Melatonin: Much More Than You Wanted To Know

[I am not a sleep spe­cial­ist. Please con­sult with one be­fore mak­ing any dras­tic changes or try­ing to treat any­thing se­ri­ous.]

Van Geijlswijk et al de­scribe sup­ple­men­tal mela­tonin as “a chrono­biotic drug with hyp­notic prop­er­ties”. Us­ing it as a pure hyp­notic – a sleep­ing pill – is like us­ing an AK-47 as a club to bash your en­e­mies’ heads in. It might work, but you’re failing to ap­pre­ci­ate the full power and sub­tlety available to you.

Me­la­tonin is a neu­ro­hor­mone pro­duced by the pineal gland. In a nor­mal cir­ca­dian cy­cle, it’s low­est (un­de­tectable, less than 1 pg/​ml of blood) around the time you wake up, and stays low through­out the day. Around fif­teen hours af­ter wak­ing, your mela­tonin sud­denly shoots up to 10 pg/​ml – a pro­cess called “dim light mela­tonin on­set”. For the next few hours, mela­tonin con­tinues to in­crease, maybe as high as 60 or 70 pg/​ml, mak­ing you sleepier and sleepier, and pre­sum­ably at some point you go to bed. Me­la­tonin peaks around 3 AM, then de­clines un­til it’s un­de­tectably low again around early morn­ing.

Is this what makes you sleepy? Yes and no. Sleep­iness is a com­bi­na­tion of the cir­ca­dian cy­cle and the so-called “Pro­cess S”. This is an un­nec­es­sar­ily sinister-sound­ing name for the fact that the longer you’ve been awake, the sleepier you’ll be. It seems to be partly reg­u­lated by a molecule called adeno­sine. While you’re awake, the body pro­duces adeno­sine, which makes you tired; as you sleep, the body clears adeno­sine away, mak­ing you feel well-rested again.

In healthy peo­ple these pro­cesses work to­gether. Cir­ca­dian rhythm tells you to feel sleepy at night and awake dur­ing the day. Pro­cess S tells you to feel awake when you’ve just risen from sleep (nat­u­rally the morn­ing), and tired when you haven’t slept in a long time (nat­u­rally the night). Both pro­cesses agree that you should feel awake dur­ing the day and tired at night, so you do.

When these pro­cesses dis­agree for some rea­son – night shifts, jet lag, drugs, ge­net­ics, play­ing Civ­i­liza­tion un­til 5 AM – the sys­tem fails. One pro­cess tells you to go to sleep, the other to wake up. You’re never quite awake enough to feel en­er­gized, or quite tired enough to get rest­ful sleep. You find your­self ly­ing in bed toss­ing and turn­ing, or wak­ing up while it’s still dark and not be­ing able to get back to sleep.

Me­la­tonin works on both sys­tems. It has a weak “hyp­notic” effect on Pro­cess S, mak­ing you im­me­di­ately sleepier when you take it. It also has a stronger “chrono­biotic” effect on the cir­ca­dian rhythm, shift­ing what time of day your body con­sid­ers sleep to be a good idea. Effec­tive use of mela­tonin comes from un­der­stand­ing both these effects and us­ing each where ap­pro­pri­ate.

1. Is mela­tonin an effec­tive hyp­notic?


That is, tak­ing mela­tonin just be­fore you want to get to sleep, does help you get to sleep. The ev­i­dence on this is pretty unan­i­mous. For pri­mary in­som­nia, two meta-analy­ses – one by Brzez­in­ski in 2005 and an­other by Fer­ra­cioli-Oda in 2013 – both find it safe and effec­tive. For jet lag, a meta-anal­y­sis by the usu­ally-skep­ti­cal Cochrane Col­lab­o­ra­tion pro­nounces mela­tonin “re­mark­ably effec­tive”. For a wide range of pri­mary and sec­ondary sleep di­s­or­ders, Buscemi et al say in their ab­stract that it doesn’t work, but a quick glance at the study shows it ab­solutely does and they are in­cor­rectly un­der-re­port­ing their own re­sults. The Psy­chi­a­tric Times agrees with me on this: “Re­sults from an­other study re­ported as nega­tive ac­tu­ally demon­strated a statis­ti­cally sig­nifi­cant pos­i­tive re­sult of a de­crease in sleep la­tency by an av­er­age of 7.2 min­utes for mela­tonin”.

Ex­pert con­sen­sus gen­er­ally fol­lows the meta-analy­ses: mela­tonin works. I find cau­tious en­dorse­ments by the Mayo Clinic and John Hop­kins less im­pres­sive than its less-than-com­pletely-nega­tive re­view on Science-Based Medicine, a blog I can usu­ally count on for a hit job on any dietary sup­ple­ment.

The con­sen­sus stresses that mela­tonin is a very weak hyp­notic. The Buscemi meta-anal­y­sis cites this as their rea­son for declar­ing nega­tive re­sults de­spite a statis­ti­cally sig­nifi­cant effect – the sup­ple­ment only made peo­ple get to sleep about ten min­utes faster. “Ten min­utes” sounds pretty pa­thetic, but we need to think of this in con­text. Even the strongest sleep med­i­ca­tions, like Am­bien, only show up in stud­ies as get­ting you to sleep ten or twenty min­utes faster; ef=“https://​ww<em>w.ny­​2007/​10/​23/​health/​23drug.html”>this New York Times ar­ti­cle says that “viewed as a group, [newer sleep­ing pills like Am­bien, Lunesta, and Sonata] re­duced the av­er­age time to go to sleep 12.8 min­utes com­pared with fake pills, and in­creased to­tal sleep time 11.4 min­utes.” I don’t know of any statis­ti­cally-prin­ci­pled com­par­i­son be­tween mela­tonin and Am­bien, but the differ­ence is hardly (pun not in­tended) day and night.

Rather than say “mela­tonin is crap”, I would ar­gue that all sleep­ing pills have mea­surable effects that vastly un­der­perform their sub­jec­tive effects. The linked ar­ti­cle spec­u­lates on one rea­son this might be: peo­ple have low aware­ness around the time they get to sleep, and a lot of peo­ple’s per­cep­tion of whether they’re in­som­niac or not is more anx­iety (or some­times liter­ally dream) than re­al­ity. This is pos­si­ble, but I also think of this in terms of an­tide­pres­sant stud­ies, which find similarly weak ob­jec­tive effects de­spite pa­tients (and doc­tors) who swear by them and say they changed their lives. If I had to guess, I would say that the stud­ies in­clude an awk­ward com­bi­na­tion of sick and less-sick peo­ple and con­fuse re­spon­ders and non-re­spon­ders. Maybe this is spe­cial plead­ing. I don’t know. But if you think any sleep­ing pill works well, mela­tonin doesn’t nec­es­sar­ily work much worse than that.

Sleep la­tency statis­tics are hard to com­pare to one an­other be­cause they’re so de­pen­dent on the study pop­u­la­tion. If your sub­jects take an hour to fall asleep, per­haps mela­tonin could shave off thirty-four min­utes. But if your sub­jects take twenty min­utes to fall asleep, then no sleep­ing pill will ever take off thirty-four min­utes, and even an amaz­ing sleep­ing pill might strug­gle to make fif­teen. I can­not di­rectly com­pare the peo­ple who say mela­tonin gives back ten min­utes to the peo­ple who say mela­tonin gives back thirty-four min­utes to the peo­ple who say Am­bien gives back twelve, but my to­tally un­prin­ci­pled guess is that mela­tonin is about a third as strong as Am­bien. It also has about a hun­dred times fewer side effects, so there’s definitely a place for it in sleep medicine.

2. What is the right dose of mela­tonin?

0.3 mg.

“But my lo­cal drug­store sells 10 mg pills! When I asked if they had any­thing lower, they looked through their stock­room and were even­tu­ally able to find 3 mg pills! And you’re say­ing the cor­rect dose is a third of a mil­li­gram?!”

Yes. Most ex­ist­ing mela­tonin tablets are around ten to thirty times the cor­rect dose.

Many early stud­ies were done on el­derly peo­ple, who pro­duce less en­doge­nous mela­tonin than young peo­ple and so are con­sid­ered es­pe­cially re­spon­sive to the drug. Sev­eral lines of ev­i­dence de­ter­mined that 0.3 mg was the best dose for this pop­u­la­tion. Elderly peo­ple given doses around 0.3 mg slept bet­ter than those given 3 mg or more and had fewer side effects (Zh­danova et al 2001). A meta-anal­y­sis of dose-re­sponse re­la­tion­ships con­curred, find­ing a plateau effect around 0.3 mg, with doses af­ter that hav­ing no more effi­cacy, but worse side effects (Brzez­in­ski et al, 2005). And doses around 0.3 mg cause blood mela­tonin spikes most similar in mag­ni­tude and du­ra­tion to the spikes seen in healthy young peo­ple with nor­mal sleep (Vu­ral et al, 2014).

Other stud­ies were done on blind peo­ple, who are es­pe­cially sen­si­tive to mela­tonin since they lack light cues to en­train their cir­ca­dian rhythms. This is a lit­tle bit of a differ­ent in­di­ca­tion, since it’s be­ing used more as a chrono­biotic than a sleep­ing pill, but the re­sults were very similar: lower doses worked bet­ter than higher doses. For ex­am­ple, in Lewy et al 2002, nightly doses of 0.5 mg worked to get a blind sub­ject sleep­ing nor­mally at night; doses of 20 mg didn’t. They rea­son­ably con­clude that the 20 mg is such a high dose that it stays in their body all day, defeat­ing the point of a hor­mone whose job is to sig­nal night­time. Other stud­ies on the blind have gen­er­ally con­firmed that doses of around 0.3 to 0.5 mg are op­ti­mal.

There have been dis­ap­point­ingly few stud­ies on sighted young peo­ple. One such, At­ten­bur­row et al 1996 finds that 1 mg works but 0.3 mg doesn’t, sug­gest­ing these peo­ple may need slightly higher doses, but this study is a bit of an out­lier. Another Zh­danova study on 25 year olds found both to work equally. And Pires et al study­ing 22-24 year olds found that 0.3 mg worked bet­ter than 1.0. I am less in­ter­ested in judg­ing the 0.3 mg vs. 1.0 mg de­bate than in point­ing out that both num­bers are much lower than the 3 – 10 mg doses found in the mela­tonin tablets sold in drug­stores.

UpToDate, the gold stan­dard re­search database used by doc­tors, agrees with these low doses. “We sug­gest the use of low, phys­iologic doses (0.1 to 0.5 mg) for in­som­nia or jet lag (Grade 2B). High-dose prepa­ra­tions raise plasma mela­tonin con­cen­tra­tions to a sup­ra­phys­iologic level and al­ter nor­mal day/​night mela­tonin rhythms.” Mayo Clinic makes a similar recom­men­da­tion: they recom­mend 0.5 mg. John Hop­kins’ ex­perts al­most agree: they say “less is more” but end up chick­en­ing out and recom­mend­ing 1 to 3 mg, which is well above what the stud­ies would sug­gest.

Based on a bunch of stud­ies that ei­ther fa­vor the lower dose or show no differ­ence be­tween doses, plus clear ev­i­dence that 0.3 mg pro­duces an effect clos­est to nat­u­ral mela­tonin spikes in healthy peo­ple, plus UpToDate usu­ally hav­ing the best recom­men­da­tions, I’m in fa­vor of the 0.3 mg num­ber. I think you could make an ar­gu­ment for any­thing up to 1 mg. Any­thing be­yond that and you’re definitely too high. Ex­cess mela­tonin isn’t grossly dan­ger­ous, but tends to pro­duce tol­er­ance and might mess up your chrono­biol­ogy in other ways. Based on anec­do­tal re­ports and the im­plau­si­bil­ity of be­com­ing tol­er­ant to a nat­u­ral hor­mone at the dose you nat­u­rally have it, I would guess suffi­ciently low doses are safe and effec­tive long term, but this is just a guess, and most guidelines are cau­tious in say­ing any­thing af­ter three months or so.

3. What are cir­ca­dian rhythm di­s­or­ders? How do I use mela­tonin for them?

Cir­ca­dian rhythm di­s­or­ders are when your cir­ca­dian rhythm doesn’t match the nor­mal cy­cle where you want to sleep at night and wake up in the morn­ing.

The most pop­u­lar cir­ca­dian rhythm di­s­or­der is “be­ing a teenager”. Teenagers’ mela­tonin cy­cle is nat­u­rally shifted later, so that they don’t want to go to bed un­til mid­night or later, and don’t want to wake up un­til eight or later. This is an ob­vi­ous mis­match with school start­ing times, lead­ing to teenagers ei­ther not get­ting enough sleep, or get­ting their sleep at times their body doesn’t want to be asleep and isn’t able to use it prop­erly. This is why ev­ery rep­utable sleep sci­en­tist and rele­vant sci­en­tific body keeps tel­ling the pub­lic school sys­tem to start later.

When a this kind of late sleep sched­ule per­sists into adult­hood or be­comes too dis­tress­ing, we call it De­layed Sleep Phase Di­sor­der. Peo­ple with DSPD don’t get tired un­til very late, and will nat­u­rally sleep late if given the chance. The weak ver­sion of this is “be­ing a night owl” or “not be­ing a morn­ing per­son”. The strong ver­sion just looks like in­som­nia: you go to bed at 11 PM, toss and turn un­til 2 AM, wake up when your alarm goes off at 7, and com­plain you “can’t sleep”. But if you can sleep at 2 AM, con­sis­tently, re­gard­less of when you wake up, and you would fall asleep as soon as your head hit the pillow if you first got into bed at 2, then this isn’t in­som­nia – it’s DSPD.

The op­po­site of this pat­tern is Ad­vanced Sleep Phase Di­sor­der. This is most com­mon in the el­derly, and I re­mem­ber my grand­father hav­ing this. He would get tired around 6 PM, go to bed by 7, wake around 1 or 2 AM, and start his day feel­ing fresh and alert. But the weak ver­sion of this is the per­son who wakes up at 5 each morn­ing even though their alarm doesn’t go off un­til 8 and they could re­ally use the ex­tra two hours’ sleep. Th­ese peo­ple would prob­a­bly do fine if they just went to bed at 8 or 9, but the de­mands of work and a so­cial life make them feel like they “ought” to stay up as late as ev­ery­one else. So they go to bed at 11, wake up at 5, and com­plain of “ter­mi­nal in­som­nia”.

Fi­nally, there’s Non-24-Hour-Sleep Di­sor­der, where some­how your biolog­i­cal clock ended up deeply and un­shake­ably con­vinced that days on Earth are twenty-five (or what­ever) hours long, and de­cides this is the hill it wants to die on. So if you nat­u­rally sleep 11 – 7 one night, you’ll nat­u­rally sleep 12 – 8 the next night, 1 to 9 the night af­ter that, and so on un­til ei­ther you make a com­plete 24-hour cy­cle or (more likely) you get so tired and con­fused that you stay up 24+ hours and break the cy­cle. This is most com­mon in blind peo­ple, who don’t have the vi­sual cues they need to re­mind them­selves of the 24 hour day, but it hap­pens in a few sighted peo­ple also; Eliezer Yud­kowsky has writ­ten about his strug­gles with this con­di­tion.

Me­la­tonin effec­tively treats these con­di­tions, but you’ve got to use it right.

The gen­eral heuris­tic is that mela­tonin drags your sleep time to­wards the di­rec­tion of when you take the mela­tonin.

So if you want to go to sleep (and wake up) ear­lier, you want to take mela­tonin early in the day. How early? Van Geijlswijk et al sums up the re­search as say­ing it is most effec­tive “5 hours prior to both the tra­di­tion­ally de­ter­mined [dim light mela­tonin on­set] (cir­ca­dian time 9)”. If you don’t know your own mela­tonin cy­cle, your best bet is to take it 9 hours af­ter you wake up (which is pre­sum­ably about seven hours be­fore you go to sleep).

What if you want to go to sleep (and wake up) later? Our un­der­stand­ing of the mela­tonin cy­cle strongly sug­gests mela­tonin taken first thing upon wak­ing up would work for this, but as far as I know this has never been for­mally in­ves­ti­gated. The best I can find is re­searchers say­ing that they think it would hap­pen and be­ing con­fused why no other re­searcher has in­ves­ti­gated this.

And what about non-24-hour sleep di­s­or­ders? I think the goal in treat­ment here is to ad­vance your phase each day by tak­ing mela­tonin at the same time, so that your sleep sched­ule is more de­pen­dent on your own sup­ple­men­tal mela­tonin than your (screwed up) nat­u­ral mela­tonin. I see con­flict­ing ad­vice about how to do this, with some peo­ple say­ing to use mela­tonin as a hyp­notic (ie just be­fore you go to bed) and oth­ers say­ing to use it on a typ­i­cal phase ad­vance sched­ule (ie nine hours af­ter wak­ing and seven be­fore sleep­ing, plau­si­bly about 5 PM). I think this one might be com­pli­cated, and a qual­ified sleep doc­tor who un­der­stands your per­sonal rhythm might be able to tell you which sched­ule is best for you. Eliezer says the lat­ter reg­i­men had very im­pres­sive effects for him (search “Last but not least” here). I’m in­ter­ested in hear­ing from the Me­taMed re­searcher who gave him that recom­men­da­tion on how they knew he needed a phase ad­vance sched­ule.

Does mela­tonin used this way cause drowsi­ness (eg at 5 PM)? I think it might, but prob­a­bly such a min­i­mal amount com­pared to the non-sleep-con­ducive­ness of the hour that it doesn’t reg­ister.

Me­la­tonin isn’t the only way to ad­vance or de­lay sleep phase. Here is a handy cheat sheet of re­search find­ings and the­o­ret­i­cal pre­dic­tions:

TO TREAT DELAYED PHASE SLEEP DISORDER (ie you go to bed too late and wake up too late, and you want it to be ear­lier)

– Take mela­tonin 9 hours af­ter wake and 7 be­fore sleep, eg 5 PM

– Block blue light (eg with blue-blocker sun­glasses or f.lux) af­ter sunset

– Ex­pose your­self to bright blue light (sun­light if pos­si­ble, dawn simu­la­tor or light boxes if not) early in the morning

– Get early morn­ing exercise

– Beta-block­ers early in the morn­ing (not gen­er­ally recom­mended, but if you’re tak­ing beta-block­ers, take them in the morn­ing)

TO TREAT ADVANCED PHASE SLEEP DISORDER (ie you go to bed too early and wake up too early, and you want it to be later)

– Take mela­tonin im­me­di­ately af­ter waking

– Block blue light (eg with blue-blocker sun­glasses or f.lux) early in the morning

– Ex­pose your­self to bright blue light (sun­light if pos­si­ble, light boxes if not) in the evening.

– Get late evening exercise

– Beta-block­ers in the evening (not gen­er­ally recom­mended, but if you’re tak­ing beta-block­ers, take them in the evening)

Th­ese don’t “cure” the con­di­tion per­ma­nently; you have to keep do­ing them ev­ery day, or your cir­ca­dian rhythm will snap back to its nat­u­ral pat­tern.

What is the cor­rect dose for these in­di­ca­tions? Here there is a lot more con­tro­versy than the hyp­notic dose. Of the nine stud­ies van Geijlswijk de­scribes, seven have doses of 5 mg, which sug­gests this is some­thing of a stan­dard for this pur­pose. But the only study to com­pare differ­ent doses di­rectly (Mundey et al 2005) found no differ­ence be­tween a 0.3 and 3.0 mg dose. The Cochrane Re­view on jet lag, which we’ll see is the same pro­cess, similarly finds no differ­ence be­tween 0.5 and 5.0.

Van Geijlswijk makes the im­por­tant point that if you take 0.3 mg seven hours be­fore bed­time, none of it is go­ing to be re­main­ing in your sys­tem at bed­time, so it’s un­clear how this even works. But – well, it is pretty un­clear how this works. In par­tic­u­lar, I don’t think there’s a great well-un­der­stood phys­iolog­i­cal ex­pla­na­tion for how tak­ing mela­tonin early in the day shifts your cir­ca­dian rhythm seven hours later.

So I think the ev­i­dence points to 0.3 mg be­ing a pretty good dose here too, but I wouldn’t blame you if you wanted to try tak­ing more.

4. How do I use mela­tonin for jet lag?

Most stud­ies say to take a dose of 0.3 mg just be­fore (your new time zone’s) bed­time.

This doesn’t make a lot of sense to me. It seems like you should be able to model jet lag as a cir­ca­dian rhythm di­s­or­der. That is, if you move to a time zone that’s five hours ear­lier, you’re in the ex­act same po­si­tion as a teenager whose cir­ca­dian rhythm is set five hours later than the rest of the world’s. This sug­gests you should use DSPD pro­to­col of tak­ing mela­tonin nine hours af­ter wak­ing /​ five hours be­fore DLMO /​ seven hours be­fore sleep.

My guess is for most peo­ple, their new time zone bed­time is a cou­ple of hours be­fore their old bed­time, so you’re get­ting most of the effect, plus the hyp­notic effect. But I’m not sure. Maybe tak­ing it ear­lier would work bet­ter. But given that the new light sched­ule is already work­ing in your fa­vor, I think most peo­ple find that tak­ing it at bed­time is more than good enough for them.

5. I try to use mela­tonin for sleep, but it just gives me weird dreams and makes me wake up very early

This is my ex­pe­rience too. When I use mela­tonin, I find I wake the next morn­ing with a jolt of en­ergy. Although I usu­ally have to grudg­ingly pull my­self out of bed, mela­tonin makes me wake up bright-eyed, smil­ing, and ready to face the day ahead of me…

…at 4 AM, in­vari­ably. This is why de­spite my in­ter­est in this sub­stance I never take mela­tonin my­self any­more.

There are many peo­ple like me. What’s go­ing on with us, and can we find a way to make mela­tonin work for us?

This bro-sci­ence site has an uncited the­ory. Me­la­tonin is known to sup­press cor­ti­sol pro­duc­tion. And cor­ti­sol is in­versely cor­re­lated with adrenal­ine. So if you’re nat­u­rally very low cor­ti­sol, mela­tonin spikes your adrenal­ine too high, pro­duc­ing the “wake with a jolt” phe­nomenon that I and some other peo­ple ex­pe­rience. I like the way these peo­ple think. They un­der­stand in­di­vi­d­ual vari­abil­ity, their model is biolog­i­cally plau­si­ble, and it makes sense. It’s also prob­a­bly wrong; it has too many steps, and noth­ing in biol­ogy is ever this el­e­gant or sen­si­ble.

I think a more par­si­mo­nious the­ory would have to in­volve cir­ca­dian rhythm in some way. Even an 0.3 mg dose of mela­tonin gives your body the ab­solute max­i­mum amount of mela­tonin it would ever have dur­ing a nat­u­ral cir­ca­dian cy­cle. So sup­pose I want to go to bed at 11, and take 0.3 mg mela­tonin. Now my body has a mela­tonin peak (usu­ally as­so­ci­ated with the very mid­dle of the night, like 3 AM) at 11. If it as­sumes that means it’s re­ally 3 AM, then it might de­cide to wake up 5 hours later, at what it thinks is 8 AM, but which is ac­tu­ally 4.

I think I have a much weaker cir­ca­dian rhythm than most peo­ple – at least, I take a lot of naps dur­ing the day, and fall asleep about equally well when­ever. If that’s true, maybe mela­tonin acts as a su­per­stim­u­lus for me. The nor­mal ten­dency to wake up feel­ing re­freshed and alert gets ex­ag­ger­ated into a sud­den ir­re­sistable jolt of awak­e­ness.

I don’t know if this is any closer to the truth than the adrenal­ine the­ory, but it at least fits what we know about cir­ca­dian rhythms. I’m go­ing to try to put some ques­tions about mela­tonin re­sponse on the SSC sur­vey this year, so start try­ing mela­tonin now so you can provide use­ful data.

What about the weird dreams?

From a HuffPo ar­ti­cle:

Dr. Ra­fael Pe­layo, a Stan­ford Univer­sity pro­fes­sor of sleep medicine, said he doesn’t think mela­tonin causes vivid dreams on its own. “Who takes mela­tonin? Some­one who’s hav­ing trou­ble sleep­ing. And once you take any­thing for your sleep, once you start sleep­ing more or bet­ter, you have what’s called ‘REM re­bound,’” he said.

This means your body “catches up” on the sleep phase known as rapid eye move­ment, which is char­ac­ter­ized by high lev­els of brain-wave ac­tivity.

Nor­mal sub­jects who take mela­tonin sup­ple­ments in the con­trol­led set­ting of a sleep lab do not spend more time dream­ing or in REM sleep, Pe­layo added. This sug­gests that there is no in­her­ent prop­erty of mela­tonin that leads to more or weirder dreams.

Okay, but I usu­ally have nor­mal sleep. I take mela­tonin some­times be­cause I like ex­per­i­ment­ing with psy­chotropic sub­stances. And I still get some re­ally weird dreams. A Slate jour­nal­ist says he’s been tak­ing mela­tonin for nine years and still gets crazy dreams.

We know that REM sleep is most com­mon to­wards the end of sleep in the early morn­ing. And we know that some parts of sleep struc­ture are re­spon­sive to mela­tonin di­rectly. There’s a lot of de­bate over ex­actly what mela­tonin does to REM sleep, but given all the re­ports of al­tered dream­ing, I think you could pull to­gether a case that it has some role in sleep ar­chi­tec­ture that pro­motes or in­ten­sifies REM.

6. Does this re­late to any other psy­chi­a­tric con­di­tions?

Prob­a­bly, but this is all still spec­u­la­tive.

Sea­sonal af­fec­tive di­s­or­der is the clear­est sus­pect. We know that the sea­sonal mood changes don’t have any­thing to do with tem­per­a­ture; they seem to be based en­tirely on win­ter hav­ing shorter (vs. sum­mer hav­ing longer) days.

There’s some ev­i­dence that there are two sep­a­rate kinds of win­ter de­pres­sion. In one, the late sun­rises train peo­ple to a late cir­ca­dian rhythm and they end up phase-de­layed. In the other, the early sun­sets train peo­ple to an early cir­ca­dian rhythm and they end up phase-ad­vanced. Plau­si­bly SAD also in­volves some com­bi­na­tion of the two where the cir­ca­dian rhythm doesn’t know what it’s do­ing. In ei­ther case, this can make sleep non-cir­ca­dian-rhythm-con­gru­ent and so less effec­tive at do­ing what­ever it is sleep does, which causes mood prob­lems.

How does sun­rise time af­fect the av­er­age per­son, who is rarely awake for the sun­rise any­way and usu­ally sleeps in a dark room? I think your brain sub­con­sciously “no­tices” the time of the dawn even if you are asleep. There are some weird path­ways lead­ing from the eyes to the nu­cleus gov­ern­ing cir­ca­dian rhythm that seem in­de­pen­dent of any other kind of vi­sion; these might be keep­ing tabs on the sun­rise if even a lit­tle out­side light is able to leak into your room. I’m bas­ing this also on the claim that dawn simu­la­tors work even if you sleep through them. I don’t know if peo­ple get sea­sonal af­fec­tive di­s­or­der if they sleep in a com­pletely en­closed spot (eg un­der­ground) where there’s no con­ceiv­able way for them to mon­i­tor sun­rise times.

Bright light is the stan­dard treat­ment for SAD for the same rea­son it’s the stan­dard treat­ment for any other cir­ca­dian phase de­lay, but shouldn’t mela­tonin work also? Yes, and there are some pre­limi­nary stud­ies (pa­per, ar­ti­cle) show­ing it does. You have to be a bit care­ful, be­cause some peo­ple are phase-de­layed and oth­ers phase-ad­vanced, and if you use mela­tonin the wrong way it will make things worse. But for the stan­dard phase-de­lay type of SAD, nor­mal phase ad­vanc­ing mela­tonin pro­to­col seems to go well with bright light as an ad­di­tional treat­ment.

This model also ex­plains the oth­er­wise con­fus­ing ten­dency of some SAD suffer­ers to get de­pressed in the sum­mer. The prob­lem isn’t amount of light, it’s cir­ca­dian rhythm dis­rup­tion – which sum­mer can do just as well as win­ter can.

I’m also very sus­pi­cious there’s a strong cir­ca­dian com­po­nent to de­pres­sion, based on a few lines of ev­i­dence.

First, one of the most clas­sic symp­toms of de­pres­sion is awak­en­ing in the very early morn­ing and not be­ing able to get back to sleep. This is con­fus­ing for de­pressed peo­ple, who usu­ally think of them­selves as very tired and need­ing to sleep more, but it definitely hap­pens. This fits the pro­file for a cir­ca­dian rhythm is­sue.

Se­cond, agome­la­tine, a mela­tonin analogue, is an effec­tive (ish) an­tide­pres­sant.

Third, for some rea­son stay­ing awake for 24+ hours is a very effec­tive de­pres­sion treat­ment (albeit tem­po­rary; you’ll go back to nor­mal af­ter sleep­ing). This seems to sort of be a way of tel­ling your cir­ca­dian rhythm “You can’t fire me, I quit”, and there are some com­pli­cated sleep de­pri­va­tion /​ cir­ca­dian shift pro­to­cols that try to lev­er­age it into a longer-last­ing cure. I don’t know any­thing about this, but it seems pretty in­ter­est­ing.

Fourth, we checked and de­pressed peo­ple definitely have weird cir­ca­dian rhythms.

Last of all, bipo­lar has a very strong cir­ca­dian com­po­nent. There aren’t a whole lot of lifestyle changes that re­ally work for pre­vent­ing bipo­lar mood epi­sodes, but one of the big ones is keep­ing a steady bed and wake time. So­cial rhythms ther­apy, a rare effec­tive psy­chother­apy for bipo­lar di­s­or­der, re­volves around train­ing bipo­lar peo­ple to con­trol their cir­ca­dian rhythms.

The­o­ries of why cir­ca­dian rhythms mat­ter so much re­volve ei­ther around the idea of pro-cir­ca­dian sleep – that sleep is more restora­tive and effec­tive when it matches the cir­ca­dian cy­cle – or the idea of mul­ti­ple cir­ca­dian rhythms, with the body func­tion­ing bet­ter when all of them are in sync.

7. How can I know what the best mela­tonin sup­ple­ment is?

Lab­door has done pu­rity tests on var­i­ous brands and has ranked them for you. All the ones they high­light are still ten to thirty times the ap­pro­pri­ate dose (also, stop call­ing them things like “Triple Strength!” You don’t want your med­i­ca­tions to be too strong!). As usual, I trust Nootrop­ic­sDe­pot for things like this – and sure enough their mela­tonin (available on Ama­zon) is ex­actly 0.3 mg. God bless them.