There's considerable amount of evidence that willpower is severely diminished if blood glucose get down, and this effect is not limited to humans. And a small sugary drink at the right time is enough to restore it.

We're talking really small numbers. Total blood glucose of a healthy adult is about 5g and it varies within fairly limited range. Then there's maybe 45g in total body waters. Then there's about 100g of glycogen in liver, plus yet larger amount in muscles and other organs, but which doesn't seem to take part in sugar level regulation. For comparison a small can of coke contains 33g - a really small amounts at appropriate times can make a big difference.

This leads to two issues. First, is blood glucose a good explanation for willpower deficiency and therefore akrasia? I'd say there's significant amount of evidence that some effect exists, but is it really the most important factor? Humans are complicated, science knows very little about how we work, and probably half of what it "knows" is false or at best only half-true. Caution is definitely warranted.

And the second issue - if this theory was true - and by manipulating blood glucose levels you could achieve far greater willpower whenever you wanted, what would you do? It seems that exploiting it isn't that easy, and I'd love to hear if any of you tried it before.

New to LessWrong?

New Comment
43 comments, sorted by Click to highlight new comments since: Today at 3:15 AM

Really, does it actually matter that something isn't a magic bullet? Either the cost/benefit balance is good enough to warrant doing something, or it isn't. Perhaps taw is overstating the case, and certainly there are other causes of akrasia, but someone giving disproportionate attention to a plausible hypothesis isn't really evidence against that hypothesis, especially one supported by multiple scientific studies.

From what I can see, there's more than sufficient evidence to warrant serious consideration for something like the following propositions:

  • Application of short-term willpower measurably expends some short-term biological resource
  • Willpower "weakens" as the resource is depleted, recovering over a longer time span
  • Resource expenditure correlates with reduced blood sugar concentration
  • Increasing blood sugar (temporarily?) restores resource availability

So, my questions are: If this is correct, what practical use could we make of the idea? What could we do as individuals or as a group to decide whether it's useful enough to bother thinking about? Particularly in cases where willpower is needed mostly to start a task rather than continue it, if there's a simple way to get a quick, short-term boost that might make the difference between several hours of productivity vs. akratic frustration, that's significant!

As an aside, I recall seeing some studies indicating that there may be more general principles in play here, regarding the mind's executive functions as a whole, but I don't have citations on hand at the moment.

I'm wondering if akrasia is partially caused by inefficient use of willpower.

I do remember these cool juice pouches I'd poke with a straw as a kid. I'd walk around with a bandolier full of them. Or maybe I'd snack on bananas throughout the day.

Actually, before taking steps to saturate myself at whatever level brings me no measurable willpower-deficiency, I'd investigate whether there's a possible training effect to enduring subjectively harder rigors while under the low-glucose condition. Plenty of people regularly don't eat at all until the evening and don't report feeling horrible after adjusting.

In other words, will I become dependent on my glucose-drip (constant snacking) so that I'm worse off if it's interrupted than if I'd never heard of this theory?

Even if I would become dependent with regular use, I'd still employ it at key moments in my life.

if this theory was true - and by manipulating blood glucose levels you could achieve far greater willpower whenever you wanted, what would you do?

I would wait for someone to provide cheap and easy implementation methods (a wristwatch that doubles as a body-monitor and expert system?), so I wouldn't have to do all kinds of primary research using very poor methods like self-experimentation, and wouldn't have to spend money on imprecise testing kits.

There is a cheap and easy way to stabilize blood sugar called ketosis. Anecdotally, suffice it to say that until I saw this article, I'd practically forgotten that I ever had an interest in akrasia.

Googling, there seems to have been a glucose monitor wristwatch since 2005 (and mentions back to 2001), eg.

So one could do this, absolutely. I don't think one needs an 'expert system' - presumably there's some level you don't want the level to fall below, and once you figure out the level...


The device didn't work very well, caused burns, and is (I believe) no longer being manufactured. Also, the level of accuracy that is useful to diabetics (who are concerned that their blood glucose may be at 200 or at 50 mg/dl) is not going to suffice for someone who is trying to stay within the optimal part of the normal range, which is probably a range of like 85-110 mg/dl. And by all accounts it wasn't even that good. Bummer, really.

And the second issue - if this theory was true - and by manipulating blood glucose levels you could achieve far greater willpower whenever you wanted

Logical error: (diminished X correlated with diminished Y) does not imply (increased X causes increased Y).

Among other things, there could be some other variable Z that causes both X and Y to diminish.

There can also be more than one resource bottleneck in the process - if other resources normally take longer to exhaust than glucose, all this does is make you run into problems later.

There may also be non-resource bottlenecks, like how much "executive" wiring you have. See the meditation-grows-your-brain studies, that seem to indicate willpower is like a muscle in terms of being able to be grown and made more efficient. So, removing the glucose bottleneck without increasing executive control, could be like having a fed runner who hasn't run a marathon before. You can give them all the glucose you want, but that won't necessarily get them to the finish line.

Following the logic of constraints, one can generate an almost unlimited number of plausible scenarios under which glucose will not produce any substantial improvements, even if glucose were a bottleneck resource in akrasia most of the time.... which it probably isn't.

Logical error: ...

You seem to have not read the linked research - consumption of sugary drink restores willpower in laboratory setting, and not one word you said makes any sense in light of this.

You seem to have not read the linked research

Actually, I was the first person to post the glucose-willpower hypothesis to lesswong, nine months ago.

consumption of sugary drink restores willpower in laboratory setting, and not one word you said makes any sense in light of this.

On the contrary - it does not in the slightest invalidate what I said. Glucose can be a bottleneck in some situations, and not in others. In those situations where it is, it will appear to be a magic bullet, and in all other cases it will fall flat.

That's not speculation, it's simple logic on the available evidence regarding the causal chain involved in willpower.

On the other hand, the reasoning you're using regarding glucose is plain old fashioned magical thinking.

Consumption of amphetamines also restores willpower in laboratory settings. The main question is what strategy is sustainable for optimizing mental performance. A one-shot lab study will not give salient answers to that.

You're technically right - laboratory experiments probe proxy measures in unrealistic settings. And that's the only kind of evidence we can really expect from them.

That's why the main thing I was asking about in this post is not if this theory is true (random speculation here will not answer this), but what would be the best ways to exploit it assuming it was true.

Anecdotally, I know that a mid-day soda seems to keep my prodcutivity higher - though if I drink too much, there's always risk of a crash. I had assumed it was the caffeine, though maybe the glucose plays a role as well. Anyone else have a similar experience?

ETA: I'll add that I've also long noticed that eating regular meals, especially relatively balanced and nutritious meals (relative to, say, an entire bag of potato chips), also seems to keep my productivity up.

It's good to study this sort of factor, but one of the points that pjeby makes in Improving The Akrasia Hypothesis is that there are multiple causes of akrasia. Dropping blood glucose may be one, and making blood glucose levels luminous may be a powerful technique, but it's not going to be a magic bullet.

but it's not going to be a magic bullet.

Why not? World is filled with magic bullets.

Since when? What are they?

What are you thinking of?

I would expect Amdahl's law (or the general principle behind it; is there some brief name for that?) to apply to this particular case.

Some magic bullets from the past:

If you look how difficult it was to figure out those, and/or how difficult it was to do such tiny changes even while knowledge of it became widespread, I don't find possibility of trivial solution to willpower problem unlikely at all by outside view.

What do you mean by "Amdahl's law"?

That essay on scurvy is fascinating, thank you for posting it.

The general idea behind Amdahl's law is that improving one thing may cause other things to be the limiting factor. The discovery that we could prevent scurvy just by getting enough vitamin C was great, and it all-but-eliminated scurvy-related problems from our lives, but it doesn't do much for other problems, like hurricanes, alcoholism, and monkey gangs. I personally would love to have a magic bullet that would prevent me from ever stubbing my toe. It wouldn't solve all our problems, but I have yet to find anybody, even masochists who get off on severe pain, who wants to continue getting stubbed toes. Amdahl's law isn't something you can get away from, but it's something you can work with.

Gustafson's law, which you linked to, isn't a rebuttal to Amdahl's law -- more like an addendum. If you have a really good solution to one problem, you can sometimes recast other problems in terms of it. If we ever get our shit together and start churning out Liquid Fluoride Thorium Reactors (PDF) in bulk, then it would do more than solve our electricity woes. We could take some other problems -- not enough fertilizer, shortages of fresh water, oil supply problems -- and recast those in a form that we can solve with LFTRs. The high temperatures from LFTRs can make ammonia, synthetic fuels for internal combustion engines, and fresh water. One solution becomes many solutions.

I personally would love to have a magic bullet that would prevent me from ever stubbing my toe.

I bought these shoes with reinforced toes. They're some sort of composite material rather than steel, so security checkpoints don't flip out too much, and no laces so they're easy to put on. Carolina makes 'em, for construction workers I think, but I wear them pretty much all the time, unless I'm sleeping or sitting in one place for hours. The only significant drawback I can think of is that eventually sweat accumulates. I suppose fashion might also be a problem, but valuing fashion over practicalities like mobility seems like a bad sign.

People have very strongly biased view against existence of magic bullets - most of the worst problem we historically had had been solved with a magic bullet. However - because the magic bullet solved them so well, the problems don't seem to have ever been that bad from today's perspective, and therefore magic bullets seem much less significant than they were.

An average person in let's say Medieval Europe's had biggest problem like:

  • risk of food shortage - magic bulletted with fertilizers
  • frequent epidemics - magic bulletted with clean water
  • death in childbirth - magic bulletted with basic hygiene
  • constant warfare and raids - magic bulletted with modern state
  • local banditry - magic bulletted with modern state
  • loss of housing due to fires - magic bulletted with building codes

If you start from a list of most serious problems, you'll see.

"Modern state" hardly seems simple and monolithic enough to be considered a magic bullet.

I'll leave it up to historians to figure out which part of modern state package caused this, yet the facts are - these problems existed since time immemorial, were some of the most severe ones, and are totally solved now.

I certainly agree that the situation has improved, but I'm not sure I'd call warfare, raids, and banditry 'totally solved.' Muggings, rape, burglary and riots still happen sometimes in civilized parts of the world. It's not just some mysterious, unpleasant thing from the increasingly distant past, like diptheria or smallpox or scurvy.

You're right Strange7, they're not totally solved. However I think taw's point has some merit. While the "Magic Bullet" didn't completely solve it's problem, it did ameliorate them to a huge degree. For example pre and post genetic-engineering farming is massively different... over 2 times as good. Doing the same for willpower would change society.

In practice, things that aren't magic bullets can't usually be fit enough symbiotic memes to overcome the noise in memetic evolution.

Thank you - this is the most insightful thing I've heard today.

I think "people" needs qualification here. I regularly encounter people who think such and such a discovery is going to be a magic bullet to cure cancer, or allow FTL travel, or some other major breakthrough (but those two most of all, I think.) So far, they've always been wrong, but that doesn't stop people writing articles on how this not-even-replicated finding is going to be the one to revolutionize our world forever.

Many people become biased against the idea of magic bullets because so many other people aren't, and so they're exposed to numerous prospective magic bullets, none of which actually work.

I would expect Amdahl's law (or the general principle behind it; is there some brief name for that?) to apply to this particular case.

It's Amdahl's law-like only if you're lucky; if it's a simple nonlinear constraint system, then you get a bullet that's magic if and only if you are currently sub-optimal for glucose and there are no other crippling resource shortages.

In all other circumstances, the "magic bullet" will do exactly nothing.

While local shortages of blood glucose in particular brain regions can cause willpower failures, you can't fix it by hacking your biochemistry because sustained global excess blood sugar has the opposite effect: crippling lethargy. Some possible workarounds do come to mind, but they all involve either nanotechnology or dangerous brain surgery.

What do you mean? Blood glucose levels can be tested at home (people with diabetes do that already) and if you somehow figure out how to eat sugar at appropriate times only you never get anywhere near sustained global excess, while limiting time spent in low blood sugar state.

Maintaining insulin sensitivity is a whole lot easier, practical and definitely healthier solution than fiddling with sugar intake. After all, in a well-rounded diet your high GI carb intake should be limited to a sane amount of sugar from fruits. Glucose and insulin spikes from consumption of simple carbs would promote insulin resistance on a longer term and render higher blood glucose levels useless from a willpower perspective. To that add hepatic damage and glycation damage from fructose intake.

In healthy individuals blood glucose is regulated extremely strictly; your brain shouldn't crash even several hours after a meal. The reason why so many people experience blood sugar crashes is that they're greatly overeating bad carbs and carbs in general. Currently I'm on a roughly 70:15:15 fat:carb:protein calorie ratio diet and my assessment of mental energy is in favor of it compared to my old ~50% carb diet.

In principle blood glucose can maintained in normal ranges even if there is literally zero carb consumption but sufficient protein consumption, as we're able to turn protein into glucose. This is a reason why carbohydrates are not considered essential nutritients by some people.

Do you have any evidence that willpower problems don't happen on lower carb diet?

My highly detailed food log says I've been eating about 33%:33%:33% calorie-weighed protein/carb/fat 1800kcal/day over the last 17 days, and I have about as many willpower crashes as ever.

Direct research evidence is pretty scarce at the moment. Anecdotal evidence is plenty from Immortality Institute, but we all know that anecdotal evidence amounts to little. On the other hand there is evidence for negative effects on cognitive performance from metabolic syndrome and diabetes, which effects can't be reversed by taking some sugar.

It's clear cut that sugar gives a cognitive boost, but we're not biologically accustomed to a constant limbo of blood sugar. Simple sugar was very scarce in the ancestral environment and in some ways it can be viewed as a psychoactive drug.

My approach is that it's better to first minimize long-term health risks and then try to find safe solutions for short-term problems. Treating mental fatigue with sugar is a huge long term risk and as a strategy it'll become unsustainable at some point. I'd note though that there are lots of other things that may affect mental energy/performance, like thyroid hormone levels, magnesium, Omega-3, etc.

33:33:33 is a very peculiar diet and I must say I've not yet encountered it before. 33% protein means 120+ grams of it per day, which is only common among heavy athletes and bodybuilders. My own 15 percent is also high-ish as I'm trying to get some muscle via strength training and thus supplementing with some whey. I suspect that 33% carbohydrate is still too much in your configuration (and 33 fat is too low).

Note that the whole idea is about taking very small amount of sugar at the right time. It's completely unrelated to levels that cause metabolic syndrome.

33:33:33 seems to arise fairly naturally when I try to eat reasonably limited amounts (for purely cosmetic temporary reasons; and all unrelated to the entire willpower business) of the tastiest food (most of which hits diminishing returns very quickly).

Here's some data on what people eat based on spreadsheetscripting out FAO data. As you can my current protein intake is indeed highly unusual - but then so are your very high fat and extremely low carb intake. Pretending what you're doing is the norm when it's really out of the norm is weird.

33:33:33 is similar to what is recommended for zone diet and paleo diet, so I cannot be the only person doing that. Similarity to zone diet purely accidental. Similarity of my constrained optimization for best taste with paleo diet is quite likely not accidental at all.

Your calorie intake is slightly high for the zone diet. That could be fine. The typical version of the zone diet is meant for weight loss and you need a higher amount of calories to maintain weight. The zone recommendation is to get those extra calories from healthy fats. The zone diet is also very concerned with maintaining the correct ratio for every meal and snack, not just as a long term running average. This makes sense if the goal is controlling insulin spikes after each meal.

I agree with Kutta that your protein consumption is much higher than is necessary. I am less clear on what the health consequences of that are.

As a 188cm tall somewhat more physically active than average male, my daily calorie expenditure is somewhere in 2500-2800kcal/day territory, so this is a decent calorie deficit to run for a few weeks.

And I wasn't aiming at any particular ratio. I just find many high protein foods like meat, cheese, eggs, and beans very tasty when cooked right. The log says nutrient ratio varies quite drastically day to day, 33:33:33 seems to be emergent longer term average.

You're missing the distinction between local and global blood glucose. A local low is when one particular area of the brain has used up its supply and has to wait for blood to circulate to replenish it. To prevent that from happening, you'd have to either increase the rate of circulation (ie, exercise in a way that increases heart rate), or increase the global blood glucose concentration to a level that's too high for the idle portions of the brain. A global low is something most people never experience, since the body stores and releases sugar to keep it from happening.

(I am a type 1 diabetic with various fancy equipment for tracking my blood sugar. I have personal experience as to what various blood sugar concentrations feel like, but they may not be representative of more typical biochemistries.)

If only there was some sort of fluid circulating in the body and providing nutrition to every cells which needs it... oh wait...

If you look at the experiments, like the one with the dogs linked above, a plain sugar drink at the right time improves willpower. These tests were all done on people and animals without diabetes, I can easily believe it won't work for you.

I didn't mean to imply that you can't increase willpower by managing blood sugar, but rather that the effect I described sets a limit on the total amount of benefit you can achieve this way. That is to say, while increasing your blood glucose from 80 to 100mg/dL is benefical for willpower, increasing it from 100 to 200mg/dL is disastrous. And most peoples' metabolism already maintains it at about the right point.

I haven't seen anything in research suggesting that above-normal sugar level is beneficial, just that below-normal sugar level harms willpower a lot.

If we could be near our best performance all the time, that would be enough.