The Science of Good Sleep

I keep staring at this old alarm clock on my nightstand, and it hit me — we set alarms to interrupt sleep, but nobody ever asks what sleep was doing right up to that moment.

That clock is actually the perfect prop, because the honest answer is — sleep was working. The brain, the heart, the lungs, metabolism, immune function, even mood — sleep supports all of it.

Working? I always pictured sleep as, like, the lights going off in a shop.

So did a lot of people, and that's the first idea worth retiring. The brain doesn't switch off. Its activity changes across the night, and at points it looks almost like it's awake.

Wait — almost awake?

That's REM sleep. But I'm getting ahead of myself. Let's build the night from the ground up, stage by stage.

Okay, so walk me through it. Suppose I fall asleep right now — what happens first?

Stage one, the doorway. Heartbeat slows, breathing slows, eye movements slow, muscles relax. It's the transition, and it's easy to be nudged back out of it.

And stage two?

Body temperature drops, eye movements stop, and the brain — which has generally slowed down — shows these brief bursts of electrical activity. Little flashes on a mostly calm readout.

Then deep sleep?

Stage three, and it's, it's the heavyweight. Longer stretches early in the night, hard to wake someone from, heart rate and breathing at their lowest. This is where the repair work happens.

Repair work — meaning what, exactly?

Think of it as maintenance. Sleep is involved in healing and repairing the heart and blood vessels. And in the brain, there's something remarkable going on.

Go on.

Mouse research found that sleep was associated with increased clearance of waste products from the brain — the space between brain cells actually expanded during sleep, which was linked to more efficient waste removal. Researchers proposed a brain-wide clearance pathway, later called the glymphatic system.

The brain cleans itself? Like a, what's the word, a self-washing oven?

Roughly, yes. And human studies since then reported that sleep-active processes were associated with overnight clearance of amyloid-beta and tau biomarkers into blood. It's an active research area, but the direction is exciting.

There's actually a clip I wanted to bring in here, because they put it beautifully. This is from the Sleepy Science for Sleep podcast.

Let's hear how they framed it. [[CLIP:2116439a-e516-4655-b590-298072b21ad8:Sleepy Science for Sleep — The Restorative Work of Sleep:22779]]

Okay, that's — I love the word they used. Reassuring.

Because it reframes everything. Sleep isn't lost time. It's the brain maintaining itself. Though one honest caveat — that mouse finding is mice, and the human picture is still being mapped.

Fair. Now you mentioned REM — where does dreaming fit?

REM's brain activity is close to wakefulness, with rapid eye movements and faster, less regular breathing. Most dreaming happens there, though dreams can occur in non-REM too.

Hold on — how long until the first REM stretch?

About an hour and a half after falling asleep. And the whole night cycles — non-REM, REM, non-REM, REM, again and again.

So one night is not one long sleep — it's loops. That means cutting the night short chops off the last loops.

Exactly. And that brings us back to the clock. So how much sleep do adults actually need? The National Sleep Foundation says seven to nine hours.

Seven to nine. And when people consistently fall short—

It compounds. Insufficient sleep makes it harder to concentrate and respond quickly — daytime performance, safety. And deficiency is associated with higher risk of high blood pressure, cardiovascular disease, diabetes, depression, obesity.

That's a longer list than most people expect.

It is. Sleep also helps the brain form and maintain the pathways needed for learning and new memories. One human study found the coordination between brain regions linked to sleep and memory differed between good and poor sleepers.

Okay, but here's my clock question again. What decides when we sleep? Is it just being tired?

Two processes, working together. Circadian rhythms — roughly twenty-four-hour cycles influencing sleep timing, body temperature, metabolism, hormone release. And sleep-wake homeostasis.

Say that second one in plain words?

Sleep pressure. The longer someone stays awake, the more it builds. After sleep loss, people sleep longer and more deeply. Like a debt the body collects on.

And light — does that matter?

A lot. Light is an important influence on the circadian clock and helps shape when people feel awake or sleepy. Which is why evening screen light and blackout curtains are real levers, not folklore.

So the clock on the nightstand is fighting the clock in my head.

Nicely put. And your habits — light, timing, consistency — are the small daily levers that genuinely shift sleep quality. No drastic overhaul needed.

One last thing — how did anyone even discover these stages? Somebody had to watch people sleep.

In nineteen fifty-three, Eugene Aserinsky and Nathaniel Kleitman reported observations of rapid eye movements during sleep, which helped establish REM as a distinct stage. Someone literally watched eyelids, all night.

Watching eyelids for science. And now we know the brain is down there, cleaning, filing, repairing.

Which is the whole reframe. The alarm clock doesn't interrupt nothing — it interrupts a shift that was mid-task. Tonight, maybe let it finish one more loop.

That streetlamp outside my window has been flickering all week, and I can't decide if it's my enemy or just... decoration.

Then let's stage a proper argument. Position one — evening light is basically harmless. Position two — darkness is essential. And I'll hold both sides honest.

Some people say they fall asleep fine with some light on, so they think light can't matter.

That is a common version of the argument, and it deserves consideration. Falling asleep is not the same as what light does to the clock afterward. The brain's central clock sits in the hypothalamus — the suprachiasmatic nucleus — and it takes its cues from light hitting the eyes.

Wait, so I can be asleep and the clock is still watching the light?

Essentially, yes. Special cells in the retina contain melanopsin, and they respond to light beyond what rods and cones handle for vision.

Through closed eyelids?

One review found a study where just five to ten lux — barely a nightlight — produced a measurable circadian response in people sleeping with eyes closed.

Five lux. Okay, but the harmless camp would say — cite the study, name it, or I'm not moving.

Fair demand. For example, the two-week inpatient study where people read on an iPad before bed — they fell asleep about five minutes later than with printed books. Five minutes. That's the harmless side's best evidence.

And here's Matt Walker on The Matt Walker Podcast with the numbers behind exactly that argument. [[CLIP:b1ea2885-bb73-4261-bfff-9f1168dccb0b:The Matt Walker Podcast — Matt Walker:39837]]

Ten minutes — so modest, but real. And you're saying the real culprit might be the overhead light, not the screen.

So... the light-is-fine people win?

Not so fast. Same research world, another controlled finding — typical room light before bedtime delayed melatonin onset by roughly thirty minutes, and in the majority of participants.

And melatonin is the, what's the word, the evening hormone —

The one that normally rises as it gets dark and is associated with sleepiness. Now, another room-light study found melatonin suppressed by more than half in most trials. That's not subtle.

So both findings are real. Five minutes later to sleep, but the chemistry takes a real hit.

And timing decides direction. Light in the evening tends to delay the clock; light in the biological morning tends to advance it. Which means the same lamp at nine p.m. and nine a.m. are different creatures.

Give me the everyday version of that. Suppose someone works late, scrolls in bed, then sleeps in on weekends—

One laboratory study linked a later last exposure to light above ten lux with later sleep onset and shorter sleep. It's an association, not proof, but the pattern is the one people live every week.

Okay, then the darkness camp's heaviest evidence — because that IARC thing unsettled me.

The International Agency for Research on Cancer classified night-shift work as probably carcinogenic to humans. Limited human evidence, but strong mechanistic evidence in animals. And it does not mean every shift worker gets cancer.

Erin O'Hara on The Sunday Session with Francesca Rudkin podcast laid out just how far those consequences reach. [[CLIP:b6322498-a7fb-466b-abf0-fc53413d0007:The Sunday Session with Francesca Rudkin — Erin O'Hara:23641]]

Eight to twenty percent higher cardiovascular risk — that's the darkness camp's heaviest artillery. But wait, why would shift workers who adapt still show problems?

Why would shift workers who adapt still show problems?

Because they alternate. Night shifts, then normal days off — the clock keeps getting dragged between schedules, and melatonin production keeps getting disrupted. NASA saw the same problem in space — dim interiors, no windows, disrupted day-night cycles — and ran melatonin-suppression trials with different light wavelengths as a countermeasure.

So entrainment — resynchronizing — how fast can that even go?

Slower than people assume. One study found the phase advance from morning bright light was less than half as large after six-hour sleep opportunities compared with nine-hour ones. A sleep-deprived clock barely moves.

So the light-deprived, sleep-deprived person can't even fix their clock quickly.

Right — and morning daylight helps synchronize the timing; NIH guidance recommends getting outside during the day, and a dark bedroom at night. Bright light for alertness by day, darkness for sleep by night.

So which claim wins — harmless evening light, or essential darkness?

In everyday life, darkness wins, on the melatonin evidence — suppressed by more than half in most trials, delayed onset, shortened duration. But it's not a blanket ban — blue-depleted LEDs of equal brightness reduced suppression in one controlled study of sixteen adults. Color, intensity, timing, all matter.

So my flickering streetlamp — enemy or decoration?

An enemy with a small army. Draw the curtain, dim the room, and let the clock tick toward midnight in the dark it was built for.

I watched a street sweeper start its route this morning. One slow pass, and it moved on — and the street behind it was still, still kind of gray. It made me think about sleep cycles.

Hold that sweeper. Because the myth I want to dismantle today is hiding in a number everyone quotes — ninety minutes.

That's the number, right? One sleep cycle equals ninety minutes, so plan your sleep in those blocks.

No. Authoritative summaries describe typical cycles as roughly eighty to one hundred minutes — about ninety on average.

So it's an average, not a setting.

And it drifts within a single night. Early cycles average about seventy to one hundred minutes. Later ones often run ninety to a hundred and twenty.

So a night isn't four identical laps — it's laps of different lengths.

Exactly. The NHLBI says people usually have four to six cycles a night. And the laps aren't identical in content, either.

Meaning what changes between laps?

The mixture. Deep sleep — N three — dominates early. REM stretches grow longer as the night goes on. First REM might be five to ten minutes; by morning it's much longer.

Wait, the first REM is that short?

Five to ten minutes, typically. It usually first appears about an hour to an hour and a half after falling asleep — which is probably where the ninety-minute folklore comes from.

So people took one true number and turned it into a timer.

Right. Near the end of the night, sleep is mostly lighter stages and REM — not an even split. That means one cycle cannot represent the whole night's pattern.

Now — back to my sweeper. Why does one lap not clean the street?

Because the heavy cleaning happens in one neighborhood. Deep sleep, the slow-wave stage, sits in the early part of the night. That's when the brain's waste-clearing work is at its busiest.

Here is how Anshul Gupta on the Reversing Hashimoto's - Heal The Thyroid podcast described what that overnight cleaning actually looks like. [[CLIP:bc6094e2-6ea5-472d-b325-c040e370945b:Reversing Hashimoto's - Heal The Thyroid — Anshul Gupta:33228]]

A sixty percent expansion and a near-total shutdown when we're awake — no wonder one early lap barely dents the workload.

And one cycle gets you —

One early lap. Mostly light sleep, a sliver of deep work, a whisper of REM. If you shorten the night, the sweeper might not reach the far streets — the long REM neighborhoods of early morning.

Hmm. But why does a ninety-minute nap feel refreshing if it barely cleans anything?

Because it pays down the immediate pressure. Sleep-wake homeostasis — the debt that builds while awake — gets its first installment. Alertness recovers before the deep overnight work does.

So a nap is like the sweeper clearing the main road.

Nice reuse of the image. And for adults eighteen to sixty, the NIH recommends at least seven hours — counting blocks isn't a substitute for duration.

What about people who wake between cycles? My partner does that constantly.

Completely normal. People may briefly wake between cycles — the night isn't one uninterrupted, identical sequence. That's not failure.

But fragmented sleep is different, isn't it?

It is. Older adults tend to have less deep slow-wave sleep and more fragmented sleep — which cuts into exactly the stage where the heavy cleaning happens.

So the street gets cleaned less, night after night.

And the stages themselves are measured, not guessed — brain activity and eye movements recorded in sleep studies, scored by the American Academy of Sleep Medicine's manual. Feeling rested is not the metric.

One thing I love from the history — Aserinsky expected no eye movements, and found the opposite.

He was observing sleeping infants, then recorded his own infant son in the lab. The recordings showed recurring bouts of vigorous eye activity nobody was looking for.

The data cleaned up the assumption. Like a sweeper.

And that's the honest rule for tonight — let the route finish. Four to six laps, uneven laps, each one doing different work.

And by sunrise... the street actually glistens.

The sweeper made all its passes. That's what a whole night buys.

I watched a guy on the train this morning, dead asleep, eight-ish a.m., coffee untouched. He looked like he'd slept a full night — and looked wrecked.

And I'd bet he'd tell you he got his eight hours. That's exactly who today's myth belongs to.

The myth being — eight hours in bed means rested.

I’ll give its strongest version, because it's not silly. The American Academy of Sleep Medicine and Sleep Research Society recommend at least seven hours regularly.

So the number isn't wrong.

No — it's just a floor, not a finish line. Their consensus explicitly says routinely sleeping six hours or less is inappropriate for optimal adult health, while healthy sleep also depends on quality, timing, regularity, and the absence of sleep disorders.

And that word "timing" is where biology really digs its heels in — here's Lisa Zawrotny on the Positively Living®: Shame-Free Productivity Conversations podcast on why when you sleep is hardwired. [[CLIP:8d97024b-0c33-4f61-b593-afb1958f85e6:Positively Living®- Shame-Free Productivity Conversations — Lisa Zawrotny:30877]]

So the clock on the wall and the clock in your genes aren't always telling the same time — which brings us back to your train guy.

Okay, but my train guy had the time in bed. So what broke?

The NIH defines sleep deficiency as more than a short night. It can mean sleep at a biologically unsuitable time, poor-quality sleep, or too little time in particular stages.

Give me a concrete case.

Sleep apnea. Breathing stops briefly or goes very shallow, repeatedly. Those pauses can last ten seconds up to two minutes, twenty or thirty times an hour.

Thirty times an hour.

And the NHLBI guide says people with apnea may bounce from deeper to lighter sleep all night and rarely get enough deep, restorative sleep. So time in bed overstates everything.

So eight hours in bed is a receipt, not a guarantee.

Exactly. And loud snoring, observed pauses in breathing, marked daytime tiredness — those are reasons to talk to a health care provider. That's the polite thing this conversation usually leaves out.

Now — what about people who genuinely sleep less, but well? Doesn't quality beat quantity?

Careful. The restriction evidence is brutal. Healthy adults limited to four or six hours for fourteen nights showed cumulative, dose-related cognitive deficits. Six hours or less matched up to two nights of total deprivation.

But didn't they feel it coming?

That's the chilling part. Sleepiness ratings barely rose after the initial day, while performance kept falling. They couldn't feel how impaired they were.

And weekend catch-up sleep — surely that resets the—

It doesn't. A review concluded one or two long nights can't fully restore the effects of chronic restriction. In one study, participants slept about an hour longer on the weekend, and it still didn't prevent metabolic disruption once restriction resumed.

And timing cuts both ways — listen to Trent Nikolic on the 2GB Weekends podcast on what late bedtimes do to metabolic health. [[CLIP:dd5b8aa5-10cf-4234-b3ef-16870bede5db:2GB Weekends — Trent Nikolic:30485]]

Same food, different schedule, worse outcomes — it really is about when, not just how much. So what actually moves the needle?

So what actually moves the needle?

Regularity, startlingly. A large prospective cohort analysis found greater sleep regularity associated with lower mortality risk — and regularity predicted all‑cause mortality more strongly than duration did.

More than duration. Wow. Any proof in the other direction — that more good sleep helps?

Cheri Mah at Stanford followed collegiate basketball players through a five‑to‑seven‑week extension, aiming for ten hours in bed. Actual sleep rose about a hundred and eleven minutes on average.

And the results?

Faster sprints, better reaction time, less sleepiness, better mood — and free‑throw and three‑point shooting each improved by about nine percentage points.

So if I'm choosing — quantity, quality, or timing?

The choice that survives every objection we raised: a regular seven‑plus hours, at a biologically suitable time. Duration without regularity lost to regularity itself.

Poor train guy. Eight hours of receipts, but not fully banked.

And as the doors close on him — may his tonight be long, dark, on schedule, and quietly breathing.

I made an espresso at two in the afternoon yesterday, felt amazing, and then slept like a broken umbrella.

Then you lived the whole puzzle in one day. Two o'clock coffee, midnight wreckage.

But everyone says caffeine wears off in a couple of hours.

That's the myth, and it deserves a fair hearing before we take it apart. Some people genuinely feel fine.

Okay, build their case for me.

Feeling alert after coffee is real, and morning caffeine rarely seems to hurt. So the claim goes — drink it early, it's gone by night.

And where does that case come apart?

At the half-life. Caffeine's average elimination half-life in healthy adults is about five hours.

So my two p.m. espresso...

At eight p.m., roughly half of it is still circulating. Feeling alert doesn't establish lost sleep was repaid — it's a feeling, not a ledger.

Half. Wait — how does it even make me alert? What's it doing in there?

Caffeine blocks adenosine receptors in the brain. Adenosine builds up during waking and contributes to sleep pressure — the debt we talked around before.

So coffee doesn't erase the debt. It, what's the word, mutes the alarm?

Beautifully put. The blockade promotes wakefulness, but adenosine is still there, stacking up underneath.

Here's how Andrew Huberman on the Huberman Lab podcast explained that exact mechanism. [[CLIP:fa05f672-1583-4139-989e-d6079a2381c4:Huberman Lab — Andrew Huberman:36127]]

So the alarm really is just muted — the sleepiness signal never gets through while caffeine is parked in that slot.

Which is why the afternoon crash hits even after morning coffee?

That's the honest reading. The pressure keeps building; the blocker just runs out of steam by mid-afternoon.

Prove it with a real case, not just chemistry.

Christopher Drake and colleagues, twenty-thirteen. A randomized crossover study — four hundred milligrams at bedtime, three hours before, or six hours before, versus placebo, sleep measured at home.

Six hours before bed still disrupted sleep?

Significantly. That dose taken six hours out cut total sleep time by about forty-one minutes compared with placebo.

Forty-one minutes... that's a whole early REM stretch gone.

Exactly the part of the night that does heavy lifting. And a later crossover study found that same dose caused greater sleep fragmentation when taken within eight hours of bedtime.

So the crash at three p.m. — people fight it with another coffee, and then?

Then the evening cup delays sleep initiation, the night fragments, and tomorrow's fatigue is worse. Coffee to fight the slump is borrowing from tomorrow.

And if they skip it instead?

Withdrawal. A systematic review validated headache, fatigue, drowsiness, difficulty concentrating — typically starting twelve to twenty-four hours after stopping.

From small habits? Like one strong cup?

The review found symptoms could follow abstinence from daily doses as low as one hundred milligrams. That's roughly one average cup.

So even quitting cold turkey has a price.

Though expectations play a role — one study found headache increases were significant mainly among people who could tell caffeinated from decaf.

And dose matters too, right? Is any amount safe late?

In that later study, a one-hundred-milligram dose didn't significantly affect objective or subjective sleep. Amount and timing both decide the outcome.

Does everyone clear it the same way?

No. It's metabolized in the liver by the enzyme CYP1A2, and effects vary considerably between people. Smoking speeds it up; late pregnancy stretches the half-life much longer.

So my neighbor's 'coffee doesn't bother me' is, it's, it's their liver talking, not the rule.

Precisely. One person's experience is not another's sleep response.

So if I wanted to actually settle this for myself, what would convince me?

The same thing that convinced researchers. Watch the clock strike three — then skip the afternoon cup for a week and time your mornings.

The espresso machine and I need to have a talk.

Have it before noon.

My phone screen dimming at eleven p.m. last night — that glow fading — it got me thinking. What if someone tracked a whole month of changing just one habit?

Then let's do exactly that. Let's follow a participant through a thirty‑day reset. One change at a time, when the month calls for it.

Okay, day one. Maya's checking her phone in bed at eleven. What's that doing?

Bright artificial light in the hour before bed can signal wakefulness. The NHLBI recommends making that hour quiet and less stimulating — so her screen is arguing with her body clock.

Right at the moment melatonin should be rising.

Exactly. And the American Academy of Sleep Medicine advises turning devices off at least thirty minutes before bedtime. The first, and really the main, big ask.

And Maya isn't alone in this — here's the Enlightened podcast on exactly what that late-night screen glow does to us. [[CLIP:3c30b311-d7be-40e5-aa5a-84a91a811e73:Enlightened — Blue light, screens, and delayed sleep:39523]]

"One check of one message" turning into 1:47 a.m. — that's her day one in a nutshell. So where does she even start anchoring this reset?

Wait — what's her anchor? Bedtime or wake time?

Wake time. The practical reset starts by choosing a wake time she can keep consistently, then working backward to allow enough hours in bed.

Enough meaning—

Seven hours minimum for adults eighteen to sixty, per the CDC. Individual needs vary, but that's the floor.

So she fixes her alarm, keeps weekday and weekend the same... that alone moves the clock?

A steady bedtime and wake time supports a regular sleep‑wake rhythm. The NHLBI specifically says keep the schedules similar across the week.

But she's not sleeping more hours at first. Why would she feel different?

Because alignment is its own currency. Same wake time, morning daylight — the NIH notes light can shift the clock earlier or later depending on when it hits. Daytime walks outdoors are literally clock‑setting.

And there's even more to that daylight habit — here's Eric Berg on the Dr. Eric Berg's Podcast on how sunshine builds melatonin. [[CLIP:f5d672fc-f1d5-449e-b82a-25c6401c3231:Dr. Eric Berg’s Podcast — Eric Berg:36519]]

So her morning walk isn't just a mood boost — it's literally stocking the shelves for nighttime sleep. So when does she actually start feeling the difference?

So week two, Maya's walking outside each morning. When does the daytime alertness actually shift?

Regular schedules are what reduce that groggy, sleep‑inertia feeling in the morning. Not one heroic early night — repetition.

Give me her week four versus day one.

Day one — wired at eleven, heavy at seven a.m., reaching for coffee before her eyes open. Week four — the quiet pre‑bed hour, a cool, dark, quiet bedroom, and mornings that don't feel like a fight.

And the setbacks? A month can have bumps along the way.

True. A late party, one glass of wine too close to bed — alcohol eases falling asleep but means lighter sleep and night wakings, per NHLBI guidance.

And when sleep just won't come?

Then sleep specialists say get out of bed. Read something calm, listen to quiet music, return when sleepy.

That feels counterintuitive — leaving the bed to fix the bed.

It protects the association between bed and sleep. Also on the list — no big meals late, limit fluids near bedtime, exercise thirty minutes most days but not vigorously right before bed.

One honest worry, though. What if hygiene habits aren't enough?

Important caveat. An AASM guideline strongly recommends cognitive behavioral therapy for insomnia for chronic insomnia — and advises against sleep hygiene alone as treatment. Habits help; they're not medicine.

So the cascade — one screen habit removed—

—opens the door for a quiet hour, a cool dark room, a steady clock. And her memory work benefits too — consolidation likely involves both non‑REM and REM sleep, cycling several times a night.

So tonight, when the screen glow fades a little earlier...

That's the whole reset, visible in one ordinary evening — a phone going dark at ten‑thirty, and a body that finally gets the signal.