Seth’s blog

Forty or fifty years ago, psychologists and other scientists talked about “genes” determining this or that. (James Watson still talks this way.) A certain percentage of the variation of this or that (e.g., intelligence) was attributed to “genes”. Hardly anyone outside genetics or behavior genetics knew what this meant, but many people thought they did. In reaction to the huge misunderstanding (e.g., those who said intelligence was “80% genetic” but did not know what this meant), psychologists began to talk about gene-environment interaction. “Is the area of a rectangle determined by its height or its width?” they like to say.

But notice how fact-free this view is. A tiny number of studies have observed gene-environment interactions but they are very difficult. I think this has made it hard to realize something basic and important. Years ago, I heard a talk about squirrel circadian rhythms by Patricia DeCoursey, the scientist who introduced the concept of phase-response curves. At her talk, she showed results from about 15 squirrels. She tested each one — with an emphasis on individual results that resembles self-experimentation — to determine how much light it needed to become entrained to a 24-hour light/dark cycle. One squirrel needed much stronger light than the others.

Here was an interesting finding that another scientist might have missed. What did it mean? Because the squirrels lived under very similar conditions (e.g., identical diets), it was almost surely a genetic difference. Let’s assume it was. In nature, sunlight is plenty strong. The lab light was weaker. In nature, the genetic difference wouldn’t make an observable difference. Only under artificial conditions did it become visible. It only became visible when the artificial conditions didn’t supply enough of something important (sunlight). In other words, the newly-visible genetic difference implied there was something lacking in the artificial conditions. The genetic difference implied the environment mattered. The opposite of the usual interpretation.

I don’t know any reason to think this is an unusual case. Aaron Blaisdell told me a story that shows its relevance to human health. Aaron is unusually sensitive to sunlight. Until recently, he could only spend 5 or 10 minutes in the sun before it became unpleasant. The condition is genetic. His mother has it; her father had it. It’s called Erythropoietic Protoporphyria. It is autosomal-dominant. Scientists even know where the gene is. That’s where the understanding of most scientists stops. A genetic condition. Recently, however, Aaron drastically changed his diet with great results, as noted earlier. At the same time as the dietary changes, his sun sensitivity got much better. He can now stay in the sun for an hour or more without discomfort. This is a gene-environment interaction, of course, but of a particular sort: The genetic effect showed there was something wrong with the environment, just as it did in DeCoursey’s experiment.

Sure, there’s always genetic variation — it’s just usually hard to see. The wrong environment makes it much easier to see. It reveals a range of genotypes, all of which would be harmless in the right environment. So when you come across a “genetic disorder” such as Erythropoetic Protoporphyria, it is likely to imply an environmental problem. No one had ever told Aaron or his mother or her father that their condition suggested that environmental changes would help them.

Song Cato, a friend of mine in Taiwan, writes:

I was very surprised that the quality of my sleep greatly improved after I switched to waking up at 5:30 am and walking in the park soon after that. I started it about a month ago. The park is packed with people doing everything from tai chi to ballroom dancing. I used to go to bed at 1 or 2 am. and wake up between 7 and 8:30 am with a foggy head. Now sometimes I feel tired and go to sleep at 10 pm which has never happened in my life since I went to middle school.

She got the idea from me. I go outside around 7 am every morning and fall asleep between 11 pm and midnight.

More. She gets up at about 5:15 am and gets outside about 5:30. She stays outside for at least 2.5 hours, mostly in the park, where she walks, talks to vendors, shops a little, and does simple stretching exercises. Talking to vendors = very good!

Surely we need sunlight to sleep properly. But how much? Rats can be synchronized to a 24-hour activity rhythm with a relatively small amount of light (such as one hour) every 24 hours. This is one reason for the emphasis on morning light by sleep doctors mentioned in a previous post.

I have agreed with them. For the last 10 years I have gotten one hour of sunlight-like light every morning from a bank of fluorescent lights on the handles of my treadmill. The lights shined up at me while I exercised and watched TV. This, I thought, allowed me to get a good dose of light with low variance in when and how much and to combine light-getting with exercise. I never questioned this routine.

Then came the event that led to this Sunlight and Sleep series: In the airport during a trip to New Orleans, a student told me when she sunbathes, she sleeps better. When I got home from my trip I tested her idea. Me, too: When I was outdoors a lot (in the shade), I slept better.

I took another trip (to Los Angeles). When I got back from that trip, I decided that I would adjust the timing of the treadmill light so that it interfered less with my day. I shifted it from 9:00 am to 10:00 am (original timing) to 8:00 am to 9:00 am (new timing).

To my surprise I started waking up too early, so often it could not be a coincidence. The only change I had made was timing of the light. So the treadmill light was making things worse! I stopped it entirely. My sleep improved — no more early awakening. Huh.

Here are details:

Period 1 (treadmill light 9-10 am, little sunlight): woke up early 29 days out of 99 (29%)

Period 2 (treadmill light 9-10 am, lots of sunlight): woke up early 1 day out of 25 (4%)

Period 3 (treadmill light 8-9 am, lots of sunlight): woke up early 4 days out 8 (50%)

Period 4 (no treadmill light, lots of sunlight): woke up early 0 days out of 8 (0%).

Lots of sunlight means 6-8 hours exposure to light of roughly 1000-2000 lux. Sitting in the shade or inside next to a big window is always enough. At the low end (1000 lux) my laptop screen is easy to read; at the high end (2000 lux), which I try to avoid, it becomes slightly hard to read.

An early section of Wide Awake (2006), Alan Berliner’s documentary about his life-long insomnia — he can’t fall asleep until 3 or 4 am — lists common folk remedies:

BERLINER Over the years I’ve tried to cry myself to sleep, to drink myself to sleep, aroma therapy, changing mattresses, changing pillows, lavender beads, massage therapy, white noise, meditation, counting sheep, melatonin, Valerian root, acupuncture, acupressure, chamomile tea, warm milk, hypnosis even, yoga, homeopathic medicines, marijuana, lots of sex, hot baths, herbal teas, biofeedback.
SISTER Okay, nothing worked.

Conspicuously absent: sunlight. At the end of the movie, however:

DOCTOR We have to reset your [internal] clock. Since you’re such a night owl, I’d like to move your sleep cycle earlier by having you get light exposure in the morning. When you wake up, throw on some clothes and go outdoors for an hour. I really want light to get into your eyes ’cause that’s what going to move your rhythm so you can fall asleep earlier.
ANOTHER DOCTOR Light is one of the most powerful cues for your internal clock to know what time it is. You see light and it tells you: be active during the day, sleep at night.

But the treatment they settle on is sleep deprivation: “I’d like you to spend just 6.5 hours in bed,” says a doctor. “Give you less time in bed than you want. . . . 2:30 to 9:00 am would be a reasonable way to go.” “You are going to be dysfunctional,” Berliner is warned. The film ends: “Now that I know what I have to do, the question is: Can I do it?”

This is a good summary of what people believe about how to cure insomnia. Sunshine is absent from the folk remedies you are likely hear. When doctors mention it, they emphasize early-morning sunlight.

Until recently, I too thought that sunlight exposure was important in the morning, but not during the rest of the day. Every morning I exercised on a treadmill with sunlight-spectrum light shining on me for an hour; I thought that was enough. Now I am adding to that sunlight later in the day — in the afternoon, for instance — and finding that it helps.

I’ve collected even more observations supporting the idea that outdoor light improves my sleep, as discussed earlier. Now I’d like to get some idea of the dose-response function. To sleep really well do I need two hours of outside light? Four hours? Eight hours?

I’ve started to rate my sleep on a scale where 50 = average sleep (average for the months before I started spending more time outside) and 100 = best sleep imaginable (which I got after standing about 10 hours). And I’ve started to use a stopwatch to measure how long I spend outdoors. I’ve also been using a light meter to measure the strength of light in various places. When I’m outdoors it’s almost always in the shade. Today I discovered that sitting indoors next to a cafe window the incident light was just as bright as when I sit outside. Great to know because indoors I can plug in my laptop.

A 1994 book chapter from Daniel Kripke’s lab reported a correlation (0.24) between low light exposure and “abnormal sleep.” So the connection I am now studying has been plausible for many years. The measurements I am now making are easy, but no one made them. Perhaps too many people believe that anything other than a double-blind trial with control and experimental groups is, as Peter Norvig, Google’s Director of Research, believes, a “mistake.”