CaffeineRevisiting a long-held position against.


After around fifteen years of avoiding caffeine, I am experimenting with maintenance intake. The galvanizing moment was when I lay awake in bed for hours after eating chili which contained less than a teaspoon of cacao powder (on the order of 5 mg of caffeine). And not for the first time. I thought, “Cacao has polyphenols and is tasty and is good for you, so it is stupid to be this sensitive to caffeine.” I also rationalized that habituation might enable me to handle some social consumption, if and when we ever go back to hanging out in coffee shops post-COVID.

Maintenance vs. abstinence

Many years ago I settled on the view that trying to use caffeine as a stimulant on a daily basis is self-defeating. Caffeine works by occupying the brain’s receptors for the neurotransmitter adenosine. Since adenosine generates a feeling of tiredness when docking at its receptor, caffeine blocks this and creates alertness — initially. But within a few days, the body builds tolerance by adding more adenosine receptors. Increased adenosine receptors means that even though the body may continue to produce the same amount of adenosine, the effect of any adenosine in the system is multiplied. So after developing dependence on caffeine, the baseline state of alertness is lowered, and caffeine intake becomes required just to restore normal wakefulness again. In order to feel a buzz, you would need to consume more caffeine than it takes to restore normal alertness, and if you do this for more than a few days, the body adapts again by growing more adenosine receptors. So you can’t continue this arms race forever, you have to level off somewhere. And once you have stabilized, if your intake diverges from your tolerance level too much it can cause either distracting jitters or withdrawal headaches. So it sure seems like the best bet is to keep intake as stable as possible, and only make changes by gradual ramping. So, I concluded that the only reasonable choice is between a consistent maintenance routine and abstinence.

In the past I presumed that of those two choices, the net effect would be roughly the same but the maintenance routine would be more trouble and more prone to error, therefore abstinence must be the better option. For example, I have had experiences where traveling for a few days disrupted my sense of routine and caused me to neglect my usual habits, which could lead to withdrawal. However, the net effect of the two options may not be the same. With caffeine intake, sensitivity to deviations from the norm is likely to be inversely proportional to one’s degree of tolerance. So, if your maintenance dose is zero (total abstinence — no tolerance), you will be maximally sensitive to any extra intake. But the higher the amount you are habituated to, the less that a given amount of variance will probably affect you. For instance, if you’re habituated to 100 mg per day, then an extra 5 mg on one occasion is probably not going to give you insomnia (stupid chili).

The major disadvantage I can think of for the maintenance approach is in case I do something like a multi-day meditation retreat or a medical procedure that requires a complete fast. In that case, I can hope to have several weeks to prepare in advance by tapering off.

Net benefit or harm from consistent daily caffeine use

There are numerous studies showing benefits of caffeine for attention, mood, reaction time, and athletic performance. But I prevously assumed that these benefits were not meaningful or relevant, for two reasons: (1) in most of the studies, the improvements they measured were only a result of withdrawal relief, and/or (2) the improvements were a result of acute intake above tolerance level, i.e. stimulant effects that would not be sustainable day to day.

First, is (1) the case? Supporting my previous opinion, this study in 2002 reported that after neutralizing caffeine withdrawal, there were no additional benefits on alertness or mood. The abstract states, “Most studies of the effects of caffeine on performance have used regular caffeine consumers who are deprived at test.” But not so fast, this study from 2007 examined the same question and found that there was no impairment in caffeine habituated individuals after overnight withdrawal. So which is it? I don’t know if there is a settled answer. So let’s assume for the sake of argument that there are some benefits which are real. On to (2): are the benefits still accessible after habituation?

The 2018 review “Effects of caffeine on sleep quality and daytime functioning” looks at this. The paper’s main concern is that typical caffeine use may create a vicious cycle in which caffeine use harms sleep quality, which in turn drives the need for caffeine in order to compensate and function adequately during the day. While the conclusion is tentative, the paper does suggest a “net benefit” (their quotes) for caffeine use. For example, the paper calls out that tolerance develops in peripheral blood pressure but not in cerebral blood flow — from which the mental performance benefits are likely to issue. This suggests that some of these benefits might be sustainable.

To get a closer look at the supposed net benefits, I forked over 40 bucks for the full text of the 2010 review “Caffeine—​Not just a stimulant.” A number of the results reported in this paper pertained to habitual use, for example “the findings of two prospective observational studies indicated that the habitual daily consumption of any caffeine decreased significantly the risk for ‘cognitive failures’ (‘forgetting where you put things,’ ‘loss of concentration,’ ‘making the wrong choice’).”[1] While the term “significantly” can sometimes describe a statistically significant but small effect, still, this is reassuring to someone mostly trying to be sure there wasn’t a net harm. I was interested to see this review because I am skeptical of self-reported productivity increases. It seems too easy for a caffeine user to assume the subjective rush they feel translates to increased effectiveness compared to an alternate universe where neither caffeine nor caffeine dependence are in the picture. Rigorously testing performance impact on oneself would be an arduous project involving many months of back and forth, and I have never quite cared enough to undertake it (yet).

Besides potential negative impact on sleep, excessive stress hormone is another risk. This study found that elevated cortisol levels were not completely eliminated with habituation, and Forbes published an article arguing that caffeine is the “silent killer of success” because it causes the body to secrete adrenaline, which sabotages emotional intelligence.

However, the widespread intake of caffeine — including in the world’s longevity hotspots, such as the “blue zones” with disproportionate numbers of documented centenarians — makes it hard to think of caffeine use as such an emotional impairment. My logic is that chronic stress is probably one of the biggest anti-longevity factors, a theme that comes out in Dan Buettner’s book. Of the five blue zones Buettner researched, coffee is regularly consumed in three of them and in a fourth, Okinawa, they “nurse green tea all day.”

Then, of course, there are the well-established benefits from polyphenol content of beverages like tea and coffee. Even caffeine itself may have benefits in staving off certain neurological disorders. I won’t link any articles here, this seems to be plenty talked about. But this does count as an important factor in the plus column. Though I guess if we want to be thorough we can note a con in that these beverages stain teeth, somewhat like the fictional juice of Sapho.

stupid things faster

Based on my previous ad-hoc use of caffeine, I’d developed the view that consuming it necessarily creates a “mind racing” or tunnel vision effect which impairs lateral or creative thinking. I thought it might be useful for cranking widgets — doing a mindless or boring task — but not for “deep work.” However, this now seems like an unfair characterization given that these are effects from acute intake above toleranace level, not necessarily seen after habituation.

So after looking into this in the last few months, I’ve had to revise my previous stance. There’s pretty good evidence that daily caffeine use can be a net positive.


I’ve had occasional migraine headaches since I was in my early twenties. The first one I ever got took place during the same summer that I started drinking coffee for the first time at work. Coincidence?

Barry Spencer doesn’t (or maybe, didn’t) think so. He wrote a website called “Caffeine and Migraine,” which seems to have fallen off the internet, but you can find a 2007 version on the Wayback Machine. Spencer’s hypothesis is that all migraine is caffeine withdrawal and they might even be the same condition. From his FAQ: “Many migrainers report caffeine abstinence. Nevertheless there is not one demonstrated example of a person who consumes no caffeine but gets migraine anyway. …​ Could so many people be mistaken about their caffeine intake? Yes. Caffeine is commonplace in foods, drinks and medicines, and is often occult (hidden).” This is an extreme view, and I include it here because such bold hypotheses amuse me. But it’s true that caffeine is hiding everywhere. It’s in freaking deodorant.[2]

In any case, most sources do acknowledge a link between caffeine and headache including migraine. An article on the American Migraine Foundation website states that “caffeine consumption is rarely the sole ‘cause’ of frequent headaches including migraine. However, it is a modifiable risk factor, unlike many other unavoidable migraine triggers.” The uncontroversial relationship between caffeine and headache was one reason that I tried to stick to abstinence. Yet over the years I have still gotten migraines, once or twice a month on average. I’ve never been able to track down any particular trigger other than caffeine. There could be a psychosomatic element, like maybe Dr. Sarno’s TMS phenomenon (and I can recall one time where I am sure I got a migraine purely from stress). But the point is that caffeine abstinence, performed to the best of my ability over many years, didn’t solve my headaches.

Spencer suggests caffeine maintenance as a valid approach, and I’d never given it a serious try.

Behavioral and evolutionary considerations

Various sources say that 85 to 90 percent of human adults in the US consume caffeinated beverages. I have to admit that there’s an iconoclast in me who hates giving up the distinction of principled caffeine abstainer.

A friend once called Seattleites “coffee dryads,” referring to mythical creatures whose essence is bonded to a particular tree, and if one of the creatures strays too far from its tree, the creature dies. It has put me off to see some avid coffee users spend time and energy throughout the day on preparation, consumption, paraphernalia, etc. related to the beverage. It does seem a bit like being chained to a tree (or a bathroom). I’m hoping to set up a routine that is dead simple, and save the ritual and paraphernalia for special occasions.

Tea and coffee plants, for their part, have managed to stumble on an evolutionary niche that has secured their survival, because our species now cultivates them on massive scale. If you’re interested in an exploration along these lines, I recommend Michael Pollan’s audio essay on Audible. He argues that the industrial revolution would have been impossible without these plants.

The experiment

Success criteria for the experiment:

  • No decrease in sleep quality or related impacts (e.g. on short-term memory, mood).

  • No increase in headache frequency.

  • Ability to occasionally ingest modest amounts of extra caffeine (<50 mg in the morning e.g. extra cup of green tea, or <10 mg in the evening e.g. cacao/chocolate or decaf coffee) without getting high or disturbing sleep.

My primary concern with taking up caffeine maintenance is preserving sleep quality. Since the half life of caffeine is about six hours, 1/4 of any consumed amount is still in the body after twelve. Sleep specialist Matthew Walker recommends your last caffeine intake occur 14 hours before you expect to turn in. So that implies a consumption window of the first 2 hours of the day. Therefore, the basic plan is to drink green tea first thing each morning upon waking up. Part of the experiment is finding the right amount, but I’m starting with one cup (~35 mg).

With the once-per-day pattern, it may create a rhythm with a daily peak and trough. It remains to be seen how well this will work or if more frequent intake through the day is more effective. Having to follow a complicated dosing schedule would start to count in the negative column for me. I might be able to handle 2x/day (adding a bedtime routine).

I started out on 8/15/20 with a teaspoon of matcha green tea, which is reportedly about two grams containing 75 milligrams of caffeine. (I didn’t find the bamboo whisk to be very effective and settled on preparing it with a blender.) This amount was a lot for me, and within a few days I cut back to 1/4 tsp. For the first week, I peed a lot and could tell I was mildly dehydrated (veins on the backs of my hands standing out). I was also easily startled, like at the grocery store I jumped when someone came up next to me at the produce aisle. I experienced short sleep every day, waking up after only about 7 hours, though this didn’t seem to be accompanied by greater fatigue during the day. In the second week, these impacts diminished. Given this experience I would recommend a more gradual taper up, just like you would want to taper down.

I’ll update as the experiment unfolds.

1. (1) Smith AP. Caffeine, cognitive failures and health in a non-working community sample. Hum Psychopharmacol 2009;24:29–34. (2) Carrier J, Fernandez-Bolanos M, Robillard R, Dumont M, Paquet J, Selmaoui B, et al. Effects of caffeine are more marked on daytime recovery sleep than on nocturnal sleep. Neuropsychopharmacology 2007;32:964–72.
2. I emailed Lumē to ask them why their dedorant contains caffeine, since it’s not addressed in the FAQ on their website. Their reply from 8/21/20: “The caffeine percentage in our formula is low and intended to have only a topical effect. It makes Lume more effective and is part of the science behind the efficacy of Lume. Caffeine’s ability to be absorbed by the skin and enter the bloodstream is poor and immeasurable at the percentages we use in our cream, so you shouldn’t have any issues if you have a sensitivity to caffeine. The amount of caffeine in an entire bottle is less than one cup of coffee.” I’m estimating 2 mg per application if we assume one cup of coffee has 200 mg and the deodorant stick has 100 uses in it. That’s indeed pretty low, and anecdotal evidence on efficacy of absorption through skin wasn’t strong. So they may be right, but the point is that total caffeine purity is a challenge in today’s world.