Sleep Is a Hormonal Event

The popular framing of sleep is that it's rest — a passive state where the body recovers and the brain recharges. This is wrong by a large margin. Sleep is one of the most hormonally active states a human body enters, arguably more active than daytime waking in terms of the absolute rate of endocrine signaling. Growth hormone, testosterone, cortisol, insulin, prolactin, leptin, ghrelin, melatonin, thyroid hormones, and sex steroids all have their daily pattern defined by sleep architecture. Miss enough of it, and every downstream system — muscle, metabolism, mood, immune function, libido — degrades in measurable ways, and the degradation shows up within a week.

The reason "just sleep more" is the boring first answer to almost every hormonal complaint is that it is the right answer more often than any of the interesting alternatives. Men walk into TRT clinics with testosterone at 300 ng/dL who are sleeping six hours a night; the sleep alone is the diagnosis in a substantial fraction of those cases. Women with irregular cycles and low energy who are up at 2 a.m. scrolling; half of them don't need a hormonal workup, they need a sleep schedule. Elite lifters taking stacks of supplements to drive recovery while sleeping five and a half hours; the supplement stack isn't the problem.

This article lays out, system by system, what sleep actually does to your hormones, how fast it breaks down when you underslept, what the evidence-backed levers are, and why the lifestyle interventions that actually move the needle are not the ones being marketed.

The orchestra, by stage

Sleep is not a single state. It is a structured sequence of stages, each with distinct neurophysiology and distinct hormonal consequences. A normal night cycles through these stages four to six times, with the proportions of each stage shifting as the night progresses.

N1, N2 (spindle memory consolidation)

N1 is the transition from waking to sleep — a few minutes of drifting, easily interrupted, not much hormonal activity. It is the doorway, not the room.

N2 is light sleep, the stage where the characteristic sleep spindles appear on EEG. Spindles are brief bursts of 11-16 Hz activity, lasting less than a second, generated by the thalamocortical system. Their role in memory consolidation — particularly declarative, hippocampus-dependent memory — is one of the better-established findings in sleep neuroscience.

Adults spend roughly 45-55% of total sleep time in N2. Hormonally, N2 is relatively quiet: no large growth hormone release, no major cortisol changes. But this stage matters for another reason — the gateway to N3, the deepest and most hormonally active stage. Fragmented sleep with frequent arousals (from alcohol, from sleep apnea, from a partner snoring) tends to reduce time spent consolidating in N2 and reaching N3.

N3 slow-wave (growth hormone, glymphatic clearance)

N3 is slow-wave sleep, sometimes called deep sleep or delta sleep for the 0.5-4 Hz waves that dominate the EEG. N3 is the most metabolically and endocrinologically active stage of sleep.

Growth hormone is released in its largest single pulse of the day at the onset of N3. In young adults, this pulse represents 60-70% of total daily GH secretion. The pulse is tightly coupled to the first appearance of slow-wave activity; if N3 is disrupted or delayed, the GH pulse is attenuated or shifted.

Glymphatic clearance — the brain's system for clearing metabolic waste, including beta-amyloid — is most active during slow-wave sleep. The spaces between neurons actually expand by ~60% during N3, allowing cerebrospinal fluid to flush through brain tissue and carry away metabolic byproducts. This is believed to be one of the major functional purposes of sleep, and chronically reduced N3 is associated with accumulation of waste products that may contribute to neurodegenerative disease.

Adults spend 13-23% of total sleep time in N3, with the majority occurring in the first third of the night. N3 declines with age — elderly adults may have almost none, which is part of why the sleep architecture of older adults is so different from younger ones.

Alcohol, certain sleep medications, and (importantly) older classes of sleep drugs like benzodiazepines suppress N3 while preserving or even lengthening total sleep time. A person "sleeping eight hours" on alcohol has a very different hormonal experience than a person sleeping eight unmedicated hours.

REM (hormonal + emotional processing, testosterone peak)

Rapid eye movement sleep is the most neurophysiologically distinct stage — brain activity resembles wakefulness, the body is paralyzed (except eyes and diaphragm), and dreaming is most vivid. REM dominates the last third of the night, with most of it concentrated between 4 and 7 a.m.

Testosterone peaks during REM in adult men. The morning testosterone peak that most people know about is a REM-driven peak; men who don't get full REM cycles (short sleepers, shift workers) have attenuated morning T. The first REM cycle of the night produces a modest T rise; subsequent REM cycles each produce larger T pulses, culminating in the predawn peak.

Memory processing, particularly emotional memory and procedural memory, is REM-dependent. The "sleep on it" effect that people describe for emotional processing is not folk wisdom; it's a real phenomenon tied to REM activity.

REM density — the amount of REM relative to total sleep time — is around 20-25% in healthy adults. REM is what gets cut first by short sleep, because it is concentrated at the end of the night: sleeping six hours instead of eight does not cost you 25% of your REM proportionally; it costs you 40-50% of your REM absolutely, because you sleep through mostly N2 and early N3 and wake up before most REM occurs.

This is why "I only need six hours" is a particularly expensive false economy hormonally. The two hours you're skipping are the two hours with the highest density of testosterone and REM-dependent memory processing.

The hormones that depend on sleep

Testosterone — the week-long sleep-restriction data

The best-known study on sleep and testosterone is Leproult and Van Cauter, JAMA 2011. Ten healthy young men (average age 24), living in a sleep lab, were restricted to five hours of sleep per night for one week after a normal-sleep baseline. Total testosterone was measured across 24 hours at baseline and after the restriction period.

After one week of five-hour nights, total testosterone fell by 10-15%. The authors noted that this magnitude is roughly equivalent to 10-15 years of normal aging, compressed into seven days. The effect was persistent throughout the day, not just at the testosterone nadir.

Subsequent studies have extended this picture. Men with chronic partial sleep restriction (6-6.5 hours) have lower morning testosterone than men sleeping 7-9 hours, in both cross-sectional and intervention studies. Obstructive sleep apnea is one of the strongest known drivers of secondary hypogonadism in men; treating the apnea (CPAP, weight loss) raises testosterone in many cases without any direct hormonal intervention.

The mechanism appears to be disruption of the pulsatile LH release from the pituitary, which in turn depends on the gonadotropin-releasing hormone (GnRH) pulse generator in the hypothalamus. GnRH pulsatility is tuned by sleep architecture; restrict sleep enough, and LH pulses decrease in amplitude and frequency, which translates into reduced testicular testosterone output.

One important practical consequence: a man with testosterone at 350 ng/dL who is sleeping six hours a night has not been adequately evaluated for hypogonadism. The sleep needs to be fixed first or at least concurrent with any other workup, because the number he's measuring is plausibly being depressed by 10-20% from sleep alone. "Fix the sleep, retest in six weeks" is the correct first move, not "start TRT."

Cortisol — CAR and its inversion in shift workers

Cortisol follows a sharp diurnal pattern in a healthy adult: lowest around midnight, a steep climb starting around 4 a.m., the cortisol awakening response (CAR) spiking in the first 30-45 minutes after waking, then a gradual decline through the day. The CAR is a 50-75% rise above pre-waking baseline in a healthy person. It is thought to be part of the system that mobilizes metabolic resources for the coming day.

Visualizer

Cortisol diurnal rhythm

Wake time: 7am

Schematic only — the shape of the diurnal rhythm, not measured concentrations. Individual curves vary with sleep timing, stress, and HPA-axis health.

The CAR is dependent on normal circadian entrainment. Shift workers, jet-lagged travelers, and people with chronic sleep misalignment lose the characteristic shape of the cortisol curve — flattened CAR, elevated nighttime cortisol, sometimes inverted patterns where cortisol is highest in the afternoon. The endocrine consequences include altered glucose metabolism, altered mood regulation, and impaired immune function.

Chronic partial sleep restriction also raises evening cortisol. Leproult and colleagues showed that one week of sleep restriction elevated evening cortisol by 37-45%, which is a substantial chronic stress signal. This is part of why sleep-deprived people often report feeling "wired" at night and have difficulty falling asleep — the cortisol curve that should be bottoming out is instead rising.

Shift work specifically has been classified by the IARC as probably carcinogenic (Group 2A), largely based on evidence of increased breast and prostate cancer rates in long-term shift workers. The mechanism is thought to include chronic circadian misalignment of hormones including cortisol, melatonin, and sex steroids. This is not a minor health issue; the combined effect on endocrine systems over years or decades of shift work is substantial.

Insulin — muscle-vs-fat sensitivity after one bad night

Insulin sensitivity breaks down faster than almost any other hormonal axis after sleep restriction. One night of four-hour sleep produces a 25-30% drop in whole-body insulin sensitivity in healthy young adults, measured by hyperinsulinemic-euglycemic clamp.

The effect on tissue-specific insulin sensitivity is more interesting. Broussard and colleagues at the University of Chicago showed that four nights of 4.5-hour sleep produced a 30% reduction in insulin sensitivity specifically in adipose tissue, with muscle insulin sensitivity dropping too but more modestly. The differential matters: adipose tissue that is insulin-resistant keeps releasing free fatty acids even when insulin is high, which drives hepatic insulin resistance, which drives higher glucose output, which eventually contributes to the metabolic syndrome pattern.

Cross-sectional data support this: chronic short sleepers (under 6 hours) have roughly 1.5-2x the risk of developing type 2 diabetes compared to 7-9 hour sleepers, adjusted for weight and other factors.

For athletes, the relevant finding is that sleep-restricted muscle takes up amino acids and glucose less efficiently. Training hard on a chronically sleep-restricted schedule is literally wasting a substantial fraction of the training stimulus because the downstream anabolic response is attenuated.

Growth hormone — why weight training without sleep is wasted

Growth hormone's largest daily pulse occurs at the onset of N3 slow-wave sleep, as discussed above. In chronic sleep restriction, slow-wave sleep is reduced, delayed, and often more fragmented, and the GH pulse is correspondingly diminished.

For muscle protein synthesis and body composition, this matters. GH and its downstream mediator IGF-1 are not the dominant drivers of hypertrophy — testosterone and mechanical tension are more important — but they play a meaningful role, particularly in recovery and in the handling of injury. Chronic sleep restriction produces a state of suppressed GH output overlaid on suppressed testosterone and impaired insulin handling, all of which combine to reduce the adaptive response to training.

This is one of the most consistent findings in sports performance research: athletes who extend sleep (typically to 9-10 hours per night for a period of weeks) show measurable improvements in performance, reaction time, recovery, and hormonal markers. The effect sizes are comparable to or larger than most legal performance supplements.

The practical consequence: weight training on five hours of sleep is not training at 100% with a 20% recovery hit. It is training at something like 70% of stimulus capture, with all the endocrine levers for adaptation partially closed. A person sleeping 5-6 hours trying to build muscle is paying for the gym and the food and the supplements, but leaving a large fraction of the adaptation on the table because the hormonal milieu for recovery is disrupted.

Melatonin — the obvious one, but the numbers surprise

Melatonin is synthesized by the pineal gland in response to darkness, peaks around 2-4 a.m., and falls with morning light exposure. It is the clearest hormonal signal of the circadian pacemaker's phase and is what mediates entrainment to the light-dark cycle.

The numbers on how sensitive melatonin is to light are striking. Melatonin suppression curves in healthy adults show:

• 50% suppression at roughly 100 lux of white-spectrum light (equivalent to a well-lit kitchen at night)
• Substantial suppression at 50-80 lux (typical living room evening lighting)
• Near-complete suppression at 200+ lux (typical office or bright bathroom lighting)

Blue-enriched light (short wavelengths, 450-490 nm) is particularly potent for suppression because of the wavelength sensitivity of melanopsin in intrinsically photosensitive retinal ganglion cells. Phone and laptop screens at normal brightness produce enough retinal illuminance to suppress melatonin substantially, even at distances of 30-60 cm.

Age changes this. Melatonin production declines progressively from adolescence onward, with older adults producing 10-30% of the melatonin they did as young adults. This is one reason sleep architecture changes with age, and part of why melatonin supplementation may work better in older adults than in young ones.

Leptin / ghrelin — appetite dysregulation

Leptin, made by adipose tissue, signals fullness to the hypothalamus. Ghrelin, made primarily in the stomach, signals hunger. Sleep restriction consistently disrupts both in the direction of increased hunger.

Spiegel and colleagues showed that two nights of four-hour sleep in healthy young men reduced leptin by 18% and increased ghrelin by 28%, with corresponding increases in self-reported hunger and appetite for energy-dense, high-carbohydrate foods. The magnitude of the hormonal shift was comparable to the effect of prolonged fasting.

Crossing this with the insulin findings produces a consistent picture: short sleep increases hunger, drives food choices toward high-GI carbs, reduces insulin sensitivity, promotes fat storage, and does all of this before anyone has stepped on a scale. The "I can't lose weight even though I'm trying" experience in people sleeping poorly is real, and it is hormonally mediated, not willpower failure.

GLP-1 agonists, interestingly, partially overcome this hunger-driver effect because they act downstream of the leptin/ghrelin signaling. This may be part of why GLP-1s work in populations that had previously been unable to sustain weight loss through diet alone.

What breaks, and how fast

Acute restriction (1-3 nights)

After one night of 4-5 hour sleep in a previously-rested adult:

• Insulin sensitivity drops 20-30%
• Glucose tolerance impaired to roughly pre-diabetic range temporarily
• Cortisol elevated in the evening of the following day
• Ghrelin up, leptin down
• Subjective sleepiness, reduced reaction time, impaired working memory
• Testosterone is largely preserved on a single night (the system has some buffer)
• GH pulse is reduced but not abolished

This is a bad but reversible state. One good night of recovery sleep restores most of these parameters, though not all — sleep debt does not fully recover in one night.

Chronic (1-2 weeks)

After one to two weeks of 5-hour sleep:

• Total testosterone drops 10-15% in men (Leproult data)
• Cortisol elevation becomes sustained in the evening
• Insulin sensitivity continues to drop; tissue-specific adipose insulin resistance emerges
• Appetite regulation persistently disrupted
• Mood, motivation, and cognitive performance degraded with limited subjective insight (people on chronic partial sleep restriction substantially underrate their impairment)
• Immune function begins to show changes; antibody responses to vaccination are reduced in sleep-restricted subjects
• Training adaptations attenuated; athletes report performance declines

This is the state most chronically sleep-restricted adults are actually in. The "I've always only needed six hours" population is largely not that; they are adults with accumulated sleep debt that has become their baseline, not a genetic short-sleeper (who are extremely rare — less than 1-3% of the population).

Shift work and permanent misalignment

Long-term shift workers represent the most extreme population. Years of working at night and sleeping during the day, fighting circadian cues, produces:

• Persistent cortisol dysregulation
• Elevated cardiovascular risk (roughly 40% higher coronary artery disease incidence in long-term shift workers)
• Elevated metabolic syndrome prevalence
• Elevated cancer incidence (IARC Group 2A designation)
• Fertility impacts in both men and women
• Persistent mood and cognitive effects

The health cost of long-term shift work is large enough that it should be treated as a serious occupational exposure, but it rarely is in practice. If someone has been a shift worker for years and is experiencing multiple hormonal symptoms, the shift work is the most likely primary cause.

The levers that actually work

Light (morning, daytime, evening)

Light is the single most powerful non-pharmacologic lever for circadian phase and for sleep quality. Three windows matter:

Morning light (within 30-60 minutes of waking) entrains the circadian clock to the coming day. 10-30 minutes of outdoor light, even on an overcast day (>10,000 lux outdoors vs. 300-500 lux indoors), is far more potent than any indoor light source. Morning light advances the circadian phase, promotes earlier sleep onset that night, and is the most cost-effective sleep intervention available.

Daytime light maintains the circadian entrainment. Chronically working in a dim indoor environment (300-500 lux) without outside exposure is a form of mild circadian desynchronization. Workers with access to daytime light exposure sleep better and have stronger circadian amplitude.

Evening darkness allows melatonin to rise. The relevant window is roughly the last 2-3 hours before intended sleep. Dim ambient lighting (below 50 lux) and avoidance of bright blue-enriched screens prevent the melatonin-suppressing effects described earlier. This does not require elaborate interventions — dim the overhead lights, reduce screen brightness, use warm-spectrum light if possible, and the effect is substantial.

The "blue-blocking glasses" industry is mixed. Good-quality orange-tinted glasses that actually block short-wavelength light do reduce melatonin suppression, but the effect is smaller than just dimming ambient light and not using screens. Light-blocking apps on phones (Night Shift, f.lux) help marginally but do not fully eliminate the effect.

Temperature (core body, sleep onset curve)

Core body temperature has a strong diurnal pattern, and sleep onset is normally coincident with the beginning of the nighttime temperature decline. Cooling the body in the evening accelerates sleep onset; keeping the bedroom cool maintains deeper sleep.

Practical numbers:

• Bedroom temperature of 65-68°F (18-20°C) is associated with better sleep architecture than warmer temperatures
• A warm shower or bath 1-2 hours before bed accelerates sleep onset — counterintuitively, through post-bath peripheral vasodilation that accelerates core cooling
• Cooling mattress pads (with water or thermoelectric cooling) have reasonably good evidence for improving sleep, particularly in warm climates or for people with hot-flash issues

Conversely, a warm bedroom or heavy bedding is a common and overlooked cause of poor sleep quality.

Timing (consistency over duration)

Consistency of sleep timing matters almost as much as duration. Going to bed and waking at the same times across days produces better subjective and objective sleep quality than variable timing, even if total duration is the same.

Weekend catch-up sleep is worse than people think. Shifting sleep by 2-3 hours on weekends produces a mini-jet-lag effect that the body spends Monday and Tuesday recovering from. "Social jet lag" — the difference between your weekend and weekday midpoint of sleep — is an independent predictor of metabolic syndrome and cardiovascular risk in large cohorts.

The practical rule: pick a sleep schedule that is sustainable on weekdays and weekends, and hold it. An hour of weekend shift is tolerable; three hours is a recurring metabolic insult.

Caffeine half-life vs. your adenosine budget

Caffeine's half-life in most adults is 5-6 hours. This means that caffeine consumed at 2 p.m. still has half its concentration at 8 p.m. and a quarter at 2 a.m. People who report that "caffeine doesn't affect me" are often wrong about this in a specific way — they can fall asleep, but their sleep architecture is disrupted in ways they don't notice.

Caffeine works by blocking adenosine receptors. Adenosine accumulates throughout the waking day and is one of the main drivers of homeostatic sleep pressure. Blocking it in the afternoon effectively delays the accumulation of sleep pressure into the evening, which is why caffeine close to bedtime reduces both sleep onset latency improvement and sleep depth.

CYP1A2 genetic variants produce meaningful individual differences in caffeine metabolism rates. Slow metabolizers (about 50% of adults) clear caffeine at half the rate of fast metabolizers, and are more affected by late-afternoon consumption. Most people do not know their CYP1A2 status but can estimate it: if caffeine at 2 p.m. still feels like it's doing something at 6 p.m., you're likely a slow metabolizer.

Practical rule: last caffeine by early afternoon, earlier if you're trying to troubleshoot sleep problems. The "I need coffee at 4 p.m. to make it through the day" cycle is often self-perpetuating — the late coffee disrupts the sleep that would otherwise make the afternoon coffee unnecessary.

Alcohol and REM destruction

Alcohol is the most common and most underestimated sleep disruptor. Even moderate amounts (one or two drinks in the evening) produce:

• Faster sleep onset (yes, this is why people use it)
• Suppressed REM sleep for the first half of the night, followed by REM rebound in the second half
• Increased awakenings in the second half of the night
• Reduced N3 slow-wave sleep
• Suppressed GH pulse at sleep onset
• Reduced morning testosterone peak

The subjective experience of sleep-after-alcohol is often "I slept well" while the objective experience is fragmented sleep with much less of the important stages. People commonly use alcohol to address sleep problems it is actually causing or worsening.

For people with chronic sleep issues, a 2-4 week alcohol-free period is a reasonable diagnostic test. A substantial fraction of "insomnia" and "can't stay asleep" complaints resolve when evening alcohol is removed.

Sleep supplements that have data (and don't)

Melatonin — what dose actually works

Most melatonin supplementation in the US is dosed at 3-10 mg, which is supraphysiologic by roughly 100-fold. The actual physiological peak melatonin concentration in a healthy young adult is around 50-300 pg/mL. Supplemental melatonin at 3 mg produces peak concentrations of 10,000-50,000 pg/mL — far above any natural signal.

The dose that actually works for circadian phase shifting and sleep onset is 0.1-0.5 mg, taken 1-3 hours before desired sleep time. Higher doses produce larger peak concentrations that:

• Do not reliably improve sleep onset or quality
• Do produce next-morning grogginess and headaches
• May suppress endogenous melatonin production over chronic use

Low-dose melatonin (0.3 mg) has the best evidence for circadian phase-shifting effects in jet lag, shift work, and delayed sleep phase syndrome. It is generally not a good treatment for general insomnia, and its acute effects on sleep onset in adults with normal circadian function are modest.

In older adults, who have reduced endogenous melatonin, physiological doses (0.3-0.5 mg) may have more meaningful effects on sleep onset and quality than in younger adults.

Magnesium (which salt, when)

Magnesium has a reasonable evidence base for improving sleep quality, particularly in people who are magnesium-deficient (which, depending on diet, is a substantial fraction of adults).

The salt form matters. Magnesium oxide is poorly absorbed and mainly laxative. Magnesium glycinate, threonate, and malate are better absorbed and produce serum changes. Magnesium threonate has some evidence for CNS penetration and may have specific cognitive effects, though the evidence is still limited.

Doses of 200-400 mg elemental magnesium in the evening, from a well-absorbed salt, is a reasonable trial. Effect sizes are modest but real in deficient populations, and the safety profile is clean (diarrhea at high doses is the main limit).

Ashwagandha (short-term anxiolytic)

Ashwagandha (Withania somnifera) has multiple controlled trials showing anxiolytic effects and modest improvements in sleep quality in stressed adults. The effect is real but modest, and the evidence is for short-term use (8-12 weeks). Long-term use data are limited.

KSM-66 and Shoden are two well-studied standardized extracts; doses in trials were 300-600 mg/day. It is reasonable to try if stress is a significant component of sleep problems. It is not a substitute for addressing the underlying stress.

The bad: valerian, "sleep stacks," ZMA

Valerian has been widely studied and has not shown consistent benefits over placebo for sleep in meta-analyses. The studies that show effects tend to be small and have methodological issues. It is generally considered safe but probably ineffective.

"Sleep stacks" — GABA, L-theanine, passionflower, chamomile, various nootropic blends — have minimal good evidence for any of their specific claims. Some components (L-theanine) have mild anxiolytic effects; most of the rest is marketing built around plausible-sounding mechanisms with thin clinical evidence.

ZMA (zinc, magnesium, vitamin B6) is heavily marketed to athletes for sleep and testosterone support. The evidence for the testosterone claim is essentially zero in zinc-replete men. The sleep benefit, if any, is the magnesium component. The branded formulation has no advantage over just taking magnesium.

CBD has emerged as a popular sleep supplement. The evidence is mixed and confounded by poor product quality — many products contain less CBD than labeled. At adequate doses (100+ mg), some people report sleep benefits, though mechanisms are unclear and long-term safety in chronic daily use is not well-established.

The deeper argument

Sleep-fix-first for every other hormonal complaint

Take the following common complaints:

• Low testosterone
• Irregular menstrual cycles
• Weight gain despite diet
• Low energy / fatigue
• Poor mood / anxiety
• Difficulty concentrating
• Poor recovery from training
• Elevated fasting glucose
• High cortisol
• Low libido

Every one of these can be directly caused or substantially worsened by chronic sleep restriction. Many of them will partially or fully resolve with sleep correction alone. Almost none of them will fully resolve with any other intervention if the sleep is not also corrected.

This is not a rhetorical claim. It is an ordering claim about the intervention pyramid. Sleep is upstream of almost every other hormonal axis, and fixing downstream systems while sleep is broken is working against a persistent headwind. The "you need to fix your sleep first" advice is boring, unglamorous, and difficult to monetize, which is why the clinics selling TRT, GLP-1s, and supplements are not usually the ones spending the first appointment on sleep architecture.

Why "I function fine on 6 hours" is almost always false

The literature on subjective vs. objective impairment in chronic partial sleep restriction is striking. Van Dongen and colleagues ran 2-week sleep restriction protocols at 4, 6, and 8 hours per night. The 6-hour group showed progressive cognitive and physical impairments equivalent to 24-48 hours of total sleep deprivation by the end of the study, while subjectively reporting only mild sleepiness and believing they had adapted.

The people in the 6-hour group didn't feel as bad as they actually were because chronic sleep restriction degrades the metacognitive ability to assess one's own impairment. By the end of the study, objectively measured performance was substantially degraded while subjective reports were "fine."

The implication is uncomfortable: most adults who report "I only need 6 hours" have miscalibrated their self-assessment because they have been chronically sleep-restricted for so long that the degraded baseline feels normal. A two-week intervention extending sleep to 8 hours typically produces substantial improvements in mood, cognition, and performance in these individuals — revealing that "fine on 6 hours" was not accurate self-knowledge.

Genuine short sleepers — people with natural sleep needs in the 5-6 hour range — do exist but are rare (probably 1-3% of the population) and are identifiable by specific genetic variants (DEC2, ADRB1). They do not sleep short because of discipline; they sleep short because their biology does not produce sleep need beyond that amount. If you needed to train yourself to sleep 6 hours and are functioning through caffeine, you are almost certainly not in this population.

Honest take

Sources

• Leproult R., Van Cauter E., JAMA (2011) — Effect of 1 week of sleep restriction on testosterone levels in young healthy men. The definitive acute sleep-testosterone study.
• Van Dongen H.P.A. et al., Sleep (2003) — The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. The 4/6/8 hour chronic restriction study showing subjective vs. objective impairment divergence.
• Broussard J.L. et al., Annals of Internal Medicine (2012) — Impaired insulin signaling in human adipocytes after experimental sleep restriction. Tissue-specific insulin sensitivity after four nights of short sleep.
• Spiegel K., Tasali E., Penev P., Van Cauter E., Annals of Internal Medicine (2004) — Brief communication: Sleep curtailment in healthy young men is associated with decreased leptin levels, elevated ghrelin levels, and increased hunger and appetite.
• Van Cauter E. et al., JAMA (2000) — Age-related changes in slow wave sleep and REM sleep and relationship with growth hormone and cortisol levels in healthy men.
• Xie L. et al., Science (2013) — Sleep drives metabolite clearance from the adult brain. The glymphatic clearance discovery.
• Zhu J.L. et al., American Journal of Epidemiology (2011) — Shift work and subfecundity / metabolic consequences; representative of a substantial body of occupational epidemiology on shift work.
• Wehrens S.M.T. et al., Current Biology (2017) — Meal timing regulates the human circadian system. Shows the interaction between circadian alignment and metabolic hormones.
• Czeisler C.A. et al., Science (1999) — Stability, precision, and near-24-hour period of the human circadian pacemaker. Foundational on human circadian physiology.
• Walker M.P., Why We Sleep (2017) — Popular synthesis of sleep physiology literature; useful for orientation though some of the strongest claims exceed what the primary literature supports.
• Kripke D.F. et al., BMJ Open (2012) — Hypnotics' association with mortality or cancer: a matched cohort study. Relevant to evaluating pharmaceutical sleep aids.