphe - but in no other neuromodulatory nucleus examined - signaled those policy changes, while multivariate patterns in dorsal anterior cingulate and anterior insular cortex tracked relative value. Human and correlational; brainstem functional MRI is technically demanding and the signal is indirect.

48. 48
    Jouvet M. Sleep and serotonin: an unfinished story.Neuropsychopharmacology · 1999 · 21(2 Suppl):24S-27Sdoi:10.1016/S0893-133X(99)00009-3 · PMID 10432485

    Review by the author of the original serotonin-sleep hypothesis, stating the contradiction plainly: destruction of the raphe serotonin neurons or inhibition of synthesis with p-chlorophenylalanine induced a severe insomnia reversible by restoring synthesis, while the electrical activity of serotonin cell bodies and the release of serotonin are increased in waking and decreased in sleep. Proposes that serotonin released during waking initiates a cascade of genomic events in hypnogenic neurons of the preoptic area, giving a homeostatic regulation of slow-wave sleep. A narrative review and a proposal, not a measurement.

49. 49
    He J, Yao T, Guo H, et al. A dopaminergic projection from the ventral tegmental area to the dorsal raphe nucleus critically regulates propofol anesthesia in mice.Biol Direct · 2026 · 21(1)doi:10.1186/s13062-026-00736-0 · PMID 41656256

    Circuit tracing in DAT-Cre and TPH2-CreER mice showed appositions between ventral tegmental dopaminergic terminals and dorsal raphe serotonergic neurons; fiber photometry showed both suppressed under propofol; chemogenetic activation of either population hastened emergence and inhibition prolonged anesthesia; optogenetic activation of the terminals produced behavioral and electroencephalographic arousal, abolished by chemogenetic inhibition of the serotonergic neurons. Mouse; propofol specifically.

50. 50
    Miyazaki K, Miyazaki KW, Yamanaka A, Tokuda T, Tanaka KF, Doya K. Reward probability and timing uncertainty alter the effect of dorsal raphe serotonin neurons on patience.Nat Commun · 2018 · 9(1):2048doi:10.1038/s41467-018-04496-y · PMID 29858574

    The qualifier. Optogenetic activation prolonged waiting in no-reward trials at 75 percent food-reward probability but not at 50 or 25 percent, and the effect grew when reward timing was unpredictable. A Bayesian decision model fits the data if serotonin activation raises the prior probability or subjective confidence of reward. Mouse; the model is an interpretation, not a measurement.

51. 51
    Worbe Y, Savulich G, Voon V, Fernandez-Egea E, Robbins TW. Serotonin depletion induces 'waiting impulsivity' on the human four-choice serial reaction time task: cross-species translational significance.Neuropsychopharmacology · 2014 · 39(6):1519-26doi:10.1038/npp.2013.351 · PMID 24385133

    Double-blind randomized dietary tryptophan depletion in healthy volunteers, tested on a human four-choice serial reaction time task designed as a translational analogue of the rodent five-choice task, plus a reward delay-discounting questionnaire. Depletion significantly increased premature responses, mirroring the rodent findings, and the increase correlated with the Barratt motor impulsivity subscale; there was no effect on delay discounting, and accuracy and response speed improved. Healthy volunteers, a laboratory task, and an acute manipulation.

52. 52
    Schweighofer N, Bertin M, Shishida K, Okamoto Y, Tanaka SC, Yamawaki S, Doya K. Low-serotonin levels increase delayed reward discounting in humans.J Neurosci · 2008 · 28(17):4528-32doi:10.1523/JNEUROSCI.4982-07.2008 · PMID 18434531

    Double-blind study in healthy male volunteers with central serotonin manipulated by dietary tryptophan depletion and loading, performing a dynamic delayed-reward choice task analyzed with a computational learning model. The low-serotonin condition increased choices of small immediate rewards and raised the discounting rate for delayed rewards relative to the control and high-serotonin conditions, with no significant differences in the speed of learning or the variability of choice. Male volunteers, small sample, model-dependent parameter estimates.

53. 53
    Varga ZK, Golla A, Kermen F. Early life chronic stress-disrupted activity of the dorsal raphe nucleus selectively drives behavioral impairments.Commun Biol · 2026 · 9(1)doi:10.1038/s42003-026-09855-w · PMID 41862617

    In vivo two-photon calcium imaging of serotonergic dorsal raphe neurons in larval zebrafish exposed to early-life chronic unpredictable stress. Stress prevented habituation of the serotonergic system to repeated stress by shifting the balance of excitatory and inhibitory responses, most markedly in serotonergic cells co-expressing GABAergic markers; chemogenetic ablation changed startle habituation and locomotor activity but not anxiety-like behavior. Larval zebrafish; the dissociation between behaviors is the interesting part and the hardest to translate.

54. 54
    Li Y, Cao Q, Hou Z, Tang B, Shen Y. Transcranial sonography as a diagnostic tool for depressive disorders.J Ultrasound Med · 2023 · 42(3):687-699doi:10.1002/jum.16081 · PMID 36047031

    Meta-analysis of the diagnostic accuracy of transcranial sonography of brainstem raphe echogenicity in depression, searching PubMed, EMBASE, the Cochrane Library and Web of Science from inception to December 2021 with QUADAS-2 quality assessment; 12 studies, 809 patients, DSM-IV criteria. Pooled sensitivity 0.66 (95 percent CI 0.61-0.71), specificity 0.84 (0.80-0.87), positive likelihood ratio 3.84, negative likelihood ratio 0.41, diagnostic odds ratio 11.45 (5.57-23.02), area under the summary receiver operating characteristic curve 0.86 (0.83-0.89); meta-regression and subgroup analyses found no source of heterogeneity. The authors' conclusion that the method has high potential for diagnosing depression is stronger than a sensitivity of 0.66 supports; the histological basis of reduced echogenicity is unknown and the reading is operator-dependent.

55. 55
    Swajkoski AR, Mayer DJ, Johnson JH. Blockade by naltrexone of analgesia produced by stimulation of the dorsal raphe nucleus.Pharmacol Biochem Behav · 1981 · 15(3):419-23doi:10.1016/0091-3057(81)90272-0 · PMID 7291246

    Bipolar stainless steel electrodes in the dorsal raphe nucleus or ventral periaqueductal gray of male rats; analgesia from electrical stimulation measured by tail flick before and 20 minutes after saline or naltrexone. Intraperitoneal saline did not alter the response; naltrexone reduced analgesia by a mean of 79 percent intravenously (3.7 mg/kg) and by 26, 52, 81 and 83 percent intraperitoneally at 0.3, 1.0, 3.0 and 10 mg/kg. The authors conclude that opiate mechanisms participate and that, under these conditions, only opiate mechanisms are involved. Rat; electrical stimulation spreads beyond the intended nucleus, which is the standing limitation of this method.

56. 56
    Hosobuchi Y, Adams JE, Linchitz R. Pain relief by electrical stimulation of the central gray matter in humans and its reversal by naloxone.Science · 1977 · 197(4299):183-6doi:10.1126/science.301658 · PMID 301658

    Six patients with intractable pain and permanently implanted periventricular and periaqueductal gray electrodes. The level of stimulation sufficient for relief appeared not to alter the acute pain threshold; indiscriminate repetitive stimulation produced tolerance to both stimulation-produced relief and narcotic analgesia, reversible by abstinence from stimulation; naloxone reversed stimulation-produced relief in five of the six patients. Six patients, open-label, and the target is the surrounding central gray rather than the dorsal raphe itself.

57. 57
    Frias AT, Hernandes PM, Zangrossi H. Blockade of acid-sensing ion channels 1a in the mice dorsal raphe nucleus inhibits panic-like responses: role of 5-HT1A receptors in the dorsal periaqueductal gray.Neurosci Lett · 2025 · 867:138389doi:10.1016/j.neulet.2025.138389 · PMID 40992456

    C57BL/6 mice given psalmotoxin-1 (12.5 or 25 nanograms in 50 nanoliters), a selective blocker of the acid-sensing ion channel ASIC1a, into the lateral wings of the dorsal raphe, then exposed to 20 percent carbon dioxide. The blockade reduced escape behavior without changing baseline locomotion, and the effect was abolished by the 5-HT1A antagonist WAY100635 (0.74 nanomoles in 50 nanoliters) into the dorsal periaqueductal gray. A rodent panic model; carbon-dioxide-evoked escape is an analogue of panic, not panic.

58. 58
    Zhang H, Li L, Sun B, et al. A GABAergic pathway from dorsal raphe nucleus to paraventricular thalamic nucleus modulates incision-related pain behaviour in mice.Brain Commun · 2026 · 8(2):fcag046doi:10.1093/braincomms/fcag046 · PMID 41777490

    Mouse paw-incision model with increased neuronal excitability in the paraventricular thalamic nucleus. Multiple tracing methods identified an inhibitory ascending pathway from dorsal raphe to that nucleus, with activity enhanced by nociceptive stimuli; inhibiting the pathway induced nociceptive sensitivity in normal mice and activating it alleviated incision pain; GABA-A, but not GABA-B, receptor antagonists injected into the thalamic nucleus blocked pathway activation and were analgesic. Mouse; one surgical pain model.

59. 59
    Song L, Ge T, Cui M, et al. An NTS-expressing inhibitory circuit from interpeduncular nucleus to dorsal raphe nucleus controls pain and comorbid depression in male mice.Cell Rep · 2025 · 45(1):116719doi:10.1016/j.celrep.2025.116719 · PMID 41411126

    Complete Freund's adjuvant mice showed increased activity of neurotensin-expressing interpeduncular neurons. Optogenetic activation of those neurons or of their dorsal raphe projections induced pain- and depressive-like behaviors in naive mice; inhibition reversed both in inflamed mice. Activation evoked co-release of GABA and neurotensin in the dorsal raphe; intra-dorsal-raphe GABA-A receptor antagonism prevented the behavioral effects, while antagonizing dorsal raphe neurotensin receptors induced pain but did not affect the depressive-like behavior. Male mice only, as the title states.

60. 60
    Wang L, Wang G, Zhao W, et al. Dopaminergic neurons in the dorsal raphe nucleus are involved in the regulation of memory impairment in chronic pain.Brain Res · 2026 · 1882:150280doi:10.1016/j.brainres.2026.150280 · PMID 41864256

    Chronic constriction injury in male C57BL/6J and DAT-Cre mice. Injured mice showed memory impairment on novel object recognition and fear conditioning with reduced dorsal raphe dopaminergic activity on c-Fos staining and whole-cell recording; chemogenetic or optogenetic activation of those neurons, or optogenetic stimulation of their terminals in the bed nucleus of the stria terminalis, rescued memory performance. Male mice, one neuropathic model, behavioral memory assays.

61. 61
    Choi JE, Kim YK, Sung Y, et al. Oxytocin signaling bidirectionally modulates serotonergic and dopaminergic neurons in the dorsal raphe nucleus.eNeuro · 2026doi:10.1523/ENEURO.0182-26.2026 · PMID 42888041

    Immunohistochemistry, tracing, chemogenetics and ex vivo electrophysiology, primarily in male mice. Dual-eGRASP labeling visualized oxytocin projections from the paraventricular nucleus onto both serotonergic and dopaminergic dorsal raphe neurons; blocking oxytocin receptors in the nucleus reduced social interaction without affecting locomotion; oxytocin inhibited the serotonergic neurons while activating the dopaminergic ones. Mouse, and almost entirely male.

62. 62
    Amin FM, Asghar MS, Hougaard A, et al. Magnetic resonance angiography of intracranial and extracranial arteries in patients with spontaneous migraine without aura: a cross-sectional study.Lancet Neurol · 2013 · 12(5):454-61doi:10.1016/S1474-4422(13)70067-X · PMID 23578775

    Cross-sectional magnetic resonance angiography during spontaneous unilateral migraine attacks (NCT01471314); 78 patients recruited, 19 women scanned during an attack and analyzed. No significant extracranial dilatation on the pain side versus attack-free days (external carotid 1.2 percent, p = 0.985; superficial temporal 3.6 percent, p = 0.532; middle meningeal 1.7 percent, p = 0.341; cervical internal carotid 2.3 percent, p = 0.093); the intracranial arteries dilated (middle cerebral 13.0 percent, p = 0.001; cerebral internal carotid 11.5 percent, p = 0.0004; cavernous internal carotid 11.4 percent, p = 0.001; basilar 1.6 percent, not significant). The authors conclude that future research should focus on peripheral and central pain pathways rather than simple arterial dilatation. Nineteen patients, all female, single attacks.

63. 63
    Bahra A, Matharu MS, Buchel C, Frackowiak RS, Goadsby PJ. Brainstem activation specific to migraine headache.Lancet · 2001 · 357(9261):1016-7doi:10.1016/s0140-6736(00)04250-1 · PMID 11293599

    A single patient with both cluster headache and migraine, scanned by positron emission tomography after glyceryl trinitrate; he developed a typical migraine without aura with activation in the dorsal rostral brainstem and no hypothalamic activation. A case report, and the strongest claim it supports is that the two headache types can be distinguished by imaging.

64. 64
    Afridi SK, Matharu MS, Lee L, Kaube H, Friston KJ, Frackowiak RSJ, Goadsby PJ. A PET study exploring the laterality of brainstem activation in migraine using glyceryl trinitrate.Brain · 2005 · 128(Pt 4):932-9doi:10.1093/brain/awh416 · PMID 15705611

    Twenty-four migraineurs and eight healthy controls, attacks induced with glyceryl trinitrate. Significant activation in the dorsal lateral pons (P < 0.05 after small volume correction) in the migraine state versus the pain-free state, ipsilateral to the side of headache in the right-sided and left-sided groups and bilateral with a left preponderance in the bilateral group; the activation again persisted after sumatriptan. The lateralization matters for this page: a lateralized pontine focus is not a midline raphe structure.

65. 65
    Maniyar FH, Sprenger T, Monteith T, Schankin C, Goadsby PJ. Brain activations in the premonitory phase of nitroglycerin-triggered migraine attacks.Brain · 2014 · 137(Pt 1):232-41doi:10.1093/brain/awt320 · PMID 24277718

    Eight patients with episodic migraine without aura who habitually had premonitory symptoms, scanned by positron emission tomography at baseline, in the painless premonitory phase and during headache. Premonitory activations in posterolateral hypothalamus, midbrain tegmental area, periaqueductal gray, dorsal pons and several cortical areas. Small n and chemically triggered attacks; the hypothalamic finding is what shifted attention away from the pons.

66. 66
    Schulte LH, Menz MM, Haaker J, May A. The migraineur's brain networks: continuous resting state fMRI over 30 days.Cephalalgia · 2020 · 40(14):1614-1621doi:10.1177/0333102420951465 · PMID 32830513

    Nine episodic migraineurs scanned daily for at least 30 consecutive days, 282 scanned days and 15 spontaneous attacks. Preictally, right nucleus accumbens connectivity rose with left amygdala, hippocampus and parahippocampal gyrus, and with the dorsal rostral pons; ictally the dorsal pons was more strongly coupled to the hypothalamic area. Resting-state connectivity is correlational and does not establish which region drives which.

67. 67
    Villalón CM, MaassenVanDenBrink A. The role of 5-hydroxytryptamine in the pathophysiology of migraine and its relevance to the design of novel treatments.Mini Rev Med Chem · 2017 · 17(11):928-938doi:10.2174/1389557516666160728121050 · PMID 27465216

    Review of serotonin pharmacology in migraine: triptans produce selective cranial extracerebral vasoconstriction through 5-HT1B receptors and inhibit trigeminovascular responses through 5-HT1D and 5-HT1F receptors; a number of patients respond poorly and the drugs are contraindicated with cardiovascular disease; newer approaches include selective 5-HT1F agonists and antibodies and antagonists against calcitonin gene-related peptide. A narrative review by pharmacologists, not a trial.

68. 68
    Deleu D, Hanssens Y. Current and emerging second-generation triptans in acute migraine therapy: a comparative review.J Clin Pharmacol · 2000 · 40(7):687-700doi:10.1177/00912700022009431 · PMID 10883409

    Comparative review of almotriptan, eletriptan, frovatriptan, naratriptan, rizatriptan and zolmitriptan against sumatriptan, locating the 5-HT1B receptors in meningeal arteries and the 5-HT1D receptors on trigeminovascular fiber endings, with sterile neurogenic inflammation as the proposed target. A review of the era in which the drug class was built; the neurogenic-inflammation model has since been qualified.

69. 69
    Donnet A, Redon S. Cyclic vomiting syndrome in children.Curr Pain Headache Rep · 2018 · 22(4):30doi:10.1007/s11916-018-0684-6 · PMID 29556858

    Review placing cyclic vomiting syndrome among the episodic syndromes associated with migraine, with stereotyped attacks and predictable, sometimes circadian or circannual, periodicity. Reports that functional magnetic resonance imaging in both cyclic vomiting syndrome and migraine has shown diminished insular connectivity with the sensorimotor network, suggesting shared pathophysiology; notes injectable or intranasal sumatriptan is often effective and that controlled trials in children are missing. A review; the shared-imaging claim is suggestive, not decisive.

70. 70
    Venkatesan T, Levinthal DJ, Tarbell SE, et al. Guidelines on management of cyclic vomiting syndrome in adults by the American Neurogastroenterology and Motility Society and the Cyclic Vomiting Syndrome Association.Neurogastroenterol Motil · 2019 · 31 Suppl 2:e13604doi:10.1111/nmo.13604 · PMID 31241819

    GRADE-based guideline. Strong recommendation for a tricyclic antidepressant such as amitriptyline as first-line prophylaxis in moderate-to-severe adult cyclic vomiting syndrome, with topiramate or aprepitant as alternates and zonisamide, levetiracetam and mitochondrial supplements (coenzyme Q10, L-carnitine, riboflavin) conditionally recommended; for acute attacks, conditional recommendation for serotonin antagonists such as ondansetron and/or triptans such as sumatriptan, or aprepitant; screening for comorbid migraine, anxiety, depression, autonomic dysfunction, sleep disorders and substance use. Conditional recommendations rest on weak evidence, as the committee states.

71. 71
    Gui S, Patel N, Issenman R, Kam AJ. Acute management of pediatric cyclic vomiting syndrome: a systematic review.J Pediatr · 2019 · 214:158-164.e4doi:10.1016/j.jpeds.2019.06.057 · PMID 31540764

    Systematic review of 799 studies published from December 1954 to February 2018, 84 taken to full review and 54 included. Ondansetron had the most quantitative and qualitative evidence as rescue therapy; sumatriptan and aprepitant were candidates for abortive therapy. The authors state plainly that the quantitative evidence comes from six case series and that controlled prospective cohort studies and randomized trials are still needed.

72. 72
    Karrento K, Rosen JM, Tarbell SE, et al. North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition 2025 guidelines for management of cyclic vomiting syndrome in children.J Pediatr Gastroenterol Nutr · 2025 · 80(6):1028-1061doi:10.1002/jpn3.70020 · PMID 40223700

    GRADE-based guideline replacing prior expert consensus, from a multidisciplinary panel with a patient representative, using Evidence-to-Decision frameworks and indirect evidence from pediatric migraine where direct evidence was lacking (PROSPERO CRD42022310108). Sixteen recommendations, split into abortive and prophylactic. One strong recommendation: anti-migraine agents to abort episodes in patients with a personal or family history of migraine. Conditional: oral and intravenous 5-HT3 and neurokinin-1 receptor antagonists, and early presentation when intravenous treatment is needed; for prevention, trigger avoidance, supplements, biobehavioral and neuromodulation interventions, beta-blockers, neurokinin-1 and 5-HT2A antagonists and tricyclic antidepressants; anticonvulsants only in refractory disease, with cautions about side effects. The conditional recommendations rest on low-certainty evidence, much of it borrowed from migraine.

73. 73
    Karrento K, Rosen JM, Gelfand AA, et al. North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition 2025 guidelines on the diagnosis of cyclic vomiting syndrome in children.J Pediatr Gastroenterol Nutr · 2025 · 81(5):1346-1359doi:10.1002/jpn3.70193 · PMID 40836301

    Companion GRADE-based diagnostic guideline, with a Delphi consensus where the literature was insufficient. Recommends screening serum and urine laboratory testing and an upper gastrointestinal radiographic series for all patients with symptoms suggestive of cyclic vomiting syndrome, and judges other diagnostic screening tests low-yield and not cost-effective; sets diagnostic criteria by attack frequency, duration and associated symptoms; treats cannabinoid hyperemesis syndrome as a related but separate entity. No diagnostic biomarker exists, so the criteria remain symptom-based.

74. 74
    Zarkadas E, Zhang H, Cai W, et al. The binding of palonosetron and other antiemetic drugs to the serotonin 5-HT3 receptor.Structure · 2020 · 28(10):1131-1140.e4doi:10.1016/j.str.2020.07.004 · PMID 32726573

    Cryo-electron microscopy structure of the palonosetron-bound 5-HT3 receptor with molecular dynamics covering granisetron, dolasetron, ondansetron and cilansetron - the whole clinical setron family. Structural and computational work in a purified receptor, not in brain tissue; it fixes what these drugs bind, not where in the nervous system the binding matters.

75. 75
    Al-Ramadan A, Kidess GG, Bahar AR, Bahar Y, Hazique M, Alraies MC. Morbidity and mortality of ondansetron in patients with non-congenital long QT syndrome: a review article.Cardiovasc Drugs Ther · 2025 · 40(3):1117-1124doi:10.1007/s10557-025-07766-2 · PMID 40856942

    Review of the 5-HT3 receptor as a ligand-gated ion channel regulating vomiting and gastrointestinal motility, and of ondansetron's prolongation of the QT interval with the attendant risk of torsades de pointes, the resulting Food and Drug Administration dosing recommendations and the patients who need monitoring. A narrative safety review, not a cohort study.

76. 76
    Lucot JB, Crampton GH. 8-OH-DPAT suppresses vomiting in the cat elicited by motion, cisplatin or xylazine.Pharmacol Biochem Behav · 1989 · 33(3):627-31doi:10.1016/0091-3057(89)90399-7 · PMID 2531422

    Cats pretreated with the 5-HT1A agonist 8-OH-DPAT and then challenged with motion, xylazine or cisplatin vomited less; the authors attribute this to stimulation of postsynaptic serotonin-1A receptors at a convergent structure at or near the vomiting center, and a supplementary experiment argued the effect was not anxiolysis, since lorazepam suppressed motion sickness only at an ataxic dose and did not suppress xylazine-induced emesis. Cat pharmacology; the receptor location is inferred from systemic dosing, not mapped.

77. 77
    Rock EM, Bolognini D, Limebeer CL, et al. Cannabidiol, a non-psychotropic component of cannabis, attenuates vomiting and nausea-like behaviour via indirect agonism of 5-HT1A somatodendritic autoreceptors in the dorsal raphe nucleus.Br J Pharmacol · 2012 · 165(8):2620-34doi:10.1111/j.1476-5381.2011.01621.x · PMID 21827451

    Cannabidiol suppressed nicotine-, lithium chloride- and cisplatin-induced vomiting in Suncus murinus (at 20 but not 40 mg/kg cisplatin) and lithium-induced conditioned gaping in rats; both effects were prevented by the 5-HT1A antagonists WAY100135 or WAY100635. Intra-dorsal-raphe WAY100635 reversed the anti-nausea effect of systemic cannabidiol, and cannabidiol injected into the dorsal raphe suppressed nausea-like behavior, reversed by systemic WAY100635; in vitro, cannabidiol enhanced 8-OH-DPAT-stimulated [35S]GTPgammaS binding in rat brainstem membranes along a bell-shaped dose-response curve. Shrew and rat; a bell-shaped curve and indirect agonism both complicate the mechanism.

78. 78
    Ray AP, Chebolu S, Darmani NA. Receptor-selective agonists induce emesis and Fos expression in the brain and enteric nervous system of the least shrew (Cryptotis parva).Pharmacol Biochem Behav · 2009 · 94(1):211-8doi:10.1016/j.pbb.2009.08.010 · PMID 19699757

    Cisplatin and selective 5-HT3, dopamine D2/D3 and neurokinin-1 receptor agonists were used to induce vomiting in the least shrew, with Fos immunoreactivity quantified as an activity marker. Enteric nervous system, nucleus of the solitary tract and dorsal motor nucleus of the vagus were activated by all emetogens; the area postrema was not activated by the dopaminergic or neurokinin-1 agents; the dorsal raphe nucleus and paraventricular thalamic nucleus, not usually discussed in emesis, also showed increased emesis-related Fos. Fos is a correlate of activity, not a causal test, and the species is a shrew.

79. 79
    Trulson ME, Jacobs BL. Dissociations between the effects of LSD on behavior and raphe unit activity in freely moving cats.Science · 1979 · 205(4405):515-8doi:10.1126/science.451617 · PMID 451617

    The first careful test of the raphe-silencing theory in awake, freely moving cats. The results gave general support to the hypothesis but with three dissociations: low doses of d-lysergic acid diethylamide produced only small decreases in raphe unit activity with significant behavioral change; the behavioral changes outlasted the depression of unit activity; and raphe neurons were at least as responsive during tolerance as in the non-tolerant condition. Cats, and the relation between firing and behavior is correlational, which is the point it undermines.

80. 80
    Trulson ME, Heym J, Jacobs BL. Dissociations between the effects of hallucinogenic drugs on behavior and raphe unit activity in freely moving cats.Brain Res · 1981 · 215(1-2):275-93doi:10.1016/0006-8993(81)90507-2 · PMID 6114779

    The extension across drug classes in freely moving cats. Lysergic acid diethylamide gave dose-dependent falls in raphe unit activity and dose-dependent behavioral change with temporally correlated peaks, but with the three dissociations of the 1979 paper; psilocin lowered firing dose-dependently while its behavioral changes were not dose-related; the phenethylamines DOM and mescaline produced large behavioral changes with no overall effect on raphe neurons, individual units increasing, decreasing or unchanged. The authors conclude that the serotonin hypothesis of hallucinogenic drug action should be re-evaluated. Cat electrophysiology with behavioral scoring, not human experience.

81. 81
    Aghajanian GK, Marek GJ. Serotonin and hallucinogens.Neuropsychopharmacology · 1999 · 21(2 Suppl):16S-23Sdoi:10.1016/S0893-133X(98)00135-3 · PMID 10432484

    Review by the laboratory that had made the 1968 raphe observation, tracing the serotonin hypothesis from the early 1950s onward and placing the common site of action of the indoleamine and phenethylamine classes at partial agonism of 5-HT2A and other 5-HT2 receptors, with prominent effects in locus coeruleus and cerebral cortex, including a 5-HT2A-mediated enhancement of late glutamatergic excitatory postsynaptic potentials at apical dendrites of layer V cortical pyramidal cells. A narrative review by interested parties, and the cortical-glutamate proposal is a hypothesis in it.

82. 82
    Beliveau V, Ganz M, Feng L, et al. A high-resolution in vivo atlas of the human brain's serotonin system.J Neurosci · 2017 · 37(1):120-128doi:10.1523/JNEUROSCI.2830-16.2016 · PMID 28053035

    Multidimensional in vivo atlas of four human serotonin receptors and the serotonin transporter built from positron emission tomography and magnetic resonance imaging in 210 healthy individuals. Regional positron emission tomography binding correlated highly with postmortem autoradiography for all five targets, which allowed transformation of the atlas into protein densities in picomoles per milliliter; regional densities were then compared with messenger RNA expression from the Allen Human Brain Atlas. Healthy volunteers and group-level maps; brainstem nuclei a few millimeters across are at the edge of what this resolves.

83. 83
    Madsen MK, Fisher PM, Burmester D, et al. Psychedelic effects of psilocybin correlate with serotonin 2A receptor occupancy and plasma psilocin levels.Neuropsychopharmacology · 2019 · 44(7):1328-1334doi:10.1038/s41386-019-0324-9 · PMID 30685771

    Eight healthy volunteers scanned with the 5-HT2A agonist radioligand [11C]Cimbi-36 at baseline and once or twice after a single oral dose of psilocybin (3-30 mg). Dose-related 5-HT2A occupancies up to 72 percent; plasma psilocin levels and occupancy conformed to a single-site binding model; subjective intensity correlated with both occupancy and psilocin levels. Eight participants in an open-label dose-finding design; it establishes that the receptor is engaged in people at psychoactive doses, not that engagement causes a clinical benefit.

84. 84
    Warren AL, Lankri D, Cunningham MJ, et al. Structural pharmacology and therapeutic potential of 5-methoxytryptamines.Nature · 2024 · 630(8015):237-246doi:10.1038/s41586-024-07403-2 · PMID 38720072

    Five cryogenic electron microscopy structures of the 5-HT1A receptor with 5-methoxy-N,N-dimethyltryptamine and related ligands, plus structure-activity analysis of 5-methoxytryptamines at 5-HT1A and 5-HT2A, receptor mutagenesis and mouse behavior, contrasted with the pan-serotonergic agonist lysergic acid diethylamide and with clinically used 5-HT1A agonists. A 5-HT1A-selective analogue lacked hallucinogenic-like effects in mice while retaining anxiolytic-like and antidepressant-like activity in socially defeated animals. Structural and mouse work; the human therapeutic claim is untested.

85. 85
    Moliner R, Girych M, Brunello CA, et al. Psychedelics promote plasticity by directly binding to BDNF receptor TrkB.Nat Neurosci · 2023 · 26(6):1032-1041doi:10.1038/s41593-023-01316-5 · PMID 37280397

    Molecular, computational and mouse work reporting that lysergic acid diethylamide and psilocin bind TrkB, the receptor for brain-derived neurotrophic factor, with affinities about 1000-fold higher than those of other antidepressants, at distinct but partly overlapping sites within the transmembrane domain of TrkB dimers. Effects on neurotrophic signaling, plasticity and antidepressant-like behavior in mice depended on TrkB binding and on promotion of endogenous BDNF signaling and were independent of 5-HT2A activation, whereas the head-twitch response depended on 5-HT2A and not on TrkB. In vitro and mouse; a direct competitor to the intracellular 5-HT2A account, and the two have not been adjudicated.

86. 86
    Højlund M, Kafali HY, Kırmızı B, et al. Efficacy, all-cause discontinuation, and safety of serotonergic psychedelics and MDMA to treat mental disorders: a living systematic review with meta-analysis.Eur Neuropsychopharmacol · 2025 · 101:41-55doi:10.1016/j.euroneuro.2025.09.011 · PMID 41205366

    Thirty double-blind randomized trials to 8 July 2025, 1480 participants, 45.8 percent female, psychological support in 83.3 percent, and high risk of bias in 83.3 percent. In major depressive disorder, psilocybin, ayahuasca or LSD reduced depressive symptoms with a standardized mean difference of -0.62 (95 percent CI -0.97 to -0.28), I-squared 55 percent, k = 8, GRADE certainty very low. Moderate certainty was reached only for MDMA in post-traumatic stress disorder against placebo. The authors name functional unblinding and expectancy as unresolved.

87. 87
    Hieronymus F, López E, Werin Sjögren H, Lundberg J. Control group outcomes in trials of psilocybin, SSRIs, or esketamine for depression: a meta-analysis.JAMA Netw Open · 2025 · 8(7):e2524119doi:10.1001/jamanetworkopen.2025.24119 · PMID 40736734

    Seventeen trials: 4 of psilocybin (n = 373), 2 of esketamine (n = 573), 11 of selective serotonin reuptake inhibitors (n = 4014). Standardized mean change in control arms was 0.50 for psilocybin trials against 1.00 for SSRI trials and 1.12 for esketamine trials, and control-arm response rates were 14 and 23 percentage points higher in SSRI and esketamine trials respectively; active-arm change did not differ significantly across populations. The authors read this as evidence that psilocybin's advantage may be overestimated. An indirect comparison across trial populations, not a head-to-head trial.

88. 88
    Yngwe H, Plaven-Sigray P, Ekman CJ, et al. Short-term and late-term effects of psilocybin on symptoms in major depression: a randomized clinical trial.JAMA Netw Open · 2026 · 9(5):e2612589doi:10.1001/jamanetworkopen.2026.12589 · PMID 42138922

    Double-blind, active-placebo-controlled trial at the Northern Stockholm Psychiatric Clinic, January 2021 to February 2024 (NCT04630964). Thirty-five participants with moderate-to-severe recurrent major depressive disorder (21 female; mean age 41.0 years), 17 to a single 25 mg dose of psilocybin and 18 to niacin 100 mg, each with 5 psychotherapeutic support sessions over 17 days. Between-group difference in Montgomery-Asberg Depression Rating Scale change: day 8 -7.27 (95 percent CI -12.89 to -1.65, p = .01), day 15 -11.03 (p < .001), day 42 -8.33 (p = .004), day 365 -3.68 (p = .20, not significant). Self-rated differences appeared by day 2 and persisted to day 102. No drug-related serious adverse events; two participants in the psilocybin group reported persistent severe anxiety requiring medical attention. Thirty-five participants at one site, and niacin is an imperfect active placebo.

89. 89
    Mertens LJ, Koslowski M, Betzler F, et al. Efficacy and safety of psilocybin in treatment-resistant major depression: the EPISODE randomized clinical trial.JAMA Psychiatry · 2026 · 83(5):448-460doi:10.1001/jamapsychiatry.2026.0132 · PMID 41848690

    Two-center, triple-blinded phase 2b active-placebo-controlled trial in Germany (NCT04670081); 144 randomized, 142 analyzed, mean age 42.6 years, 59 percent male, all withdrawn from antidepressant medication. Primary endpoint, 50 percent or greater reduction in HAMD17 at week 6: 17.0 percent with psilocybin 25 mg, 12.5 percent with 5 mg, 10.6 percent with nicotinamide. The first hierarchical comparison, psilocybin 25 mg versus nicotinamide, gave an adjusted odds ratio of 1.73 (95 percent CI 0.53-6.23, p = .19 against a one-sided alpha of .03), so formal testing stopped there and the secondary mean-change findings are exploratory. Suicidal ideation was reported on 4 percent of psilocybin 25 mg dosing days versus 1-2 percent in comparators; two serious adverse reactions including one case of hallucinogen persisting perception disorder. The authors describe the trial as inconclusive.

90. 90
    Retraction Note: Mithoefer MC, Feduccia AA, Jerome L, et al. MDMA-assisted psychotherapy for treatment of PTSD: study design and rationale for phase 3 trials based on pooled analysis of six phase 2 randomized controlled trials.Psychopharmacology (Berl) · 2024 · 241(11):2405doi:10.1007/s00213-024-06666-x · PMID 39126501

    Retraction notice, November 2024. The PubMed record is the notice itself and carries no abstract; what it fixes is that the pooled phase 2 analysis underpinning the phase 3 program was retracted. Authors were affiliated with the Multidisciplinary Association for Psychedelic Studies and MAPS Public Benefit Corporation, the program's commercial sponsor.

91. 91
    Retraction Note: Jerome L, Feduccia AA, Wang JB, et al. Long-term follow-up outcomes of MDMA-assisted psychotherapy for treatment of PTSD: a longitudinal pooled analysis of six phase 2 trials.Psychopharmacology (Berl) · 2024 · 241(11):2407doi:10.1007/s00213-024-06665-y · PMID 39126500

    Retraction notice, November 2024, for the long-term follow-up pooled analysis of the same six phase 2 trials. No abstract in the record; the authors' affiliations are the sponsor and its public benefit corporation.

92. 92
    Retraction Note: Feduccia AA, Jerome L, Mithoefer MC, Holland J. Discontinuation of medications classified as reuptake inhibitors affects treatment response of MDMA-assisted psychotherapy.Psychopharmacology (Berl) · 2024 · 241(11):2403doi:10.1007/s00213-024-06671-0 · PMID 39126502

    Retraction notice, November 2024, for the third of the sponsor's retracted analyses. No abstract in the record.

93. 93
    Caspi A, Sugden K, Moffitt TE, et al. Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene.Science · 2003 · 301(5631):386-9doi:10.1126/science.1083968 · PMID 12869766

    Prospective-longitudinal study of a representative birth cohort reporting that a functional polymorphism in the promoter region of the serotonin transporter gene moderated the influence of stressful life events on depression: carriers of one or two short alleles showed more depressive symptoms, more diagnosable depression and more suicidality in relation to stressful events than long-allele homozygotes. Presented by the authors as evidence of a gene-by-environment interaction; it did not survive replication in large samples, which is why it belongs on this page.

94. 94
    Border R, Johnson EC, Evans LM, Smolen A, Berley N, Sullivan PF, Keller MC. No support for historical candidate gene or candidate gene-by-interaction hypotheses for major depression across multiple large samples.Am J Psychiatry · 2019 · 176(5):376-387doi:10.1176/appi.ajp.2018.18070881 · PMID 30845820

    Preregistered analyses in population-based and case-control samples (subsample n from 62,138 to 443,264) of the 18 genes studied ten or more times as depression candidates, testing polymorphism main effects, polymorphism-by-environment interactions and gene-level effects across several definitions of depression (lifetime diagnosis, current severity, recurrence) and several moderators (childhood sexual or physical abuse, socioeconomic adversity). No clear evidence for any candidate gene association or any gene-by-environment moderation; candidate genes were no more associated with depression than non-candidates, and phenotypic measurement error was shown unlikely to explain the nulls. The authors conclude that the earlier candidate-gene literature is largely false positives.

95. 95
    Lepri A. The myth of serotonin theory of depression: an analysis of umbrella review methodologies and clinical implications.Psychiatr Danub · 2025 · 37(Suppl 1):187-188PMID 40982824

    A short supplement article reviewing the umbrella review alongside two major commentaries. It identifies post-hoc protocol amendments, selective reporting and down-weighting of serotonergic findings as weakening the review's conclusiveness, and argues for serotonin as a modulatory factor in vulnerable populations rather than a sole cause. Two pages in a conference supplement; no DOI in the record; low evidentiary weight, cited as a marker of where the debate stood in 2025.

96. 96
    Cipriani A, Furukawa TA, Salanti G, et al. Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis.Lancet · 2018 · 391(10128):1357-1366doi:10.1016/S0140-6736(17)32802-7 · PMID 29477251

    Systematic review and network meta-analysis of published and unpublished double-blind randomized trials to 8 January 2016 (PROSPERO CRD42012002291): 522 trials, 116,477 participants, 21 antidepressants. All were more effective than placebo, with odds ratios from 2.13 (95 percent credible interval 1.89-2.41) for amitriptyline to 1.37 (1.16-1.63) for reboxetine; only agomelatine and fluoxetine had fewer dropouts than placebo. Forty-six trials (9 percent) were rated at high risk of bias, 380 (73 percent) moderate and 96 (18 percent) low, and the certainty of evidence was moderate to very low. Response-rate outcomes over acute treatment only, with wide credible intervals on most comparisons.

97. 97
    Grinberg LT, Rüb U, Ferretti REL, et al. The dorsal raphe nucleus shows phospho-tau neurofibrillary changes before the transentorhinal region in Alzheimer's disease. A precocious onset?Neuropathol Appl Neurobiol · 2009 · 35(4):406-16doi:10.1111/j.1365-2990.2009.00997.x · PMID 19508444

    The dorsal raphe of 118 brains from the Brain Bank of the Brazilian Aging Brain Study Group, 38 staged Braak 0 and 80 Braak 1 or above. Neurofibrillary changes were present in the supratrochlear subnucleus of the dorsal raphe in all Braak 1-and-above cases and in more than a fifth of Braak 0 cases, which the authors read as involvement before the transentorhinal cortex and as support for transneuronal spread. Cross-sectional autopsy data; the order of appearance across individuals is inferred, not observed in any one person over time.

98. 98
    Hary AT, Chadha S, Mercaldo N, et al. Locus coeruleus tau validates and informs high-resolution MRI in aging and at earliest Alzheimer's pathology stages.Acta Neuropathol Commun · 2025 · 13(1):44doi:10.1186/s40478-025-01957-6 · PMID 40022196

    High-resolution postmortem magnetic resonance imaging with validated tau histopathology in the same samples (n = 11), from controls through the earliest Braak stages, with multiple histology sections per sublevel and tau severity spatial matrices for the locus coeruleus and the caudal dorsal raphe. Locus coeruleus volume was 30.0 percent smaller between Braak 0 and II (p = 0.0381), affecting the rostral-most portion most (49.2 percent smaller, p = 0.0381); locus coeruleus tau rose 37.6 percent between Braak I and II (p < 0.0001); the caudal dorsal raphe showed a rostrocaudal gradient with 58.2 percent lower tau severity caudally (p < 0.0001). Eleven samples, cross-sectional, so stage order is inferred.

99. 99
    Kolling LJ, Balasubramanian N, Ismail S, et al. Selective reduction of KCNA4 in vulnerable glutamatergic-serotonin neurons of the dorsal raphe nucleus in Alzheimer's disease.Alzheimers Dement · 2026 · 22(5):e71317doi:10.1002/alz.71317 · PMID 42117485

    Computational, molecular, biophysical and behavioral work in htau mice with post mortem human comparison. The centromedial dorsal raphe is composed predominantly of neurons that are both serotonergic and glutamatergic; in the mouse these differentially express ion-channel genes and are more excitable, which the authors propose promotes phosphorylated tau accumulation, and in human Alzheimer brain at Braak stage II the potassium-channel gene KCNA4 is reduced in those dual-phenotype neurons, which are more vulnerable than the serotonin-only neurons. A mechanistic hypothesis with mouse causal data and human correlational data.

100. 100
     Siletti K, Hodge R, Mossi Albiach A, et al. Transcriptomic diversity of cell types across the adult human brain.Science · 2023 · 382(6667):eadd7046doi:10.1126/science.add7046 · PMID 37824663

     Single-nucleus RNA sequencing of more than three million nuclei from about 100 dissections across forebrain, midbrain and hindbrain in three postmortem human donors, identifying 461 clusters and 3313 subclusters organized largely by developmental origin, with high diversity among midbrain and hindbrain neurons and regional diversity in astrocytes and oligodendrocyte-lineage cells. Three donors and whole-brain sampling rather than nucleus-targeted dissection, so it bounds rather than closes the question of human dorsal raphe subtypes.

101. 101
     Gray NA, Milak MS, DeLorenzo C, Ogden RT, Huang YY, Mann JJ, Parsey RV. Antidepressant treatment reduces serotonin-1A autoreceptor binding in major depressive disorder.Biol Psychiatry · 2013 · 74(1):26-31doi:10.1016/j.biopsych.2012.11.012 · PMID 23374637

     Nineteen subjects with no recent antidepressant exposure, scanned with [11C]WAY-100635 positron emission tomography and a metabolite-corrected arterial input function before and after 5 to 9 weeks of selective serotonin reuptake inhibitor treatment (mean 47 +/- 8 days). Raphe 5-HT1A autoreceptor binding (BPF) fell 18 percent (df = 1,18; F = 5.12; p = .036), but the degree of reduction was unrelated to improvement in depression (df = 1,16; F = 1.27; p = .276). The authors read downregulation as necessary but insufficient for response. Small, open-label, no placebo arm; an agonist ligand would be needed to show desensitization rather than downregulation.

102. 102
     Lemonde S, Turecki G, Bakish D, et al. Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide.J Neurosci · 2003 · 23(25):8788-99doi:10.1523/JNEUROSCI.23-25-08788.2003 · PMID 14507979

     The human risk allele behind the mouse autoreceptor models. The homozygous G(-1019) allele was enriched about twofold in depressed patients versus controls (p = 0.0017 for genotype, 0.0006 for allele distribution) and fourfold in completed suicide cases (p = 0.002 and 0.00008); the C(-1019) allele sits in a 26-base-pair imperfect palindrome bound by the repressors NUDR and Hes5, and the G allele abolishes NUDR-mediated repression, which the authors propose derepresses autoreceptor expression and lowers serotonergic transmission. Case-control genetic association from one era and population; candidate-gene findings of this kind have often failed to replicate, which this page says.