The chemical imbalance explanation was never really the science, and the story now replacing it is being oversold in the opposite direction. The useful question was never how much serotonin you have. It is whether you can manufacture it — and whether inflammation is quietly diverting the raw material before it arrives.
Nothing in this article is a reason to stop taking an antidepressant. The evidence that the serotonin theory of depression is unsupported is not evidence that antidepressants do not work. Those are separate questions with separate literatures, and the second is considerably more favourable than the first.
Stopping an SSRI abruptly can produce severe discontinuation effects and relapse. If this article makes you want to change your medication, that is a conversation with your prescriber, not a decision to make from a website. Mine included.
Somebody arrives feeling flat, unmotivated, foggy, and anxious in a low-grade constant way. They have read that this is serotonin, and they would like to know what to take.
The honest answer is that nobody is measuring their serotonin, and that even if we could the number would probably not tell them what they want to know. But that is not the same as saying there is nothing biochemical to look at — and the space between those two positions is where the useful clinical work sits.
In 2022, Moncrieff and colleagues published an umbrella review in Molecular Psychiatry covering every main line of serotonin research: metabolite concentrations, 5-HT1A receptor binding, serotonin transporter levels, tryptophan depletion studies, and SERT gene associations. They concluded there was no consistent evidence of an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin. The two largest genetic studies — one with 115,257 participants — found no association and no gene-by-stress interaction.
It became a very large news story. Within academic psychiatry it was much less surprising than the coverage suggested; Cowen and Browning had written in World Psychiatry back in 2015 that the hypothesis had not been clearly substantiated.
Jauhar and colleagues responded in the same journal in 2023, arguing the conclusion was overstated — citing methodological weaknesses, selective reporting, and errors in interpreting tryptophan depletion and serotonergic molecular imaging. Their critics also make a fair point in return: the profession largely abandoned the simple chemical imbalance model years ago, even while it remained the explanation routinely given to patients.
So we have an explanation that specialists do not really hold and patients are still handed. I am not in a position to adjudicate between the two camps, and I would be suspicious of anyone in my field who claimed to be. What I would say is that the confident public verdict — serotonin theory debunked — is a good deal firmer than the actual state of the argument.
Whether serotonin deficiency causes depression and whether serotonergic drugs help depression are different questions. A drug does not have to correct a deficiency to work; paracetamol is not correcting a paracetamol deficiency.
Hieronymus and colleagues reanalysed 18 placebo-controlled SSRI trials covering 6,669 adults and found that when the single item “depressed mood” was the outcome rather than the full 17-item Hamilton sum, 29 of 32 drug-placebo comparisons detected an antidepressant signal — against 14 of 32 using the conventional total. Kendrick and colleagues put it plainly in the BMJ: antidepressants remain effective even without the chemical imbalance explanation.
The more current framing, set out by Page and colleagues in 2024, is that depression involves inflexibility in cognitive and emotional circuits with a persistent negativity bias, and that effective treatments — monoaminergic drugs, ketamine, psychotherapy, neuromodulation — work by enhancing neuroplasticity. Serotonergic drugs are on that list. They are just not on it for the reason people were told.
Here is the part that belongs to me rather than to psychiatry, and I want to be careful about the boundary.
Neurotransmitters are not stored in a reservoir that runs low. They are synthesised continuously, from amino acids in your food, by enzymes that each require specific cofactors to function. None of this is controversial — it is in every biochemistry textbook — and none of it amounts to a theory of depression. It is a set of places where synthesis can be constrained, which is a smaller and more defensible claim.
Tryptophan → 5-HTP → serotonin. The first step needs iron and tetrahydrobiopterin (BH4). The second needs vitamin B6 in its active P5P form.
Tyrosine → L-DOPA → dopamine. Tyrosine hydroxylase needs iron and BH4. The second step uses the same enzyme as serotonin, requiring the same B6.
Dopamine → noradrenaline requires copper and vitamin C. Noradrenaline → adrenaline requires methylation, and therefore folate, B12 and B2.
Glutamate → GABA, the main inhibitory signal, needs — again — B6.
Look at what is actually going on there. It is not four independent systems. It is one production line with shared inputs.
Why do so many people present with anxiety and low mood and poor focus and poor sleep, all at once?
A single-molecule model struggles with that. Four different neurotransmitters, all simultaneously low, for no coherent reason.
The cofactor picture handles it easily. If B6 is depleted, serotonin synthesis is constrained, dopamine synthesis is constrained, and glutamate cannot be converted to GABA. One bottleneck, four symptoms.
If iron is low — tissue-depleted, not necessarily anaemic — both tryptophan hydroxylase and tyrosine hydroxylase are affected. A ferritin of 20 µg/L sits inside the laboratory range while supporting those enzymes poorly. Fatigue, low mood, poor cognitive stamina and hair thinning are all consistent with that number, and the report says normal.
An important caveat before this becomes a theory of everything. That this can happen is textbook biochemistry. That it commonly does explain a given person's low mood is a much larger claim, and the trial evidence for supplementing these nutrients in unselected people with low mood is generally weak. Zinc has the most supportive signal; magnesium is inconsistent; the rest is largely extrapolated from mechanism rather than demonstrated in trials.
The defensible use is narrower and more honest: where someone has a documented deficiency, correcting it is worth doing on its own merits, and it is reasonable to expect mood and cognition to be among the things that improve. Iron deficiency without anaemia is the clearest example, and ferritin is cheap to measure.
There is a second route out of tryptophan, and under the right conditions it takes most of the traffic.
Tryptophan can go toward serotonin. Or it can enter the kynurenine pathway — and once there it splits again: toward kynurenate, which is broadly neuroprotective and an NMDA antagonist, or toward quinolinate, an NMDA agonist that is neurotoxic at sustained concentrations.
What determines the split? Inflammation. Inflammatory cytokines — interferon-gamma and TNF-alpha in particular — induce IDO, the enzyme that pulls tryptophan down the kynurenine route. Cortisol drives the liver equivalent, TDO.
So an inflamed system diverts its own serotonin substrate away from serotonin and toward a neurotoxic metabolite instead.
In an inflamed person, low mood may not be a manufacturing fault at all. The production line may be perfectly functional and simply receiving no raw material, because the raw material was requisitioned upstream.
This is well characterised. The clearest human demonstration is interferon-alpha therapy, historically used for hepatitis C, which induces depression in a substantial proportion of patients alongside measurable tryptophan diversion — about as close to an experiment as this field gets.
It also explains something that frustrates a lot of people: why loading up on 5-HTP or tryptophan sometimes does nothing at all. If inflammation is diverting the pathway, you can add substrate all day. It goes down the same diverted route.
Organic acid testing reports both quinolinate and 5-HIAA, the serotonin metabolite. The ratio between them shows how tryptophan is being routed — a rising quinolinate alongside a falling 5-HIAA is the diversion, visible on a page.
The honest caveat: showing that inflammation diverts tryptophan is not the same as showing that reducing inflammation improves mood. Anti-inflammatory approaches to depression have produced mixed trial results, with signals mostly in subgroups who had elevated inflammatory markers to begin with. Which is itself an argument for measuring rather than assuming.
The obvious question is why anyone would be short of these things while eating adequately.
Sustained stress is expensive. Producing catecholamines continuously consumes B6, vitamin C, magnesium and copper. The adrenal cortex holds one of the highest tissue concentrations of vitamin C in the body, and it is not holding it decoratively.
Poor absorption undoes adequate intake. Low stomach acid, insufficient pancreatic enzyme output, an inflamed intestinal surface — all mean what is on the plate is not what reaches the bloodstream.
Medications deplete specifically. Oral contraceptives and B6. Metformin and B12. Proton pump inhibitors and B12, magnesium and iron. None of these is a reason not to take a needed medication. All of them are a reason to know what is being consumed, and to check rather than assume.
The gut may contribute, though here I want to mark the evidence honestly. Certain Clostridium species produce HPHPA, a compound reported to inhibit dopamine beta-hydroxylase — the enzyme converting dopamine to noradrenaline — which would produce dopamine accumulating while noradrenaline depletes. It is a mechanistically coherent story and I have seen the marker fall alongside clinical improvement. But the dopamine beta-hydroxylase inhibition claim traces largely to one laboratory's work and has not been independently validated. I include it because the marker is measurable and the pattern is recognisable, not because the mechanism is settled.
And inflammation is the multiplier, because it runs the diversion described above while simultaneously increasing demand.
This is the practical payoff of thinking about pathways rather than a single molecule. Take three people who all describe low motivation and flat mood.
I am not claiming these are separate diseases, or that sorting people this way is validated. It is not. It is a way of deciding what to measure first, and it earns its place by occasionally finding something correctable that nobody had looked for.
Not a mood questionnaire.
Blood chemistry first, and often from the GP for free. Ferritin read at functional rather than merely non-anaemic levels, B12, folate, thyroid function, vitamin D, and the inflammatory markers — CRP, ESR, homocysteine — that indicate whether the diversion is likely to be running at all.
Organic acids give the actual metabolites: 5-HIAA for serotonin, homovanillate for dopamine, vanilmandelate for noradrenaline, the quinolinate-to-kynurenate picture for the diversion, functional B6 status via xanthurenate, and HPHPA where the gut may be contributing.
Dried urine hormone testing gives the cortisol pattern — cortisol being one of the inducers of that diversion — and melatonin, which is made from serotonin and therefore shares every bottleneck above.
Stool testing gives the gut contribution and, more basically, whether absorption is plausible in the first place.
This is not an argument against antidepressants. SSRIs help a great many people, and the mechanism being less simple than “topping up serotonin” does not make the benefit less real. If you are taking one and doing well, this article is not a reason to stop.
And it is not an argument that depression is purely biochemical. I would rather say this plainly than leave it implied: most low mood is not a nutrient deficiency. Grief, isolation, financial fear, an unbearable job, an unresolved trauma and simple exhaustion produce these symptoms far more often than any of the pathways above, and no supplement addresses any of them. Circumstance, loss and meaning matter enormously, and no amount of correcting cofactors substitutes for a life that has become unliveable.
What I am describing is one layer. It happens to be the layer that gets skipped, and it happens to be measurable.
The version of this article I could have written — where the chemical imbalance story collapses and functional testing rides in with the real answer — would be more shareable and would be making the same mistake in a different costume. Replacing one oversimplified molecular story with a slightly more elaborate one is not progress.
Depression is not one thing. That is the actual finding, and it is much less satisfying than either the serotonin story or the story people want to replace it with.
The useful question was never “is my serotonin low.”
It is: can this person manufacture what they need, are the cofactors present, and is inflammation diverting the raw material before it arrives?
Those are three answerable questions with three measurable answers — and a shorter list of things worth ruling out. That is a considerably better position than being told you have a chemical imbalance with no way to check. It is not a theory of mood, and I would not want it sold as one.
Educational content, not medical advice, and not a substitute for mental health care. If you are struggling, please speak to your GP or a qualified therapist — that is the appropriate first step, not testing. Never stop or change prescribed medication without talking to whoever prescribed it.
Ferritin, inflammatory markers and organic acids answer different questions. Knowing which question you are asking comes first.
Ask AIdan → See the TDG Programme