Why You Can't Sleep — and the Retraining That Actually Works
Sleep researcher Michael Grandner explains why chronic insomnia has essentially one engine — conditioned arousal — and why the fix is retraining rather than sedation. The conversation also covers how often sleep apnea hides behind 'stress' awakenings, what melatonin actually signals, which wearable numbers are worth reading, and how to use sleep as a performance lever.
Overview
Michael Grandner directs the Sleep and Health Research Program at the University of Arizona, and this long-form conversation moves from the clinic outward. He separates insomnia the complaint from insomnia the disorder, then makes the case that while short-term sleep loss has a thousand causes, the chronic version has one: a learned association between the bed and being activated.
From there he walks through the toolbox of cognitive behavioural therapy for insomnia — stimulus control, restricting time in bed, and the counterintuitive skill of surrendering control — with a string of analogies (the dentist's chair, the snow globe, the child who won't eat broccoli) that make the mechanics unusually easy to hold on to. A large middle section is devoted to sleep apnea: how common it is, why 'stress woke me up at 3am' is so often a breathing event, and what the options look like beyond CPAP. He then covers sleep architecture in detail, including what deep sleep and REM appear to be doing differently and why he thinks we dream at all.
The supplement segment is unusually candid — melatonin as a darkness signal rather than a sedative, why less is often more, and why good research on supplements is so scarce. On wearables he is blunt: trust the sleep-wake and heart-rate data, treat sleep staging as a fuzzy but useful picture, and largely ignore the composite scores. The closing third turns to performance — sleep banking, extending sleep in 15-minute steps, adolescent chronotypes and school start times — and finishes with a rapid-fire round of practical questions.
Key quotes
5Sleep is not something that you do. Sleep is something that happens to you when the situation allows for it.
The enemy of sleep is effort. If you're engaging in effort, you're adding energy into the system.
Melatonin is the hormone of darkness. It's a nighttime signal.
A bathroom scale is not a weight loss program.
You cannot caffeinate away complex decision-making. You just make bad decisions faster.
Key ideas
9Insomnia with a small i, insomnia with a capital I
Trouble sleeping and an insomnia disorder are not the same thing. The disorder means persistent difficulty falling or staying asleep, at least three nights a week for three months, with daytime consequences — and with adequate opportunity to sleep in the first place. A rough working threshold is roughly thirty minutes to fall asleep or thirty minutes awake in the night.
Chronic insomnia has essentially one engine
Acute insomnia has almost unlimited causes, because prolonging wakefulness under threat is a protective design feature. But the switch from short-term to chronic runs through conditioned arousal: the bed becomes predictably stressful, and the prediction itself produces the activation. By the time someone reaches the clinic, whatever started the problem is usually no longer relevant.
The bed can only mean one thing at a time
Stimulus control, first published in 1972, works on the same logic as a gym or a dentist's chair: a place where only a few things happen becomes predictable, and predictability primes the response. If bed means sleep, scroll, worry, work and TV in unpredictable order, the brain has nothing to predict — so the cue stops working.
Sleepiness and wakefulness are two separate dials
Sleep-wake isn't one line from awake to asleep; think treble and bass rather than mono. Sedatives push the sleepiness signal up hard enough to steamroll whatever is in the way, but in many people the sleep signal was fine and the wake signal was too high. That is why behavioural work often targets making you less awake rather than more sleepy.
'Stress woke me up' is often a breathing event
Grandner describes a common presentation: falls asleep fine, wakes in the middle of the night apparently from stress, can't get back down. Stress doesn't wake you up — the brain reads elevated heart rate, breathing and muscle tension after an arousal and fills in stress as the explanation. In his clinic that pattern raises the probability of a respiratory event above fifty-fifty.
Deep sleep sorts, REM connects
Slow-wave sleep appears to handle pruning and clearance — the day's material is sorted, the unimportant allowed to fade, and the spaces between brain cells widen so waste can clear. REM then takes what survived and builds connections, which is what dreams look like from the inside. Stage two, often dismissed as 'light sleep', is more than half the night and where most of the rest of the work happens.
It's the intermittency, not the hypoxia
In untreated apnea, oxygen usually dips only a few points before recovering — then dips again, dozens of times an hour. The cost is the repeated oxidative stress cycle rather than sustained low oxygen: cells spend the night putting out small fires instead of doing recovery work. Severe apnea in particular is a recognised risk factor for neurodegeneration.
Daytime light inoculates you against evening light
Morning outdoor light does three things: it starts a predictable clock roughly 16–17 hours before your sleep window, it raises the amplitude of the whole circadian curve, and — least known — it buffers you against light later in the day. When the system already knows what time it is, conflicting evening light carries much less weight.
Wearables: trust the layers unevenly
Wrist movement has estimated sleep versus wake at over 90% accuracy since the 1970s, and heart-rate optics are well developed — those are the layers Grandner reads. Sleep staging lands somewhere around 60–80%, which he calls a fuzzy but genuinely useful picture. The composite scores — readiness, recovery, sleep score — he largely ignores, because nothing tells him what went into them.
Practical takeaways
7- 1
Move the phone out of the bed, not out of the evening 17:00
Telling people to drop screens an hour before bed rarely survives contact with real life. Grandner's version: keep scrolling if you want, but do it standing next to the bed or sitting up on top of it, so the bed stays paired with sleep and you stay in touch with your body's signals.
- 2
If you wake in the night, take a break instead of a battle 21:00
Give yourself two or three minutes; if sleep isn't coming, get up, have some water, and return when you feel ready. The activation settles at its own pace like a shaken snow globe — you can slow it down by panicking, but you cannot speed it up.
- 3
Anchor the morning, not the bedtime 1:29:00
Fifteen minutes of outdoor light early is probably enough, thirty is better, and a walk or run makes it easy. Grandner puts the regularity on the morning precisely because getting into bed at a fixed hour when you aren't ready is bad stimulus control.
- 4
Delay the first coffee, and know your cut-off 1:30:00
Adenosine is at its lowest on waking, so caffeine has little to block — the lift most people feel is sleep inertia fading plus placebo. Waiting about an hour puts the peak, which arrives roughly 30 minutes after drinking, where it's useful; most people should stop four to six hours before bed.
- 5
With melatonin, less is usually more 1:49:00
Roughly half a milligram taken a few hours before your natural rise acts as a clock signal; three to five milligrams closer to bed is the sleep-promoting range. If you wake groggy, that's a sign the dose was more than you could clear overnight — halving it is the standard first move.
- 6
Extend sleep in 15-minute steps 3:15:30
Adding a full hour in bed can backfire if you can't fill it, since unfilled time in bed is exactly what creates arousal. Move bedtime earlier by fifteen minutes at a time — that's a block almost anyone can find, and the extra sharpness tends to buy the next fifteen.
- 7
Bubble-wrap the sleep you already get 3:25:00
A plain cloth eye mask and earplugs are, in Grandner's view, some of the best cheap sleep technology available — in one study an eye mask improved sleep consolidation and next-day test scores against a cut-out placebo. Sharing a bed with a restless partner or a pet is a bigger drag on sleep quality than most people assume.
Topics & chapters
15Opening: what actually goes wrong with sleep
How common sleep complaints are, an introduction to Michael Grandner's research programme, and how a clinician separates an ordinary sleep complaint from an insomnia disorder: frequency, duration, daytime impact and adequate opportunity to sleep.
Conditioned arousal and stimulus control
Why the original trigger stops mattering once the ball is rolling, and how a place with only one predictable outcome becomes a cue. The dentist's chair, the gym and the dining table that became a desk.
Restricting time in bed
The badly named 'sleep restriction': match time in bed to the sleep you can already produce, then expand. Explained through a child, a plate and twenty pieces of broccoli.
Sedatives, CBT-I and how common insomnia is
Why prescriptions are easier to obtain than therapy, what the pooled trial data show, and the roughly one-in-three / one-in-ten prevalence picture.
Sleep apnea: prevalence and the quiet signs
Why the upright human airway has a vulnerable 90-degree bend, why mild cases hide in fit young people, which awakening patterns raise suspicion, what home testing looks like, and what untreated apnea does to deep sleep and REM.
Sleep stages and dreaming
Stage one through REM, what each appears to be doing, and Grandner's own view on why we dream: the brain reading between the lines of the day.
Beyond CPAP
Mandibular advancement devices, myofunctional therapy, tongue stimulation, implantable options — and an even-handed take on mouth taping.
Advanced sleep hygiene: predictability and light
Building predictability when your schedule can't be regular, what blue-blocking glasses have to actually do, and the three ways morning light sets up the night.
Caffeine timing and the commute to tomorrow
Why the first coffee is mostly placebo, how to think backwards from tomorrow's wake time, and the university athlete programme that added over an hour of sleep without asking for it.
Melatonin: what it signals and how to dose it
Darkness hormone rather than sedative, the low clock-shifting dose, why over-the-counter bottles are stronger than the label, and the grogginess fix.
Supplements, THC, CBD and alcohol
Why good supplement research is so scarce, what magnesium, glycine and calming compounds do and don't do, and the rebound problems with THC and alcohol.
Late-night eating, shift work, naps and jet lag
The extra calories that show up after dinner when people are sleep deprived, the 'mind after midnight' window, napping as a snack versus a meal replacement, and how to brute-force a new time zone.
Wearables: what to trust and what to ignore
Movement and heart rate versus staging versus composite scores, orthosomnia, and how to actually read a night of data for signal.
Sleep as a performance tool
Resilience, adolescent chronotypes and school start times, sleep banking before competition, and the sleepiness questionnaire that predicted careers.
Rapid-fire questions
Fastest route to falling asleep, the best pre-bed routine, deep sleep, nighttime urination, how to tell if you're getting enough, and whether eight hours is really the number.