SSRIs and their cousins are among the most commonly prescribed drugs. But despite something like 15% of Americans using “antidepressants” for depression, generalized anxiety, panic disorder, PTSD, social anxiety, or OCD, we can’t actually explain how these medications work. This is a problem! It’s hard to employ medications appropriately if we don’t know what we’re targeting. It plays into the hands of anti-psychiatry critics. And it’s part of psychiatry’s broader failure to develop a physiology of emotions.
Serotonin reuptake inhibitors presumably increase serotonin in the brain. But this doesn’t shed much light, since “attempts to identify a unified theory of brain serotonin function have largely failed to achieve consensus.” Unlike dopamine or GABA or norepinephrine, we lack a plausible model here.
But there is a good theory available: serotonin inhibits emotions, and most antidepressants rely on this for therapeutic effect. This is an old idea, and controversial. Neuroscientists and research psychologists don’t like it, since they’d rather talk about threat responses and reinforcement contingencies than emotions and feelings. Psychiatrists don’t like it, since it makes us look like 9-foot-tall aliens with the face of John Travolta. But emotions and feelings are as real as any biological process, and the ability to inhibit, reduce, attenuate, regulate, or modulate them is profoundly useful, just like medicine’s ability to control heart rate, pain, or tremor.
“Serotonin as an emotion inhibitor” explains a great deal of evidence in a parsimonious way. It’s plausible on its face and fits clinical experience. It’s supported by a raft of basic research and grounded in principles of physiology and evolution. In fact, though couched in arcane behaviorist and computational language, most existing theories of serotonin say something similar.
There’s no conceptual problem here
Regulating emotions with pharmaceuticals is an ancient pastime. In modern psychiatry, benzodiazepines (think Xanax, Valium, Ativan) are anxiolytics - they “lyse” (break) anxiety. The anti-seizure medication Depakote breaks people out of mania, an intense mood state. Dopamine blockers reduce agitation, anger and fear, causing flat affect and low motivation at high doses. In the other direction, amphetamines induce active euphoria, opiates calm bliss.
SSRIs and similar medications are pretty good at treating depression without simply raising mood across the board. They’re antidepressants, not “mood boosters.” Unlike Adderall, a normal person taking fluoxetine won’t feel any better than usual. To be fair, 40% of depressed people don’t feel any better either.
But we must also account for their use in anxiety disorders, PTSD, OCD, and others. The obvious interpretation is that SSRIs et al reduce various bad feelings: sadness, guilt, anxiety, panic, irritability. This is uncontroversial, but it’s also special pleading: doctors like the idea that meds reduce unpleasant emotions but hate the idea they inhibit emotions more broadly.
The dissonance persists despite people constantly saying that SSRIs blunt their emotions. Patients commonly report feeling numb or like a zombie, with a compressed range of feeling on both positive and negative ends. This effect shows a dose response, with higher doses causing more emotional blunting. Don’t forget decreased libido; lust qualifies as a pleasant emotion.
We view this as a side-effect, rather than the therapeutic effect, only because it’s undesirable. They are inseparable, though, like a beta-blocker medication to lower heart rate that also makes it harder to exercise. One follows from the other. Bigger picture, this is wishful thinking. The point of emotions is the results they produce, not how they feel. We should be skeptical that evolution provides a mechanism for only turning off the unpleasant ones.
Regardless, the effect of serotonergic medications presumably follows the function of serotonin itself. There’s no grand theory of serotonin, but there is a lot of data, and it mostly points in the same direction.
Serotonin turns down activity
Serotonin promotes rest and digest mode in animals. It terminates feeding, suppresses appetite, ramps up intestinal movement, and stores energy (review). Conversely, depleting serotonin makes rats eat more carbs. Along the same lines, a high serotonin:dopamine ratio is associated with fatigue and greater perceived exertion. Serotonin inhibits sexual behavior and delays orgasm in mammals.
There’s a million papers premised on the idea that serotonin reduces negative feelings. This shows up in many ways. Perhaps serotonin diverts attention to nice things like smiling faces, turns up calming parasympathetic tone, improves coping, changes risk assessment of an elevated maze for mice, reduces rumination, and so on. It’s tempting to conclude that low serotonin causes low mood or anxiety, but this isn’t borne out.
Instead, low serotonin and serotonin depletion mostly correlate with impulsivity and aggression. Serotonin metabolites are lower in people who commit impulsive vs premeditated violence, as well as those who die by suicide. SSRIs reduce agitated behavior in people with dementia. Primates with high serotonergic sensitivity are more tolerant of peers, while low serotonin sensitivity correlates to greater vigilance and threats toward others.
A common interpretation is that serotonin constrains behavior. Not coincidentally, most serotonin receptors are inhibitory at the cellular level. Across the board, serotonin reduces reactivity and various forms of impulsive behavior while promoting “waiting.” This review notes the view that serotonin acts “possibly by enhancing aversion and increasing behavioral inhibition.”
What’s the bigger picture? Why is serotonin “involved in behavioral control when there is a prospect of reward or punishment?” Perhaps it promotes a “satiated waking state.” Perhaps it activates a “primitive drive to withdraw” from high stimulation to safety. But how do we square these ideas with serotonin improving depression? Depression is not an impulsive state, it involves too much withdrawal, and it doesn’t feel like satiety in the slightest. These interpretations don’t quite hang together.
Serotonin turns down emotions
The picture is clearer if we accept that serotonin does reduce (all or most) emotions, both pleasant and unpleasant. This isn’t a crazy idea; The Archaeology of Mind offhandedly mentions that “serotonin tends to reduce all forms of emotional arousal.” And as noted, it readily explains the effectiveness of medications for various emotional disorders, the phenomena of emotional blunting and reduced libido, and the connection of low serotonin to impulsivity.
A fair term for impulsive, heuristic, quick responses is “emotions,” though many scientists find this term improper. This review posits two systems for behavior: “a lower-order system that responds quickly to associative cues of the moment and a higher-order system that responds more reflectively and planfully.” Sure sounds like emotion and reason, not to mention Kahneman’s System 1 and System 2. The authors further argue that low serotonin hands control to the lower-order system (emotions), consistent with all the evidence above. You can see this as serotonin either reducing emotions or promoting executive function; it’s two sides of the same coin.
Biology is chock full of opponent processes
Since the 1950s, we’ve known that serotonin opposes the effect of activating neurotransmitters. Acetylcholine, norepinephrine, and dopamine basically correlate with increased arousal, awareness, and movement. Serotonin supports “functions roughly opposite to those of [dopamine].” Why is this good? Low serotonin would be adaptive when exploring a new environment, requiring vigilance and quick reactions. High serotonin helps in familiar surroundings, where arousal and motion just waste energy.
The idea of one brain system inhibiting others is unremarkable. Evolution frequently arrives at a scheme of regulation by opponent forces, with a tension of on-off or promote-inhibit mechanisms. Familiar examples include insulin and glucagon managing blood glucose, or alternating contraction and relaxation of muscle pairs. This dynamic is ubiquitous in nervous systems and creates robust systems with stable oscillations. Like a water faucet with separate hot and cold handles, a good way to control behavior is with a balance of opposing drives.
This regulatory dynamic works for emotion and behavior too. Emotions are action modes, akin to other modes like feeding and sex, necessary because different activities demand different mental and physiological settings. They arise in response to environmental and homeostatic needs, but these needs are constant and overwhelming. The risk is that emotions and drives pull the animal in too many directions, or too far down one path. There must be mechanisms to adjust the threshold for activation, the size of response, duration, and so on.
Getting more specific
“Regulate” and “modulate” are weasel words, with “inhibit” and “reduce” only slightly better. Here are specific ways I can imagine serotonin affecting an emotion circuit:
Raise the threshold for activation (wider deadband)
Add lag or delay
Lower the gain of the response
Dampen the whole system
Reduce noisy input via a low-pass filter effect
I see arguments for all of these. I like the idea of serotonin reducing noise, because this fits the apparent use of SSRIs in reducing worry (noise from the future), rumination (noise from the past), and hypervigilance (noise from the present). On the other hand, a wider deadband fits really well with the unique efficacy of serotonergic meds for OCD, a disorder of repetitive thoughts and behavior that resembles a thermostat hunting its setpoint in repeated activation-deactivation cycles. But surely there are different effects at different receptors, and this is all complicated.
Critiques and Rejoinders
Emotional numbing is a symptom of depression, not medications.
This article wears its argument on its sleeve: “Antidepressants Do Not Work by Numbing Emotions.” The authors summarize three studies, all of which conclude that emotional blunting is better explained as a symptom of depression rather than a medication effect. Obviously anhedonia and flat affect are often part of depression, and these can improve with treatment.
The two big problems are that 1) people report persistent emotional blunting long after the depression has resolved and 2) non-depressed patients taking SSRIs for GAD, Panic Disorder, and OCD also report emotional blunting. This is hard to square with the view above. What about totally normal people? This small study weakly supports my claim. I say weakly since it involved 31 men filling out questionnaires and an indirect measure of serotonin, but the authors conclude that serotonin “exerts an inhibitory influence over both [Positive Affect] and [Negative Affect] in humans.”
On the other hand, this article finds that patients report some emotional blunting even on the antidepressant bupropion, which has no direct serotonergic activity. The rates are lower than patients on SSRIs, but still; this is evidence against my position.
About 40% of people don’t respond to SSRIs.
This is true. I think this is best explained by the well-accepted idea that Major Depressive Disorder is actually multiple conditions we can’t distinguish. For instance, if somebody is “depressed” due to an absence of good feelings rather than an excess of bad feelings, SSRIs would not help them in my model. Something like this definitely seems true, but it is hard to study.
Some people have a paradoxical response to SSRIs with worse depression, anxiety, mood swings, or irritability.
This is true. The output of even a single control system can be hard to predict, and the brain is an absurdly complex network of interdependent control systems. Also, if SSRIs add lag (phase shift) to an emotional circuit, this could easily cause instability. Hand-wave, hand-wave, shrug.
Why don’t SSRIs help mania, then?
The pathophysiology of bipolar disorder is different from unipolar depression. I still see all these conditions in terms of control systems, but perhaps for mania the main issue is excessive gain, so SSRIs would not help and may even worsen things.
OK, but this makes meds and psychiatrists look bad
To be clear: neo-Szaszians, the Mad in America crowd, and Scientologists are dumb. Of course people want to selectively ablate certain emotions, but the brain is complicated. TANSTAAFL.
Your theory still doesn’t explain why SSRIs take 4-6 weeks to have an effect
This is conventional wisdom about SSRI/SNRIs, but I am not sure it is true. It may be an artifact of study design and need for statistical power. In the real world, it is definitely the case that some patients respond to medication within a few days or a week. It happens all the time. Generally a subset of patients respond in 1-2 weeks but take longer for full remission. This meta-analysis concludes “Treatment with SSRIs is associated with symptomatic improvement in depression by the end of the first week of use, and the improvement continues at a decreasing rate for at least 6 weeks.”
If this is so obvious, why is it not the consensus?
It sort of is, as I’ve tried to show. However, most researchers view talk about emotions and feelings as invalid and prefer to drone on about delayed reward discount rates or whatever. I blame the dead hand of behaviorism. In addition, I suspect researchers and psychiatrists are reluctant to make the argument since “SSRIs actually do numb emotions” is unappealing, and the handful of researchers with this position are viewed as distasteful. Luckily I am a bold iconoclast.
Conclusion
In 2024, Psychiatry still can’t explain the mechanism of our most commonly used drugs. This is embarrassing! More importantly, it opens the door to widespread use of medications in situations where they don’t help people. I think we’re well past that point.
We need better understanding of our medications, and that requires high level models of emotions, feelings, and behavior in addition to details about receptors and genes. Sometimes reductionism only gets you so far. My view is that serotonin in the brain acts something like a rheostat for emotions. This is simplistic, but it’s also a coherent account that fits in a broader model of affective neuroscience. It’s the best explanation we’ve got.
What do you think about short-term recreational drugs that impact serotonin (e.g., MDMA, psychedelics)? These and particularly MDMA do not blunt emotion. Would you have to just say that serotonin has many roles and these modulate some other role?
It's more complicated than "serotonin inhibits emotions". SSRI do, yes. But serotonin acutely, itself? No, analogy from psychedelics and MDMA that suddenly increase serotonergic function - emotions are amplified massively! Only with consistent stimulation of serotonergic receptors by SSRI, after 2 weeks or so their sensitivity decreases and this coincides with therapeutic effectiveness. So does serotonin inhibit emotion? No, not really, it amplifies them, but if you overkill the system, then it shuts down, and emotion is reduced.