How Antidepressants Actually Work

Mechanism, Drug Classes and the Newer Paradigm

A small note before we start. This is a text about how these drugs work and how the classes differ, it is not a guide for dosing, starting or stopping anything. Every single one of those decisions belongs with the treating physician & nowhere else. I am serious about that one, it comes back a few times.

1 - The Old Myth and Why It Sticks

Almost everyone has heard the story by now. Depression is a chemical imbalance in the brain, a serotonin deficiency, and antidepressants simply top up the missing chemical back up. It’s a lovely, tidy story. It is also wrong as a model of cause.

The idea goes back to the late 1950s & 60s. People noticed that the first antidepressants raise the availability of monoamines in the brain, serotonin, noradrenaline, and dopamine, and the tempting reverse conclusion followed almost by itself. If a drug that helps depression raises serotonin, then depression must be a lack of serotonin. That conclusion got popularized for decades, the pharmacological industry included, because it explains so neatly and makes treatment feel obvious.

The trouble is the logic. The fact that a drug acts on a system does not prove that a fault in that system caused the illness. Aspirin helps a headache without the headache being an aspirin deficiency. That’s exactly the mix-up sitting inside the imbalance myth.

What survives is a lot more nuanced. Serotonin clearly does matter for how we process emotion, how we weigh up threat, how reward and reinforcement learning work, and as the entry point into a whole cascade of slower changes we will get into. Depression itself we now understand as multifactorial. Genetic vulnerability, early & current stress, inflammatory processes, and shifts in brain networks & plasticity all feed into each other. No single messenger carries that weight.

1.1 - The Serotonin Debate

In 2022 a much-discussed review pulled together six research areas on the serotonin hypothesis, from serotonin & its breakdown products in body fluids through receptor & transporter imaging to the genetics. The verdict was that there is no consistent evidence that depression is caused by low serotonin activity. It got downloaded over a million times & ran in the press, usually flattened into the headline that antidepressants don’t work at all.

That headline is false, & this is where the pushback comes in. Several groups argued the review overreached. Methodologically, it is a summary of existing reviews rather than fresh analysis. The imaging findings on receptors & transporters are genuinely complex & don’t collapse into a clean yes or no, & the core point stays logical: the absence of a proven serotonin deficiency does not show that the serotonin system is uninvolved or that serotonergic drugs don’t work.

However you turn it, both camps agree on one thing. The simple deficiency model is dead. What is alive is a system where serotonin is one important lever inside a bigger machine, never the single cause on its own. That distinction matters, because the death of the myth keeps getting sold as proof the drugs are useless, & that just isn’t what the evidence says.

2 - How They Actually Work

So if depression isn’t a simple shortage of serotonin, what is the drug actually doing up there? The honest answer comes in layers. What happens in the first hours, what slowly changes over the following weeks, & how that finally turns into a lift in mood. We walk through them in that order, from the synapse out to the whole circuit.

2.1 - What Happens Acutely in the Synapse

The acute pharmacology is quick to tell. An SSRI blocks the serotonin transporter at the nerve ending. Serotonin that would normally get hoovered straight back into the cell now lingers in the synapse for longer. An SNRI adds blockade of the noradrenaline transporter, the tricyclics pile a broad receptor blockade on top, & the MAO inhibitors raise monoamines by blocking the enzyme that breaks them down.

The interesting part is where the rise lands first. not out in the target regions at the front of the brain, but right at the cell bodies of the serotonin neurons themselves. Sitting there are inhibitory autoreceptors, the 5-HT1A autoreceptors. When serotonin climbs at those cell bodies, they act like a brake & slow the firing rate of the neurons. So the early net effect is paradoxically damped rather than boosted. That early brake is part of why the effect doesn’t show up straight away & why the first days often bring a bit of restlessness or jitter that settles later.

2.2 - The Delay Paradox

Here is the puzzle the whole story hangs on.
The change in the synapse happens within hours. The clinical improvement only arrives over weeks. If the drug were just topping up a missing chemical, the effect should be quick. It isn’t.

To be fair to the data, measurable improvement does begin for many people inside the first week or two; the old rule of strictly four weeks is outdated. But the basic gap between the pharmacology & the benefit is real & it stays. & that gap forces a conclusion. The thing that matters is not the acute rise in transmitter; it is what that rise slowly sets in motion.

2.3 - The Slow Adaptation: Autoreceptors, BDNF & Plasticity

Over days & weeks a few things shift. The inhibitory 5-HT1A autoreceptors grow less sensitive & downregulate, the early brake comes off, & serotonin release in the forward regions finally does climb. Alongside that, intracellular signaling cascades switch on & end up changing the activity of genes that govern plasticity & cell survival.

The key building block of the last years is BDNF & its receptor TrkB. BDNF is a growth factor that strengthens synapses & enables plasticity. A landmark study showed that antidepressants bind directly to TrkB & make its signaling easier. That’s elegant because it explains the timing. Classic antidepressants bind TrkB only weakly, so their concentration in the brain has to build up over days, hence the slow onset. Ketamine reaches effective levels fast, hence the rapid effect. One mechanism, two speeds.

The upshot of the whole cascade is remodelling. more working synapses, stronger connectivity, more plasticity in exactly the circuits that drift out of balance in depression. So antidepressants don’t refill a tank. They open a window in which the brain can relearn & rewire.

2.4 - 🧠 Neurogenesis and Limits of the Model

A related strand is the birth of new neurons in the hippocampus. In animals, certain behavioral effects of antidepressants disappear if you block that neurogenesis, which made hippocampal neurogenesis a popular piece of the explanation.

Staying honest means naming the limits, though. the findings dont line up cleanly. Other work found antidepressant effects that did not depend on neurogenesis at all but on the remodelling of existing neurons. the dependence seems to vary with the model & the drug. & Above all, the translation to humans is unproven, because these processes are nearly impossible to measure directly in the living human hippocampus. Neurogenesis is a plausible part of the picture, not the whole of it.

2.5 - The Psychological Bridge

There’s a model that ties the neurobiology to the lived experience, the cognitive neuropsychological model. The core idea is that antidepressants very early, sometimes after a single dose, shift automatic emotional processing. Negative cues get less preferential handling, and the reading of faces & memories loses some of its negative tilt. That shift happens before mood lifts.

Mood then follows later, because the person has to go out with the changed perception, gather new experiences & relearn the old negative associations. that takes time & contact with the world. So the model gives a second, psychological reason for the delay paradox, & it sits beautifully next to the plasticity story. Imaging fits too, with overactive amygdala responses settling down & the circuits between the frontal cortex & limbic system rebalancing.

3 - The Drug Classes

Now to the actual drugs. From the outside, it looks like an endless list of names, but they fall into a handful of families, & the whole thing gets a lot calmer once you know what to look for in each one. So we start with how to read a class, then take the families one by one.

3.1 - How to Read a Class

One trick makes this whole field readable. You don’t have to memorize every single drug; you ask two questions of each class. Which transporters does it block, & which receptors does it hit on the side? The wanted effect mostly comes through the transporters & the plasticity downstream. The side effect profile falls almost entirely out of the receptors it bumps into along the way.

One thing to put up front. In raw efficacy the drugs differ surprisingly little on average. The big differences lie in tolerability, safety & which person & which surrounding situation a drug fits. And deciding that fit is a job for the treating physician who weighs the whole person, not a checklist & definitely not a forum text.

3.2 - SSRIs

The selective serotonin reuptake inhibitors block the serotonin transporter on purpose & leave most other receptors fairly alone. Hence the comparatively clean profile & the role as first line in basically every guideline. there are still differences inside the group, a more anticholinergic & metabolism relevant streak with paroxetine, a very long duration of action with fluoxetine, a dose dependent effect on cardiac repolarisation with citalopram & escitalopram.

The side effects follow the pharmacology. Nausea & early inner restlessness track the serotonergic activation in the gut & brain. Sexual dysfunction is common & comes through several routes at once: stronger serotonergic action at certain receptors, a damping of the dopamine system & inhibition of nitric oxide-driven arousal. Emotional blunting gets reported too, a sense of flatness or a narrower emotional range. Other relevant points are close monitoring for suicidality in under 25s at the start, a raised risk of low sodium in older people & a slightly higher bleeding risk.

A delicate one to name, honestly, is sexual dysfunction that persists after stopping. The European regulator added a warning about this in 2019. The evidence is limited & leans heavily on case reports; the true frequency is unknown. It belongs in the picture, without drama & without hiding it.

3.3 - SNRIs

The serotonin-norepinephrine reuptake inhibitors block both transporters. With venlafaxine, which is dose-dependent, the serotonergic part leads at a low dose, & the noradrenergic part comes through clinically as the dose rises. Clinically they sit as an alternative or second choice, with an extra edge where pain comes along for the ride, neuropathic pain or fibromyalgia for instance.

The noradrenergic part gives the side effects their own flavor, more sweating, dry mouth & a dose-dependent rise in blood pressure, most of it with venlafaxine. Nausea is the most common one. It is worth noting that venlafaxine’s short duration of action gives it a higher risk of withdrawal phenomena & it is more toxic in overdose than the SSRIs. With duloxetine, a rare strain on the liver is worth keeping in mind.

3.4 - Tricyclics

The tricyclics are the old guard. They block reuptake of serotonin & noradrenaline but also hit a broad spread of receptors, which explains their whole profile. Blocking acetylcholine receptors brings dry mouth, blurred vision, constipation & cognitive effects. Blocking histamine receptors makes you sleepy & drives weight up. Blocking alpha receptors drops blood pressure on standing.

The genuinely critical bit is the blockade of cardiac sodium channels. That’s where the danger in overdose comes from, arrhythmia & the high lethality. Tricyclics are clearly effective; some even rank among the most effective drugs we have. But because of tolerability & overdose risk, they are not the first choice & stay reserved for severe or treatment-resistant courses, with a side role at low dose in pain. That risk profile is one more reason the call belongs to a doctor who knows the person’s full history.

3.5 - MAO Inhibitors

The MAO inhibitors take a different road; they block the enzyme that breaks monoamines down & so raise serotonin, noradrenaline & dopamine. That makes them powerfully effective, especially in atypical & treatment-resistant depression. The price is a set of precautions.

The famous one is the cheese reaction. If the enzyme is blocked in the gut for good, tyramine from food can’t be broken down; it floods in & can set off a dangerous blood pressure crisis. Here’s where reversible versus irreversible matters. A reversible inhibitor gets pushed off the enzyme again by rising tyramine, which cuts the risk a lot. Selectively blocking the B type sidesteps the problem too, since gut tyramine is still cleared. On top of that, the interactions are touchy, above all the risk of serotonin syndrome when combined with other serotonergic drugs. It is these constraints that limit their use, not any lack of effect.

3.6 - Atypicals

This is the grab bag, & each one earns its place through its receptor profile.

Mirtazapine goes the other way around & blocks inhibitory autoreceptors, which indirectly releases more serotonin & noradrenaline, & on top of that, it blocks several serotonin & histamine receptors. So the typical picture follows: sedating & appetite-raising through the histamine route, little sexual dysfunction & few gut problems through the serotonin receptor blockade. Funny enough, it gets used mostly to make people sleepy even though it lifts noradrenaline, because at the relevant doses the histamine effect simply wins. It is a favorite where depression comes with insomnia, agitation, or loss of appetite & weight.

Bupropion works through noradrenaline & dopamine & has essentially no direct serotonin action. It is activating rather than sedating, causes neither sexual dysfunction nor weight gain & is even used on purpose to offset exactly those SSRI effects. The important caveat is it lowers the seizure threshold & is contraindicated in seizure disorders & in eating disorders.

Agomelatine is the conceptually most interesting one, the first licensed antidepressant with a non-monoaminergic component. It acts at melatonin receptors & at the same time blocks a particular serotonin receptor, which steadies the circadian rhythm & improves sleep. Sexual function & discontinuation both look favorable. The central caveat is a possible strain on the liver that calls for liver value checks.

Trazodone is dose-dependent. At low doses it mainly promotes sleep; a real antidepressant effect only shows at higher doses once the serotonin transporter is occupied enough. Typical are sedation, a drop in blood pressure on standing & rarely priapism. In practice it gets used a lot as a sleep aid; as a standalone antidepressant, it counts as less well accepted.

Vortioxetine is multimodal; it blocks the serotonin transporter & also acts at several serotonin receptors, partly activating & partly blocking. That broad profile is credited with pro-cognitive effects. Tolerability is comparable to the SSRIs with relatively little sexual dysfunction.

Tianeptine we keep short & as a special case on purpose. For a long time it looked pharmacologically odd; today we know it acts as an agonist at the opioid receptor. That explains both the antidepressant action & the serious dependence & abuse potential that comes with it. In parts of Europe & Asia, it is licensed as an antidepressant; in other countries, there are warnings about misuse. It needs no more framing than that here.

4 - The New Paradigm

Everything up to here worked through the monoamines, serotonin & noradrenaline & their relatives. The last years opened a genuinely different door: agents that work through the glutamate system & can act in hours instead of weeks. Here’s the short tour, kept strictly to mechanism.

4.1 - Ketamine and Esketamine

Ketamine leaves the monoamine world entirely & acts on the glutamate system, the main excitatory messenger of the brain. It blocks a particular glutamate receptor, the NMDA receptor, & does so preferentially on inhibitory interneurons. That disinhibits the excitatory neurons downstream; a brief surge of glutamate follows. It activates another glutamate receptor & through exactly the BDNF TrkB cascade we met earlier, it grows new synapses very fast.

The remarkable thing is the speed. The effect comes on within hours, a real break with the weeks-long delay of the classic drugs. Esketamine, one mirror image form of the molecule, is licensed as a spray for treatment-resistant depression.

as promising as that is, the evidence deserves an honest look. the licensing trials are contested, several short term studies showed no clear edge over placebo, & proof that esketamine actually lowers suicidality is missing despite the indication carrying it. it is given only in a supervised setting because of dissociation, sedation, blood pressure rise & abuse potential. durability beyond the acute phase & the comparison to plain ketamine are still open. a real shift in mechanism then, handled with real caution, & firmly inside specialist care.

4.2 - 🧠 A Bridge to Psychedelics

purely on mechanism, no use recommendation, & i have a separate text on this if you want the proper deep dive [https://discord.com/channels/292845551110717441/292845551110717441/1463961581187760234]. classic psychedelics like psilocybin act through a particular serotonin receptor & also set off a wave of cortical glutamate release & rapid neuroplasticity. At the same time they acutely change how the brains large networks talk to each other. One reading is that this opens a window of heightened plasticity in which stuck negative patterns of thought and feeling can be revalued.

what is openly contested is whether the subjective trip is even needed for the effect or whether the plasticity could be triggered without it. the clinical data are early & methodologically hard, because real blinding is nearly impossible. fascinating as a research field, not a settled therapy.

5 - How Agents are Actually Chosen

So with all of these options on the table, how does a clinician actually land on one for a given person? Not the way most people picture it, & that gap is worth spelling out because it also says something about what these drugs really are.

5.1 - Side Effect Profile and Comorbidity, Not Neurotransmitter Matching.

Here’s where we clear up a second myth: There is no serious method that picks the right drug from some guessed transmitter deficiency. Because efficacy barely differs on average, the choice runs on other criteria.

In the front and center are the side effect profiles that fit the person & the conditions they bring with them. A sedating, appetite-raising drug can be an advantage where there is insomnia & weight loss. Someone who badly wants to avoid sexual side effects has other options. Where pain comes along, a dual-acting drug makes sense; where the trouble is circadian, a melatonergic one. Then come overdose safety, interactions with other medicines, response to earlier treatments, age, pregnancy & in the end the person’s own preference. None of this is a lookup table the reader can run on their own; it is the reasoning a clinician walks through with you.

One point matters enough clinically to get a fixed place here. The choice of drug hinges on the depression being correctly placed in the first place. The most dangerous mistake is to miss a depression that sits inside a bipolar illness & treat it with an antidepressant on its own, which can worsen the course. I wrote a separate piece on antidepressants tipping people into mania [https://discord.com/channels/292845551110717441/292845551110717441/1474970517394952374]. So the question of the right antidepressant always hangs on the question of the right diagnosis. Treatment decisions are a shared process with the treating physician, & that part is not up for negotiation.

5.2 - What the Efficacy Data Show & Where They Stop

The largest comparative analysis pools hundreds of trials with over a hundred thousand participants. The core result: every antidepressant studied beat placebo on average; the differences between drugs were small, and tolerability differed more clearly than effect did.

That work is important & criticized at the same time. Many of the included trials had methodological weaknesses or were funded by manufacturers, & part of the critique argues the gap to placebo might sit below the threshold of clinical meaning. The German guideline puts it well: in mild depression, an advantage over placebo is barely demonstrable statistically, & the benefit grows as severity rises. that is not an argument against antidepressants, it is an argument for careful indication.

6 - Coming Off: Discontinuation and Withdrawal

When antidepressants are stopped after longer use, withdrawal, or discontinuation phenomena can show up. That is real & deserves an honest hearing. the clear line first, stopping always runs planned & supported with the treating physician, never solo. This text gives no schedules on purpose, & that is a deliberate choice, not coyness.

How common & how severe these phenomena are genuinely contested. One large analysis landed on a rate around one in six or seven people, with severe courses in roughly one in thirty to forty, & argued earlier numbers were inflated by self-selection in online surveys. Other researchers counter that the short follow-up of many trials misses the dragged-out cases & arrive at clearly higher figures.

In practice the field converges despite the fight. The British guideline now deliberately says “withdrawal” rather than “discontinuation,” stresses a stepwise & individually fitted reduction, flags the higher risk with short-acting drugs, & warns to tell withdrawal apart from relapse. The underlying pharmacological idea is that receptor occupancy does not fall in a straight line with dose, which is why a slow reduction, especially at the low end, makes sense. & Again, that is a matter for medical guidance with your doctor steering, never something to improvise alone.

7 - What we still don’t know

an honest close, for all the decades of research, we don’t fully know the causal chain from that first molecular binding to the lift in mood. We have good building blocks, autoreceptors, BDNF & TrkB, plasticity & emotional relearning, but no seamless model.

Several big questions stay open. Why do the classic drugs need weeks while ketamine needs hours if both end up working through similar plasticity routes? How much does neurogenesis really matter in humans? Are there reliable markers that predict who responds to what, because for now a lot of it stays trial & error? How common & how lasting are persistent sexual dysfunction & dragged-out withdrawal really? & how well do the new glutamatergic & psychedelic approaches hold up in everyday care?

And one bit of humility overall of it: a sizeable share of the comparative & the newer evidence is industry funded. That doesn’t make the findings worthless; it does mean reading them with a careful eye. Antidepressants are effective tools with a mechanism we understand better every year & still don’t understand all the way down. That honesty, clear strengths & openly named gaps are what separate a serious account from a brochure. And the single most important line in the whole thing is the quiet one I keep repeating: what you take and how you ever come off it are decided together with your doctor.