This experiment isn’t finished but these results are too fascinating not to share:
Each point is a mean of 25 trials. The bars are standard errors.
Every evening around 9-10 pm I drank 2 or 3 tablespoons of flaxseed oil (with my nose clipped shut). Every morning around 7-8 am I measured my balance — how long I could balance on one leg on a small platform. (My balance-ometer.) I drank 2 tablespoons of flaxseed oil for 7 days then switched to 3 tablespoons.
The morning after the first 3-tablespoon dose was magical. From the very first trial I could tell my balance was better. It had always been hard to balance for much longer than 4 seconds. Now, all of a sudden, I could balance for 6 seconds, and even longer. More impressive to me than the conclusion that omega-3 (flaxseed oil is high in omega-3) had improved my brain function is (a) how easy it was to detect the difference between 2 and 3 tablespoons and (b) a fairly high dose (2 tablespoons/day of flaxseed oil contain about 14 g ALA, the shorter-chain omega-3) was less than optimal. Flaxseed oil labels recommend 1-2 Tablespoons/day.
Wow! I just increased my daily DHA supplement to 600mg, I guess I will up it some more
Exactly what is being compared when a dosage of”14g omega-3″ daily is said to be an amount that “everyone would consider a very high dose”?
Flaxseed oil contains about 7g of alpha linolenic acid (ALA) per tablespoon, fair enough. But what counts is the conversion of that ALA in the body into long-chain eicosapentaenoic acid (EPA) and docosapentaenoic acid (DHA). The standard literature suggests that such conversion is “unreliable and restricted”, and usually no better than 5% to 15% (and in some people, especially men, close to 0%).
That rate of conversion would yield:
2 Tbsp flaxseed oil => 14g of ALA => 0.7g-2.1g EPA+DHA (or less)
A daily dose of 0.7g to 2.1g of combined EPA and DHA is a very standard recommendation for distilled-fish-oil supplements, with package inserts recommending up to twice that much (and other sources still more).
Not knowing your personal conversion efficiency, is it not possible that 2 or even 3 Tbsp of flaxseed oil would give you amounts of EPA and DHA less than a standard dosage of marine-source capsules? And that the dramatic change in your data for 2 Tbsp vs. 3 Tbsp is further evidence for that possibility?
I’ve sincerely enjoyed reading your blog, but I’m still baffled that someone with your solid training in experimental design would report this data–with the lack of controls and unexplored measurement tool–as being meaningful.
If you were a doing peer review on an article with this research design, would you honestly green-light it for publication?
that’s a good point, diarist. I don’t trust those conversion ratios because the more of something to digest, the more the body makes the enzyme necessary to digest it, and the conversion ratios may have come from short-term feeding studies. Thus they would underestimate the conversion rate with long-term intake. But I haven’t looked at them. I read a paper about omega-3 requirements that recommended about 4 g/day — but they did not include ALA in that, I now see. I will fix the text.
“If you were reviewing an article with this research design, would you honestly green-light it for publication?”
This particular design is incomplete, as I said in the first sentence. Your comment is a bit like asking what’s the use of a newborn baby. I intend to add many more conditions to aid in the interpretation. However, ABA designs or ABABA or ABCDA or similar designs (of which the data I give could be a part) are common in the field of animal learning. Peer-reviewed articles in the best journals.
In my field (experimental psychology) few papers contain a single experiment. You do multiple experiments to answer the questions of interpretation that a single experiment usually raises. There’s no need for any experiment to be perfect. Because very few experiments stand on their own.
Lots more to be done — but this is an amazing result. Who would have thought that there would be such a short-term, dramatic benefit from increasing omega-3s? Congratulations on finding this.
You mentioned that you were going to do some tests of mental acuity as well as of balance — how are these looking?
Perhaps your balance got better because you have been practicing every day for a week? May be you should alternate 2 tbsp /3tbsp days and see if that makes a difference for you? Or start with 3 tbsp then go to 2 tbsp and see if your balance decreases. Or better yet, have someone else administer it alternating 2 tbsp of flax oil + 1 tbsp olive oil vs. 3 tbsp flax, etc…
The practice explanation is unlikely — my balance suddenly got quite a bit better. Practice improvements are more gradual.
Yes, I plan to do many more tests with different designs.
I’m still in the preliminary stages of the mental acuity measurements.
What about using something like chess for mental acuity? Unless you play regularly, it’s probably not suitable, but if you do, you might have reached a skill plateau, and that would make it suitable. You don’t have to go just by win/loss ratio, either. There are lots of chess programs that will give detailed number ratings for each move you make, which could probably be used to get a continuous performance number.
What’s the difference between flaxseed and fish oil? I note that my fish oil bottle says each gram tablet has 340 mg of omega-3, including 180 mg EPA and 120 DHA.
I don’t know how long it takes to replace enough of the lipids in the brain membranes to make a difference. I have a feeling it’s much more than one night, though. The placebo effect definitely cannot be ruled out here.
You can improve this experiment by making it effectively “double blind”:
Prepare doses of the omega-3 oil and a suitable placebo in small containers labelled with random numbers, mix them up and put the “decoding key” away until you have completed a trial run. You might want to put them in pairs where you only switch randomly within each pair, using the two doses of each pair on consecutive days.
Chess is a good idea. It requires heavy continuous computation, which is what I’m looking for and which most cognitive tests lack. I don’t think it’s practical because of the time-consuming opening game, unfortunately. I want something that will take no more than 20 minutes per testing session. Some other game?
Flaxseed oil has lots of the shorter-chain omega-3, ALA. Which is converted in the body to the longer-chain omega-3s, such as EPA. Fish have lots of EPA.
“I don’t know how long it takes to replace enough of the lipids in the brain membranes to make a difference. I have a feeling it’s much more than one night, though. The placebo effect definitely cannot be ruled out here.”
The initial improvement in my balance was a total surprise and cannot be due to a placebo effect. The other part of this comment — shouldn’t it take longer than one night to replace enough of the lipids to make a difference? — is what I thought. It is what all experimenters have thought, apparently, because these day to day measurements have never before been made. This is part of what is so interesting here, that the effects are much faster than expected.
Well, you could always use short time controls. On internet chess servers, twenty minute games (ten per player) are actually considered kind of long. If you’re not used to fast play like that, it might take a bit of getting used to. Blitz games, which are very popular, are 5/5 or 5/0 (five minutes per player, plus another 5 seconds per move, or 0 seconds per move).
But if you use computer analysis to rate the quality of your moves and give you a score that way, it doesn’t matter whether you finish a game (or get beaten early, and start another). It doesn’t even matter as much whether you win. So you could play for ten minutes at a stretch and not worry about finishing.
The way you’d do that is to have your chess engine think for some standard amount of time for each move, say ten seconds, and it would tell you who it thinks has the better position after (1) your move and (2) its favorite move. You’d subtract the scores to get the difference, and add up the difference over the whole game. The closer to 0 you get, the better. I think I could pretty easily set up an engine to do this to an arbitrary game with some scripts.
Another possibility for mental acuity metrics is to time the arithmetic sets in “Train Your Brain” by Kawashima. These take under 2 minutes per day and have been pretty stable for me once I got up the learning curve. (I had my best times ever the last couple of days since I increased my DHA supplements.) He also has some Stroop tests in this book, which could also be a useful metric and take less than a minute.
For the arithmetic sets, I just xerox them and start over after I get to the 60th one, seems to work ok.
Any comments about the idea of doing a blinded self-experiment?
These are all excellent ideas.
1. Computer chess might be fun. That would be a big advantage. of course. Other pluses: lots of mental computation; easy to describe. Minuses: Most people don’t play chess. Unclear how to deal with speed/accuracy tradeoff issue (if I take more time my moves improve).
2. Arithmetic sets. Pluses: easy to describe, easy to generate (I assume), easy to measure performance, little worry about speed/accuracy tradeoff (because accuracy always close to perfect). Minuses: Less fun than chess.
3. Blinded self-experiment. Pluses: More convincing. Minuses: Requires swallowing a lot of capsules.
“The practice explanation is unlikely – my balance suddenly got quite a bit better. Practice improvements are more gradual.”
I do not get your argument here. When it comes to balance you can dramatically improve your results during a single session. Trying to balance on a beam 5 times in a row most people will normal physical health will have a significant improvement from the 1st try to the fifth try. Going from 4 to 6 seconds should be no problem. I would expect something like, for example, 4 seconds to 30 seconds for the unathletic types and 30 seconds to 1.5 minutes for athletic types, if you get my drift.
Your general claim that practice improvemnets are gradual if we are talking about big improvements does not tell us anything about the specific case at hand. A fifty percent imporvement is big if were are talking about a change in how fast you can run 100 meters, but not if we are talking about balacing on a beam. A big improvment in the later case might be on the order of 500%
The improvements I am talking about are between session, not within session.
I have been measuring my balance for months. I have done it thousands of times. And then it suddenly got better. This is not how practice works. Between-session practice effects — such as the improvement from Day 1 to Day 2, or from Day 2 to Day 3 — get smaller as practice continues. For example, the improvement from Day 1 to Day 2 will be much larger than the improvement from Day 100 to Day 101. At the time this experiment began, the day-to-day improvements were quite small, as the slope of the lines shows.
OK
But why do think that you demonstrated such poor improvement when you had practiced the task thousands of times.
I am sure you almost always get the comment that your own expectations are affecting the results of the experiment (and that you are tired of hearing it), but with these kinds of physical tasks the degree to which you believe that you should be able to do it or to improve can make a huge difference. And this confidence factor can often explain a big jump in performance.
I think most climbers will tell you that they have had both periods of gradual improvement and leaps in performance, the latter having to do with attitude changes.
I am sure you are also aware of this potential explanation, but I just do not understand how you are controlling for this factor. Can it be done, or is attempting to demonstrate a link between Omega-3 and something as attitude dependent as balance not the kind of thing that can be shown with any confidence via self-experimenting. Again here it makes a big difference whether or not a jump from 4 seconds to 6 seconds is big for a person with your physical capabilities, i.e. not for you specifically but the average for all people with comparable physical capabilities to you since we want to rule out your expectations about what your body can do.
Could one have a control were you try to think that you will be able to do 8 seconds tomorrow and visualise yourself actually doing it and honestly try to build confidence that you can do it. Such a control might also require trying to convince yourself that your Omega-3 results are not real but due to an attitude change, and it is not obvious that it would be easy to detach yourself from previous results in this way.
“demonstrated such poor improvement when you had practiced the task thousands of times” — don’t understand.
Changing the dose from 2 T/day to 3 T/day did not change my attitude. And balance is remarkably unconscious. Wanting to balance better has no effect, as far as I can tell.
Throughout my balance testing I have always tried to do the best I could possibly do. To minimize the possibility of effects due to changes in motivation.