"Rebuild your baseline" is everywhere and defined almost nowhere. Worth pinning down, because the vague version leads directly to the wrong interventions.
Think of a baseline as a tank and you try to refill it. Think of it as a lab value and you go looking for a test. Neither works.
What it isn't
It isn't a dopamine level, and nothing you own measures one.
Dopamine concentration in the brain isn't a single quantity with a resting value you could read off. It varies by region, by receptor type, and by timescale. Fast bursts and slower background activity do different jobs. There's no consumer device, wearable or app that measures any of it, and anything claiming to is either using the word loosely or selling something.
So when the word "baseline" is used well, it refers to a state of the system, not a chemical measurement.
Two kinds of dopamine signal
Most confusion about baselines comes from collapsing two signals into one.
In 1991 the neuroscientist Anthony Grace proposed that dopamine works in two modes. Tonic dopamine is the low, steady level that sits outside the cells all the time. Phasic dopamine is the brief burst that fires when something happens. His key idea was that the tonic level sets how strongly the system responds to phasic bursts. Background shapes the size of the spikes.
Phasic bursts are the part most people have heard about, usually wrongly. Wolfram Schultz's recordings from monkey dopamine neurons showed that bursts don't track pleasure. They track surprise: a reward that is better than expected produces a burst, an expected reward produces almost nothing, and a missing reward produces a dip below the background rate. That is a reward-prediction error, and it's a teaching signal, not an enjoyment meter.
Put those together and you get a sensible working definition. When people say "baseline", the honest version points at the tonic side of this picture: the background state against which every new event is scored. Not the spike after a notification. The floor the spike starts from.
What it does refer to: a setpoint
The most useful version of the idea comes from the allostasis literature: George Koob and Michel Le Moal's account of how reward regulation changes under chronic load.
The idea: reward function normally operates within a stable range, held there by opposing processes that push back against deviation. Under sustained load, those opposing processes stop returning to that range, and the system re-stabilizes somewhere else. They describe this as a chronic shift in the reward setpoint, with stress systems recruited to hold the new position.
A setpoint is a reference value the system defends. Not a fuel gauge. A thermostat setting.
That's what "baseline" should mean: the level your reward system is currently defending as normal. It's where you sit when nothing in particular is happening. Flat mornings are what a lowered setpoint feels like from the inside.
The opposing-process part has an older ancestor. Richard Solomon and John Corbit's opponent-process model says every strong feeling triggers a slower counter-feeling, and with repetition the counter-feeling grows while the original shrinks. The opponent process post covers that loop. Here it matters for one reason: a setpoint is what you get when the counter-process stops fully switching off.
Why the distinction has consequences
The thermostat framing predicts different things than the fuel framing does, and the predictions are testable against your own experience.
A setpoint is defended, so it resists change in both directions. One great day doesn't raise it. One bad day doesn't lower it. Only sustained load moves it, which is exactly why the change was invisible while it was happening.
A setpoint is the reference everything else is measured against. This is why good things stop registering. They're not smaller; they're being compared against a different number. The subjective flatness is a comparison artefact, not a report on the events. If that sounds familiar, the feeling flat guide collects the related reading.
A setpoint moves on the timescale of the load that moved it. Months down, months back. Which is why fixed-length challenges misfire: they use a refill clock for a re-adaptation problem.
A setpoint can be lowered by things that aren't pleasure. The allostasis account recruits stress systems, not just reward circuitry. Chronic stress and poor sleep count as load. You can be doing nothing "wrong" by the usual accounting and still be carrying plenty.
How you would actually measure it (and why nobody does at home)
Researchers who want to see the human dopamine system use positron emission tomography. A radioactive tracer that binds to dopamine receptors is injected, and the scanner shows how much binding is available in each region. Less available binding can mean fewer receptors, or more dopamine already occupying them, and teasing those apart needs extra steps.
That is a hospital procedure, not a wristband. And even PET gives receptor availability in one region on one afternoon, not a single "dopamine baseline" number.
So there is no at-home dopamine test, and here is what a self-check can and can't tell you. It can tell you how your days are going on the dimensions the research ties to reward state: sleep, effort, how much of your reward arrives free, how flat the mornings are. It can show direction over weeks. It cannot tell you your dopamine concentration, your receptor density, or whether you are "depleted". Anything that reports those from a questionnaire is guessing in a lab coat.
What the research shows
The evidence for a baseline as a setpoint is real, but most of it was gathered somewhere other than your phone.
- Two modes exist. Grace 1991 laid out the tonic and phasic distinction and argued that tonic levels regulate phasic responsiveness. Limit: it was a hypothesis paper about schizophrenia. Using it to tell a tidy "baseline versus spike" phone story stretches it, so treat the analogy as an analogy.
- Bursts encode surprise, not pleasure. Schultz, Dayan and Montague 1997 showed in monkeys that dopamine neurons fire to unexpected rewards, shift to the predicting cue after learning, and dip when an expected reward is withheld. Schultz 2016 reviews the same coding across humans, monkeys and rodents. Limit: simple cues in a lab, not a day of mixed inputs.
- Setpoints shift under chronic load. Koob and Le Moal 1997 and Koob and Le Moal 2001 describe reward thresholds rising and anti-reward systems being recruited. Limit: this is a model of drug addiction. Nobody has shown that phones produce the same allostatic shift, and this post doesn't claim they do. What transfers is the shape of the idea: a defended reference that moves slowly.
- Ordinary life can move receptor measures. Volkow et al. 2012 scanned 20 healthy adults with PET and found that one night without sleep lowered D2/D3 receptor availability in the ventral striatum, in step with how sleepy people felt. Dopamine release did not differ between conditions, so the authors read it as receptor downregulation. Limit: 20 people, one night of total deprivation, and an indirect inference. It is still clear human evidence that something as mundane as sleep shows up in this system.
- Opponent processes as a model. Solomon and Corbit 1974 proposed that repeated strong states breed a growing after-effect. Limit: a 1970s theory from animal and anecdotal data. Useful as a model, not a measured mechanism.
And the null that matters most here: no study has shown that a "dopamine detox" lowers or resets dopamine. The one peer-reviewed commentary on the practice (in Lifestyle Medicine, 2022) reports mixed, anecdotal outcomes and no dopamine data at all. The benefits people describe fit ordinary habit change and stimulus control. The dopamine fasting post goes through where that term came from.
What actually moves a setpoint
If the baseline is a defended reference shifted by sustained load, the levers are the ones that change sustained load. Nothing here is dramatic.
- Protect sleep first. It is the one lever with direct human PET evidence. A week of short nights is load you can remove without giving anything up.
- Change the share of free reward, not the total. A setpoint responds to what keeps arriving. Feeds, autoplay and delivery give reward with no cost in front of it. Adding a small cost to some of it changes the daily pattern without a heroic ban. The effort post explains why cost matters.
- Watch the cue, not just the craving. A drifting baseline makes cues pull harder. The wanting vs liking post covers why the pull can grow while enjoyment doesn't.
- Judge over weeks, never days. Defended systems resist single good days. Track direction across two to four weeks and ignore the daily noise.
- Count stress as load. A hard week at work belongs in the ledger next to the scrolling. Recovery time after it does too.
If you want a quick read on which of these is most out of line for you, the 60-second audit asks 12 questions and points at one lever. It is a self-report index, not a test.
How Myo uses this
Myo carries an explicit setpoint in its model. You log your day in plain sentences, and the model scores each entry: reward that arrives without effort feeds a chronic-load term, and that term pulls the setpoint down. Load accumulates faster than it clears, deliberately, so a bad week costs more than a good week pays back. The modelled tonic level then drifts toward the setpoint slowly, which is why recovery looks like drift rather than a jump.
What you see is direction: whether load is accumulating or clearing. It is derived only from what you logged. Myo does not measure dopamine, and the weights are Myo's modelling choices, labelled as such on the science page.
FAQ
Can you test your dopamine baseline at home?
No. There is no blood, saliva or urine test that reflects dopamine in the brain's reward circuits, and no wearable measures it. Researchers use PET scans, which show receptor availability in a brain region and need a radioactive tracer and a hospital scanner. At home, the honest option is a structured self-check of sleep, effort, mood and habits, read as a pattern over weeks rather than a chemical number.
What is the difference between tonic and phasic dopamine?
Tonic dopamine is the low, steady background level in the space around neurons. Phasic dopamine is a short burst fired by dopamine neurons when something unexpected happens, usually a reward that is better than predicted. Anthony Grace proposed in 1991 that the tonic level controls how strongly the system responds to phasic bursts. When people talk about a "baseline", the closest scientific match is the tonic side.
How long does it take for a dopamine baseline to recover?
Nobody has measured this for everyday habits like scrolling, so any precise number is invented. The setpoint model predicts that recovery runs on roughly the timescale of the load that caused it, which means weeks to months rather than days. Sleep is the exception with direct evidence: one lost night measurably changed receptor availability in a small PET study, which suggests sleep changes show up faster than most.
Is a low dopamine baseline the same as depression?
No. A lowered reward setpoint is a model of how reward regulation shifts under load. Depression is a clinical condition with its own diagnosis and treatments, and flat mood can come from many causes, including thyroid problems, medication and grief. If low mood, loss of interest or hopelessness last more than two weeks, talk to a doctor rather than a habit app.
Does a dopamine baseline actually exist?
Not as one measurable number. Dopamine varies by region, receptor and timescale, so there is no single resting value to read. What does exist, in the research, is a reward system that defends a reference point and can shift that point under sustained load. "Baseline" is useful shorthand for that defended reference, as long as nobody pretends it is a lab value.
Sources
- Grace, A. A. (1991). Phasic versus tonic dopamine release and the modulation of dopamine system responsivity: a hypothesis for the etiology of schizophrenia. Neuroscience, 41(1), 1 to 24. https://doi.org/10.1016/0306-4522(91)90196-U
- Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275(5306), 1593 to 1599. https://doi.org/10.1126/science.275.5306.1593
- Schultz, W. (2016). Dopamine reward prediction error coding. Dialogues in Clinical Neuroscience, 18(1). https://doi.org/10.31887/DCNS.2016.18.1/wschultz
- Koob, G. F., & Le Moal, M. (1997). Drug abuse: hedonic homeostatic dysregulation. Science, 278(5335), 52 to 58. https://doi.org/10.1126/science.278.5335.52
- Koob, G. F., & Le Moal, M. (2001). Drug addiction, dysregulation of reward, and allostasis. Neuropsychopharmacology, 24(2), 97 to 129. https://doi.org/10.1016/S0893-133X(00)00195-0
- Volkow, N. D., et al. (2012). Evidence that sleep deprivation downregulates dopamine D2R in ventral striatum in the human brain. Journal of Neuroscience, 32(19), 6711 to 6717. https://doi.org/10.1523/JNEUROSCI.0045-12.2012
- Solomon, R. L., & Corbit, J. D. (1974). An opponent-process theory of motivation: I. Temporal dynamics of affect. Psychological Review, 81(2), 119 to 145. https://doi.org/10.1037/h0036128
Myo models dopaminergic state from behavior; it does not measure dopamine. Not medical advice.