Response to criticism of our serotonin in depression paper

We recently published a paper finding that the serotonin hypothesis of depression (the idea that depression is caused by low serotonin or reduced serotonin activity) is not supported by scientific studies that have been conducted over the last few decades. The serotonin hypothesis was communicated to the public as the “chemical imbalance” theory of depression. In surveys, 85 to 90% of people in western countries report believing that depression is caused by a chemical imbalance. We suggest that the idea that depression is caused by low serotonin or a chemical imbalance should no longer be communicated to patients as it is not supported by research.

This also brings into question what antidepressants are doing: if they are not correcting an underlying chemical problem, as people are often told (“like insulin for diabetes”), then other ways of understanding what they are doing, such as providing hope (the placebo effect) or numbing emotions (a common report by patients) may be more accurate descriptions.

Psychiatrists in Britain, some of them with career-long relationships with the pharmaceutical industry, responded to our paper’s findings and its implications here, which were then reported in several media pieces covering our paper. We would like to respond to these criticisms.

We would like to say first that no one should stop their antidepressant medication abruptly—this can be dangerous and is known to cause withdrawal effects, which can be severe and long-lasting in some people, especially those using the medications long-term. If anyone is considering this choice, we advise you to discuss it with your doctor and, if you go ahead, to undertake a gradual and supported reduction as advised by recent Royal College of Psychiatry guidance.

Contradictory responses—the serotonin imbalance theory was both never supported and also is still supported 

There are a few remarkable aspects of the criticisms to note before we address specific critiques. The first is that the criticisms of our paper were contradictory, with some prominent psychiatrists saying that there was nothing new in our review (“really unsurprising,” “not news”) as it was already understood that depression was not caused by low serotonin. However, other psychiatrists said it was “premature” to dismiss the serotonin hypothesis and that further studies are required (despite this hypothesis having been studied for more than 50 years now). The existence of contradictory viewpoints reveals the cognitive dissonance in the field.

Disconnect between the public and psychiatrists 

The second notable fact is how disparate the response was from psychiatrists and the public, with most psychiatrists shrugging off the finding as a straw man by saying “I don’t think I’ve ever met any serious scientists or psychiatrists who think that all cases of depression are caused by a simple chemical imbalance of serotonin” and that they are “broadly in agreement with the authors’ conclusion about our current efforts.”

In contrast, there has been an avalanche of interest from the public—with more than a million reads on The Conversation, and widespread media coverage, so that our paper is now amongst the top 600 papers ever shared (out of 21 million papers that have been tracked). This interest likely stems from how widespread the message is that depression is caused by a chemical imbalance and that antidepressants work by fixing this imbalance. Many people, including journalists, have been shocked to find out this is not true, with one presenter commenting “it blows your mind.”

It may well be the case that psychiatrists have a more “sophisticated” understanding of the role of serotonin than the lay public, but psychiatrists have failed to correct this misunderstanding. Several academics have said that “we never told anybody this explanation.” However, the public clearly is being given this explanation: in England last week, a doctor on BBC radio told the public that in depression “there is a chemical imbalance and antidepressant given at the right time will help with that chemical imbalance.” The same message was given to the public on a leading UK morning television programme earlier in the year by another doctor. It is not surprising that the vast majority of the general public (as shown in surveys) believe this message to be an established scientific fact.

We know from our analysis of textbooks and journal articles that the idea of low serotonin (the “chemical imbalance”) was widespread in medical literature and remains so in many current textbooks. It was only recently that the Royal College of Psychiatrists in Britain removed its reference to chemical imbalances, describing the theory as an “oversimplification,” but without explaining that there is no proof for low serotonin levels, or indeed for any other neurochemical theory of the causes of depression. The American Psychiatric Association continues to tell the public that “differences in certain chemicals in the brain may contribute to symptoms of depression.”

Psychiatrists fail to appreciate the enormous impact for patients of being told that depression is caused by a chemical problem in the brain and that antidepressants can fix this problem. Skating over this issue to turn to alternative hypotheses about the cause of depression or the mechanism of action of antidepressants neglects to address the fact that patients have been misled. It is alarming to hear that there is a problem in your brain and it is misleading to suggest that we know there are drugs which can fix it.

This narrative strongly encourages people to take antidepressants because it seems wholly rational to take a drug which reverses an underlying chemical problem; indeed, it seems irresponsible not to do so. What is being dismissed as trivial semantics by experts has had consequences for the life choices and self-perception of hundreds of millions of patients worldwide. Imagine being told that you had a major problem in your heart that required medication to fix—only to find out that that problem was not truly there.

For the public, the chemical imbalance has been no straw man or semantic approximation, but something that has guided the direction of their lives, choices, and health. We know that believing that your depression is caused by a chemical imbalance tends to make people more pessimistic about recovery (seeing their symptoms as more chronic and intractable), leads them to believe they have less ability to regulate their moods, and also leads them to believe that medication is a more credible solution than therapy. We should actively counter this myth and remove it from medical information conveyed to patients because it is not supported by evidence.

The mechanism of action of antidepressants does not matter, as we know that they work 

The other argument raised by critics was that even if antidepressants are not rectifying an underlying chemical problem, they can still be effective by modifying neurotransmitters—and we use many medications whose mechanisms we don’t understand. Some critics have said “Many of us know that taking paracetamol can be helpful for headaches and I don’t think anyone believes that headaches are caused by not enough paracetamol in the brain. The same logic applies to depression and medicines used to treat depression.”

First of all, the analogy is misleading because we know that paracetamol works by targeting the mechanisms that produce pain, and it does not produce an alteration in normal emotions and mental experiences. With antidepressants, we do not have evidence that they target the underlying biological basis of depressive symptoms, and they do produce mental and emotional changes which can account for their effects.

Secondly, we would suggest that how a drug works, or what exactly it does, is crucially important in evaluating whether it is useful or not. With a drug that modifies brain chemistry in ways we do not fully understand, it would be wise to take a cautious approach and be wary of using it for long periods of time on a continuous daily basis. This is a very different proposition from taking a drug that reverses an underlying deficiency.

With antidepressants, we are looking for heuristics, or rules of thumb, to make sense of what these medications are doing in the context of short-term randomised trials that show marginal differences from placebo (with the vast majority of studies lasting less than 12 weeks). The idea that medications work by rectifying an underlying imbalance is very reassuring. Indeed, who would decline such a “lock and key” treatment? And this seemed to have been the marketing strategy of drug companies in propagating this line. For instance, we are not overly worried about the long-term use of insulin in diabetes because supplementing a naturally occurring chemical back to normal levels seems unlikely to be a harmful approach.

However, if the approach to treatment is now reframed as altering brain chemistry in a system that has no underlying detectable problem (or involves a complex and nuanced alteration of serotonin as yet poorly understood) then we are faced with a very different proposition. The human brain has evolved over millions of years and involves thousands of inter-dependent chemical systems to regulate processes in the body and brain. It is a valid question to ask: what is the effect on the brain of modifying the action of a neurotransmitter in this complex, inter-dependent system, especially in the long term?

We may be guided in answering this question by the effects of other substances that affect mental processes such as thoughts and feelings, including recreational drugs like alcohol. These tend to cause tolerance from repeated use, and withdrawal effects when they are stopped; this combination is usually termed physical dependence (a state distinct from addiction). Most such drugs also have detrimental effects on things like concentration and memory when they are used frequently or continuously. We know that these theoretical concerns are borne out in practice with the use of antidepressants: there are withdrawal effects—which can be severe and long-lasting in some people—and negative impacts on memory, concentration and sleep, not to mention sexual and other physical adverse effects.

We should use antidepressants because we know they work, even if we don’t understand their mechanism of action 

Many critics have put forward the argument that it does not matter that antidepressants are not rectifying a chemical imbalance because we know that they are effective from clinical trials (and the mechanism of action is a secondary concern).

First, it’s important to remember that the majority of the effect of an antidepressant is due to a combination of the natural course of our moods and placebo effects. When you look at all the randomised controlled trials that have been performed together (as in this meta-analysis paper) they show that antidepressants are a little bit better than a placebo (an inactive sugar pill), but not much. Meta-analyses routinely find that placebos produce a 10-point improvement, while antidepressants produce a 12-point improvement, on a 52-point depression scale after 6 weeks of treatment. Many have argued that this 2-point difference between antidepressants and placebo does not amount to a worthwhile difference.

In fact, it is not certain that there is even as much difference as this because there are methodological problems with these studies that may explain this small difference between drugs and placebo. These include the possibility that people on antidepressants have an enhanced placebo effect because some will be able to guess that they got the real drug due to side effects and other subtle cues. Indeed, in one trial, in which patients were all given an antidepressant, but half told that it was a placebo and half told the truth, those who were been told that they had received an antidepressant showed twice the change in anxiety and depression scores compared to those who believed they had received the placebo. Expectations can have a powerful effect on outcomes.

Recent criticisms of antidepressant trials are detailed in this paper and this one. Other important points are that these trials are almost all conducted by drug companies, and the vast majority of them last only a few weeks, whereas of course many people end up taking antidepressants for months and frequently years. In general, the effects of drugs tend to diminish over time, especially for drugs associated with withdrawal effects, such as antidepressants.

Even if there are small differences between antidepressant and placebo that are not explained by artefacts of the trial methods, there are other mechanisms that can explain their effects, and we cannot, therefore, assume that they work by fixing an underlying chemical problem (as outlined further below).

Antidepressants might work via a mechanism other than serotonin 

Several critics have said that although antidepressants don’t work by correcting a serotonin deficiency there are many other possible biological mechanisms of depression that they may be targeting. The possible mechanisms included: acting through neurogenesis, “due to complex changes in neuronal functioning,” increasing neurotransmitter levels or changing cognitive biases by acting on the brain. One psychiatrist pointed out that there are 59 biological hypotheses for why depression may occur and antidepressants might be working on any of these proposed abnormalities. One or more of these hypotheses may pan out, but at the moment they remain hypotheses—that is, they are speculative, unproven ideas about things that might be relevant, and most of them come from work in animals or cells in a dish.

This line of argument illustrates how most critics just assume that there must be something wrong with the brain: “it is very clear that people suffering from depressive illness do have some abnormality of brain function, even if we do not know what it is.” They also assume that antidepressants must be acting on the biological processes that underpin depression and this reveals how they are wedded to what has been called a “disease-centred” model of drug action. This is the idea that drugs for mental health problems can only work by reversing underlying brain abnormalities that are responsible for producing the symptoms of mental health problems.

However, one of us has been arguing for many years now that there is an alternative explanation for how psychiatric drugs work—the “drug-centred” model. This suggests that psychiatric drugs affect mental symptoms and behaviour through altering normal brain functioning and, through this, altering normal mental experiences and activity. When alcohol, for example, reduces social anxiety because of the typical mental and behavioural changes it produces, we recognise that these effects occur in anyone, regardless of whether they suffer from a diagnosed social anxiety disorder or not.

Any drug that changes normal brain activity is likely to have some impact on mood, and indeed, drugs with many different sorts of chemical actions have been shown to have comparable effects to drugs that are classified as antidepressants, including opiatesbenzodiazepinesstimulants, and antipsychotics.

By virtue of changing brain chemistry, antidepressants also produce changes to normal mental activity and experiences. The nature of these changes depends on the type of antidepressant—some antidepressants are strongly sedative, for example, but others are less so. Sedative drugs might improve sleep and reduce anxiety, which may be reflected in a lowering of depression symptom scores (because depression scales include several items on sleep and anxiety), but they may also make people feel groggy during the day.

Antidepressants are widely recognised to numb emotions (in a dose-related manner), including not just sadness and anxiety but welcome emotions like happiness and joy. Numbing emotions is also likely to reduce depression scores, and it may be experienced as useful by someone with a mental health problem, but it may not be.

All these effects may account for the small difference found between antidepressants and placebos in randomised trials (if these are not due to methodological artefacts). Hence the difference between placebo and antidepressants does not demonstrate anything about the basis of depression unless you make the surely indefensible assumption that all the effects described above are not relevant.

Taking a drug that numbs emotions may feel like a relief for someone who is deeply unhappy, fearful or confused. But in the long term taking a drug that alters normal brain chemistry may have harmful effects. In fact, we know that antidepressants cause physical dependence. The brain alters to try to counteract the effects of the drug, and then when people miss a dose or stop taking the drug they experience withdrawal effects, which are a consequence of the brain changes no longer being opposed by the drug. These can be severe and prolonged, especially if people have used the drugs for a long time.

Long-term use of drugs that numb emotions may also have harmful psychological consequences because it may prevent people from finding other, potentially more lasting ways of managing their emotions. It may also prevent people from identifying and addressing the problems that made them depressed in the first place.

But SSRIs work so there must be a serotonin problem of some sort 

Just because SSRI antidepressants show marginal benefits over placebo in randomised trials (as above), does not logically mean that depression is related to serotonin. For example, the fact that alcohol improves social anxiety does not mean that social anxiety is caused by an alcohol deficiency. And we do not think that headaches are caused by a paracetamol deficiency, as even many of the critics pointed out. This line of reasoning is so common there is even a term for this fallacy—the ex juvantibus fallacy (making inferences about the causes of an illness from response to treatment).

The relationship of serotonin to depression is more nuanced 

Many of the expert critics have suggested that although they recognise that a simple serotonin deficiency does not explain people’s depression, “changes in the serotonin system may be contributing to their symptoms,” in a more nuanced, complicated, and still poorly understood way. In one sense this is probably true—that serotonin in some complex way is involved in depression—and we agree with one critic who said, “it would be surprising if a such a widely distributed brain neuromodulatory system was completely uninvolved in the complex experiences that make up clinical depression.”

Indeed, it is also probably true that noradrenaline, dopamine, inflammation, cortisol, glutamate, and substance P in various interlinking neural networks are all involved in some nuanced, complicated, and poorly understood way in depression—because of course the brain works with electricity and chemistry and so these will be involved in different mood states. It would be equally true to say that serotonin (and all these other substances) is involved in some complex, nuanced way in hunger, fear, joy, thinking, walking, talking, and sleep. It is essentially an untestable and unfalsifiable claim to say that a given chemical is involved in a complex, nuanced way in depression.

However, it is a very different sort of claim to say that a specific neurotransmitter is altered in depression and provides a target for treatment. The non-specific argument that serotonin is involved in some complex, nuanced manner is not a sound basis for manipulating serotonin as a treatment for depression. This is similar to making a general claim that biology is involved in depression (as it surely is) to justify the use of any biological treatment. Biology is involved in diabetes but this does not justify any biological treatment (e.g. blood pressure medication). Instead, a specific biological problem (insufficient insulin production) is used to justify a specific remedy (exogenous insulin).

Alternative understandings of depression 

None of the experts who criticised our research or rushed to defend the use of antidepressants acknowledged that there are other ways of understanding depression, and other approaches to helping people who suffer from it.

There is abundant research that shows that stressful life events strongly predict depression. One study found that combining this with personality structure (“neuroticism,” which could be understood as essentially sensitivity to stress) shows an incredibly strong relationship to the risk of depression—a strength of relationship entirely absent from research into brain changes.

This is not to dismiss the idea that biology is involved in our moods in some shape or form—genetics plays a significant role in shaping our personality, for example, along with upbringing and perhaps particularly childhood experiences. However, the role of biology in some general way is not the same as proposing a specific biological problem that can be reversed with a supposedly targeted biological treatment.

Some people suggest that even if environmental factors precipitate depression, depressed feelings are still produced by brain chemicals, and therefore modifying these chemicals can help alleviate these feelings. An analogy might demonstrate the limitation of this approach. We know that learning Japanese will produce changes in the electrical signals and chemistry of the brain. However, we would think it odd if a student of Japanese decided that they would like to work out what those chemical and electrical changes were rather than attending more Japanese classes. In the same way, if we know that situations that produce insecurity and stress lead to depression, trying to delineate the specific chemical correlates of depression might be less productive than addressing the challenging situations that are the root cause of it.

Overall, the search for the brain basis of depression in chemistry may be committing a category error, mistaking trouble in the mind for trouble in the brain, like opening up the hood of a computer when a piece of software crashes.


Overall, although academic psychiatrists may hold a more sophisticated view of the role of serotonin in depression than that it is simply lowered (although some continue to advocate for this explanation), this explanation for depression has been widely communicated to the public as the “chemical imbalance” theory of depression and this has affected their treatment choices and how they view themselves. This likely explains the considerable interest generated by our paper.

Despite opinions to the contrary, being told that a drug acts on the underlying chemical cause of depression is quite distinct from being told that it changes the brain in ways we don’t understand and may act via placebo effects or numbing. This information is likely to have a profound effect on how people evaluate antidepressants and the decisions they make about them. The effectiveness of antidepressants in clinical trials is still highly contested, and other proposed theories of how antidepressants might target hypothetical biological processes underpinning depression have not been proven or demonstrated in humans.

Our overall approach to searching for the chemical equation of depression may not be the most fruitful manner in which to understand depression, given there is such strong evidence that stressful life events are closely linked to the onset of depression.