There's a version of this you've already read. Delete the apps, sit in a grey room, eat plain rice, don't listen to music. Seven days later your brain is supposedly clean and a walk in the park feels like a rave.
It doesn't work, and the reason it doesn't work is more interesting than the fact that it doesn't.
The chemical claim is the broken part
"Dopamine detox" contains a hidden premise: that dopamine is a substance that accumulates from overuse, and that abstinence drains it back down. Under that premise, deprivation is the treatment, and severity is the dose.
Neuroscientists have been consistent about this. Dopamine isn't a toxin, doesn't pool up, and isn't flushed by sitting still. It's a signalling molecule doing continuous work in motivation and learning — and abstaining from pleasant things does not "reset" its levels. A 2024 literature review concluded the practice remains scientifically unproven, and warned that extreme versions push people toward isolation, anxiety, and undereating. Harvard Health and the Cleveland Clinic have both published versions of the same correction.
The best evidence against the chemical reading comes from the person who coined the term. Cameron Sepah, the psychologist behind "dopamine fasting," has said repeatedly that the name was never meant literally — it was a memorable label for a cognitive-behavioural technique for reducing compulsive behaviour. The internet kept the name and threw away the method.
So the trend is a mistranslation of a reasonable idea. Which raises the real question.
Something is genuinely wrong, though
Nobody searches "dopamine detox" at 1am for fun. They search it because the phone opens itself. Because the fourth hour of scrolling produces less than the first minute did. Because things that used to be enough aren't, and the gap keeps widening.
That experience is real, and it has a name in the literature. It just isn't depletion.
What's actually happening: the second process
In 1974, Richard Solomon and John Corbit published an opponent-process theory of motivation. The claim is simple and, once you see it, hard to unsee.
Every affectively charged experience triggers two processes, not one. A fast process — the hit, the rush, the pleasure — and a slower opposing process that starts afterward and pulls in the other direction. What you feel at any moment is the sum of the two.
Then repetition does something asymmetric. The fast process weakens. The slow opposing process strengthens. Same stimulus, less high, more comedown. Solomon and Corbit used it to explain tolerance and withdrawal in drug use, and the same shape in ordinary acquired motivations — the skydiver whose terror becomes exhilaration, the routine that stops delivering.
That's the mechanism behind "it doesn't hit like it used to." Not less dopamine. More opponent.
And then the baseline moves
Twenty years later, George Koob and Michel Le Moal took the next step, and it's the one that matters most.
Opponent processes normally keep reward function inside a homeostatic range. In their model, under chronic load, those processes stop returning to that range. The system stabilizes somewhere new. They called it an allostatic state — a chronic deviation of the reward setpoint, sustained by both reward-circuit changes and recruited stress systems.
This is the part the detox framing can't express. The problem isn't that a tank is empty. It's that the floor has moved, and the floor is what you spend your day standing on. That's why a week of grim abstinence changes so little: a setpoint that took months of load to shift doesn't move because you had a bad weekend on purpose.
It also explains something detox gets exactly backwards. If your baseline is already depressed, stripping every source of reward out of your life adds load. Restriction isn't neutral.
Wanting and liking come apart
One more piece, and it's the one that explains the specific cruelty of the whole thing.
Terry Robinson and Kent Berridge's incentive-sensitization theory — first published in 1993, reviewed by them again in 2025 after thirty years — separates two things we normally treat as one. "Wanting" (incentive salience, dopamine-mediated) and "liking" (hedonic impact) run on dissociable systems. Sensitization amplifies the wanting without increasing the liking.
Which is the experience exactly. The pull gets stronger while the payoff gets weaker. You are not weak-willed for still wanting something that stopped being good. Those are two different systems, and only one of them sensitized.
So what actually moves a setpoint
If the problem is a setpoint under chronic load rather than a full tank, the intervention changes shape. Less "abstain from everything." More "change what the load is made of."
Three things carry most of the weight, and none of them is deprivation.
How the reward arrives matters as much as how big it is. Ferster and Skinner mapped this in 1957: variable-ratio reinforcement — unpredictable payoff per action, the slot machine structure — produces the highest response rates and the most resistance to extinction of any schedule. Your feed is a variable-ratio schedule. So is refreshing anything. Making one frequent unpredictable reward predictable will do more than deleting three predictable ones.
Effort changes what a reward does to you. Work from Salamone, Correa and colleagues reframed dopamine's role away from pleasure and toward effort and behavioural activation: accumbens dopamine depletion impairs animals' willingness to work for a reward while leaving their preference for it intact. Effort and reward are separable, and dopamine sits on the effort axis. The practical form: a reward you had to climb toward behaves differently from one that arrived because your thumb moved a centimetre. This is why "want less" is the wrong instruction and "earn more of what you get" is the right one.
Recovery is a real, slow variable. Setpoints move on the timescale they were moved on. Sleep is the highest-leverage input we have — sleep deprivation is associated with striatal dopamine-receptor downregulation — and it works by accumulation, not by heroics.
What we built
Myo is that model, made explicit.
Every activity you log becomes an event with a magnitude, a novelty, a reinforcement schedule, and an effort cost. The engine runs a fast process and a slow opposing process against each other, exactly as Solomon and Corbit described — the "felt" line you see is the first minus the second. Unprotected reward feeds a chronic-load term that drags a setpoint downward, which is Koob and Le Moal's allostasis expressed as arithmetic. Variable-ratio events cost more than predictable ones. Effort-gated events are treated as protective. Craving risk rises with accumulated load rather than with pleasure, because wanting and liking are different systems.
If you connect a wearable, HRV, resting heart rate and sleep enter as corroborating physiology — objective evidence about recovery. They are not a dopamine readout. Nothing is. Anyone selling you one is selling you the same broken premise in a nicer package.
Two honest notes. The specific coefficients are our parameterization, not measured constants — the structure follows the literature, the numbers are our model. And Myo isn't a clinical instrument; it's a tracker built on mainstream reward neuroscience instead of on a chemistry metaphor that was never true.
We publish the actual constants and label which parts are ours on the science page.
The short version
Dopamine detox asks you to want less. The literature says the reward system doesn't drain and refill — it moves a setpoint, under load, over months, and it moves back the same way.
You don't need a grey room. You need to know which of your rewards are unpredictable, which ones you actually worked for, and whether your floor is rising or falling.
That's a measurement problem. So we measured it.