Explainer

Sensitivity beats level

One of the cleanest experiments in psychiatry gave two groups of women exactly the same hormones. One group's mood collapsed. The other group felt nothing. Nobody's levels were abnormal.

4 min readUpdated 2026-09-16

Almost everything written about brain chemistry is about levels. Low serotonin. High testosterone. A dopamine deficiency. The whole vocabulary assumes that the interesting difference between two people is how much of something they have.

The best-designed experiments in the field say otherwise. When researchers control hormone levels precisely — giving two groups identical doses — the groups still diverge. Same chemical, same amount, opposite effect. What differs is not the level but the sensitivity to a change in it.

That has a specific consequence for personality tests, and for anyone who has been told their problem is a deficiency.

The experiment that shows it

In the cleanest version of this study, researchers worked with two groups of women: one with premenstrual dysphoric disorder (PMDD), a severe cyclical mood condition, and one without.

First they shut down ovarian hormone production entirely, using a drug that suppresses the cycle. Both groups became hormonally flat. The PMDD group's symptoms went away.

Then, blind, they added the hormones back — oestradiol or progesterone, at ordinary physiological doses.1

Neither group had abnormal hormone levels at any point. The difference was entirely in how each group's brain responded to the hormones changing. A follow-up refined it further: the trigger is the change in level, not the level itself. Stable high hormones were fine. Stable low hormones were fine. Movement between them was not.2

The same shape keeps appearing

Once you know to look for it, the pattern recurs across unrelated literatures:

Dopamine drugs. Dopamine agonists given for Parkinson's disease cause impulse-control disorders — gambling, compulsive shopping, hypersexuality — in roughly 17% of patients taking them, two to three times the rate in patients who aren't.3 The striking part is the 83%. Same drug, same receptors, same doses; most people are unaffected and a minority's behaviour changes profoundly.

Serotonin depletion. Lowering the precursor that the body makes serotonin from increases aggression — but mainly in people who were already aggression-prone. In people low in trait aggression, the same depletion does comparatively little.4

Testosterone. Baseline testosterone correlates with aggression at about r = .07, which is close to nothing, and a preregistered trial in a thousand men found no effect of administration on economic behaviour.5 Effects show up under provocation, in men already high in dominance — context and sensitivity, not level.

Hormonal contraception. A Danish cohort of more than a million women found a small but real increase in later depression diagnoses, concentrated in adolescents.6 Most users are unaffected. A minority are affected considerably.

In each case the average effect is small and the individual effects are large. Reporting the average, which is what a "levels" framing does, describes almost nobody.

Why this breaks the deficiency story

Tests like the Braverman Nature Assessment ask you a few dozen questions and tell you which neurotransmitter you're deficient in. Set aside for a moment that no questionnaire can measure a neurotransmitter at all.7 The deeper problem is that the model is looking for the wrong variable.

If two people can have identical hormone levels and opposite responses, then knowing someone's level — even if you could measure it perfectly, with a blood draw and a mass spectrometer — would not tell you how they feel or behave. The thing that predicts their experience is their reactivity, and reactivity doesn't show up in a level at all.

What to do with this instead

The useful question isn't what am I low in. It's what moves me, and how much.

That question is answerable, and unlike a neurotransmitter level, you can answer it about yourself:

• Sleep. After a short night, how different is the next day — mood, patience, ability to start things? For some people this is the single largest lever in their life. For others it's a mild inconvenience.
• Caffeine and alcohol. Including the day after, which is where the interesting variance usually lives.
• Cycle phase, if it applies. The PMDD research is unambiguous that this is real for some people and not others, and that prospective daily ratings across two cycles is the actual diagnostic standard — not a recollection.
• Stress and provocation. How hard, and how long the recovery.
• Reward cues — food, spending, games — and whether their pull varies with hunger, tiredness or mood.

The honest summary

Individual differences in how people respond to their own biology are real, large, and consequential. They're also not what "high dopamine" or "low serotonin" describes.

The level framing survives because it's simple and sounds mechanistic. The sensitivity framing is harder to say in two words, but it's what the controlled experiments actually found — and it points at questions you can do something about.