Warning, Soybean Eaters: Tofu Made Me Stupid

I’ve been testing my brain function daily for the last six years. I use a reaction-time test (see digit, type digit as fast as possible) that takes about five minutes. I have gradually improved the test over the years — this is about version 8. One reason for this testing is that I might observe a sudden change. That could suggest a new factor that affects brain function — whatever was unusual before the change (e.g., a new food). This is how I discovered the effect of butter. My score suddenly improved, I investigated. Another sudden change (improvement) happened soon after I switched from Chinese flaxseed oil to American flaxseed oil. I hadn’t realized that something was wrong with the Chinese flaxseed oil. I started brain tracking after I noticed a sudden improvement in balance the morning after I swallowed about five flaxseed oil capsules. Millions of people had taken flaxseed oil capsules, but no one, it seemed, had noticed the balance improvement. Maybe other big changes in brain function go unnoticed, I thought.

A month ago, my score suddenly got worse (= I became slower). There was one unusual thing that day, before the test: I had eaten a piece of bai fu ru, a kind of fermented tofu popular in China. To make it, “cubes of tofu are first fermented, then soaked in brines that contain a number of ingredients: rice wine, vinegar, chili peppers, cinnamon, star anise, and red yeast rice, the last which imparts the deep red hue that you’ll see in certain varieties.” Here is a discussion of fermented tofu in Los Angeles. It’s popular in China. Most people use it as a condiment, but I ate small amounts (one cube)  alone.

To find out if the bai fu ru caused the worsening, I did a test. I deliberately ate one 20 g cube at 11 am. I do the brain tests in the afternoon, usually 4-5 pm. What happen to my brain score that day? Here are the results.

 photo 2013-12-01tofuslowsdownbraintofuplot_zpsceffdd58.jpeg

Each point is a different test. I conclude that the tofu slowed me down by about 20 ms and the effect lasted two days. I didn’t notice the change in other ways — I didn’t feel tired or slow, for example. Presumably that is why this has gone unnoticed.

I think these results reflect cause and effect for several reasons:

1. Clarity. A t value would be very large.

2. Surprising prediction. A surprising prediction turned out to be true. Sharp drops like this are rare. Perhaps they happen once every 3-6 months.

3. Repetition. A research assistant found similar results, although not as clear. She is Chinese — quite different genetically.

4. Other evidence that tofu is bad for the brain. After I found these results, I remembered an old study. It found an association between midlife tofu consumption and late life cognitive decline among Japanese-American men in Hawaii (more tofu, more decline). A related study of the same men found a correlation between cognitive decline and miso consumption. A later study of Indonesian men and women also found a correlation between tofu consumption and cognitive decline (more tofu, more decline). The same study also found a beneficial weak correlation between tempeh consumption and cognitive decline (more tempeh, less decline). A weak epidemiological association found only once means little — epidemiologists, in my experience, do not adjust for the number of tests done and usually ignore the problem.  More recently, an experiment using a mouse model of Alzheimer’s disease found that a high-soy diet made brain function worse.

Because of the other evidence, I conclude that all tofu (and other soybean foods, such as soy milk) probably impair brain function, not just this version, which contains slightly more than tofu. The other evidence involves lots of non-fermented tofu. Because of the other evidence and my assistant’s results, I believe these results will be true for other people.

I can’t explain the effect. Tofu is high in omega-6, but the amount of omega-6 in 20 g of tofu is small. Others think that tofu impair brain function due to its isoflavones.

Most nutrition experts say tofu is good for you. Catherine Newman, in O Magazine, raves about it. “In addition to being wonderfully inexpensive, tofu is high in protein, low in fat, and very low in saturated fat. . . .  One daily four-ounce [= 120 g] serving is an excellent addition to a healthy diet.” The Mind Health Report (October 2012) says tofu is good for the brain. “For vegetarians, good choices are tofu, beans and eggs.”

Better informed experts criticize soy but say fermented soy is good. Joseph Mercola wrote,”For centuries, Asian people have been consuming fermented soy products such as natto, tempeh, and soy sauce, and enjoying the health benefits.” The Weston Price Foundation website says a lot about the badness of soy but claims fermented soy is healthy. In an article about why unfermented soy is bad, the Healthy Home Economist says, “Please note that fermented soy in small, condimental amounts as practiced in traditional Asian cultures is fine for those who have healthy thyroid function [as I do — Seth].” My 20 g dose was a small, condimental amount. John Robbins, author of The Food Revolution (2001), dismissed the association of tofu and dementia. He argued:

That’s not all we know. We know, for example, that dementia rates are lower in Asian countries (where soy intake is high) than in western countries. We know that the Japanese lifestyle (with its high soy intake) has long been associated with longer life span and better cognition in old age. And we know that Seventh Day Adventists, many of whom consume soyfoods their whole lives, have less dementia in old age than the general population. . . . A number of clinical studies have shown that soy and isoflavones from soy are actually beneficial for cognition. . . . . Having studied the literature, soy researchers Mark and Virginia Messina conclude that “there is no reason to believe that eating soyfoods is harmful to brain aging.” [Robbins failed to mention that Mark and Virginia Messina own a “nutrition consulting company specializing in soyfoods nutrition” — Seth]

I don’t know when the Messinas said this. Maybe there was once “no reason” to think soyfoods bad for the brain, but there is now.  The new evidence, epidemiology, and mouse evidence make a good case.

Here are two interesting things. 1. A very popular food is dangerous, maybe harmful. 2. Nutrition experts had claimed the opposite: the food is good for you. Even the better ones (Mercola, Weston Price Foundation) got it wrong, ignoring evidence (the epidemiology). It is a good example of experts overstating their understanding. The Shangri-La Diet is another example (every expert said sugar was fattening, I found it caused me to lose weight).

Even more interesting is the methodological implication. Anyone can do what I did, with any food. It’s easy, safe, cheap, and takes little time. The bad effects were large enough, and the method sensitive enough, that the bad effects were well above noise. And this is a non-trivial case. Tofu is popular.

We eat thousands of foods. We have millions of genotypes. We eat our food in millions of environments. Your genotype and environment affect how a food will affect you. So a good understanding of how foods affect us would seem to require thousands times millions times millions of tests. It is absurd to assume that anyone else (government, academia, industry) will do the necessary tests. They can’t begin to do the tests. In contrast to people doing a job (for example, people in government responsible for food safety) you have a much simpler problem: Is my food safe for me? You only care about one genotype (yours) and one context (yours) and you eat far fewer than thousands of foods. You can do good tests. You can test exactly what you eat in exactly the context you eat it. Compared to the present, where we extrapolate from epidemiology, animal tests, and the rare human experiment, the reduction in uncertainty and increase in generalizability is immense.

If you want to do experiments like this — test foods one by one — with my software (which requires Windows), please contact me. The tests require a training period of 1-2 months so that the scores during an experiment are roughly constant.

62 Replies to “Warning, Soybean Eaters: Tofu Made Me Stupid”

  1. Hello Seth,
    As an avid reader of your blog, I’m interested in your cognition tracking software. Please do send me a link.

    Thank you.

  2. Hi Seth, very interesting!

    I sent this to my sister, who has been a vegetarian for several years. Her comments:
    “I would maybe give more credit had there been any reference to GMOs. Because of how heavily this crop is subsidized, I will only purchase items with Organic Tofu/Soy. There are correlations that clearly show how GMO products affect performance, underlining how dangerous they are – including soy/tofu, but addressing many other items as well. Corn, anyone?”

    So just wondering if you noticed whether the tofu you had was organic or possibly GM? I noticed some of the studies you cite are older so perhaps those were done before there was much modification of soy products?

    Seth: The label is in Chinese but I doubt that it was organic. Whether it is GM, I have no idea. As you say the studies I cite involve people eating tofu 40 years ago, long before GM became widespread.

  3. Could it have something to do with the red yeast rice? RYR has statin-like properties; my GP is trying to get me to go on it to lower my LDL. Statins have a long history of negative cognitive effects — perhaps it’s not the tofu here? Could you try again using tofu w/o RYR?

    Seth: I used white fu ru. No red yeast rice. That’s red fu ru.

  4. Weston A Price foundation summary of harms of soy: http://www.westonaprice.org/soy-alert

    Chris Kresser discussing soy (repeats a lot of the points from WAP; +F for “Soy:”; toward the bottom): http://chriskresser.com/9-steps-to-perfect-health-1-dont-eat-toxins

    Pages of studies showing harm from soy: http://www.westonaprice.org/soy-alert/studies-showing-adverse-effects-of-soy

    Seth: Yeah, I said WAPF says soy is bad (“The Weston Price Foundation website says a lot about the badness of soy”). They also say that fermented soy is good — and I tested fermented soy. WAPF ignored the epidemiology in the sense that the epidemiology didn’t show that fermented soy was safe.

  5. Interesting, I noticed that when I came back to complex fiction books after NOT eating loads of soy I was able to read them faster, gain more info from them, and not struggle. As a soy-product eating vegan I had a VERY hard time with some cognitive tasks, now omnivorous and avoiding soy, I find these same tasks to be VERY easy. It’d be great to try your software and get some numbers to support this… 🙂

  6. Seth,

    Very interested in the possibilities of your software if you have a link and are willing to share I would like to test and give my feedback/story.

    I have been very closely watching my health and can feel the difference many times when I eat bread or pasta it slows me down but to what quantitative degree I don’t know? Yet.

    Austin Lutz

  7. Definitely interested in trying this. Would the software, as-is, work with supplement experiments as well? Either way, this is awesome. I’d love to give it a try if possible.

    I could see a community get behind this to create dozens of different heuristics that can test the effect of a single dietary choice on an array of different performance tests. There would be so much to discover, especially once you have a pile of data.

  8. I stopped at
    ‘A research assistant found similar results, although not as clear. She is Chinese — quite different genetically’

    ridiculous statement

    Seth: Like “loud” and “soft” and “heavy” and “light” “different genetically” is a relative term. There is always an implied comparison. I meant compared to a random person.

  9. Very interesting, but you really cannot draw conclusions based on an individual datapoint. You say the t-value would be very large, but if you mean the results of a statistical t-test, this is wrong, you have no degrees of freedom. Why not repeat the experiment a couple more times? Granted you’d be sacrificing a bit of reaction speed for a few days, and the test would still be imperfect because you have expectations about the results and couldn’t be blinded to the treatment, but I would find this much more credible if you repeated it even once more and saw the same spike in reaction speed.

    Seth: This was a repetition. I had seen the same drop earlier when I had eaten the same tofu in the same amount. A t test would compare the tofu day results with the baseline days before and after. I have no idea why you think there are no degrees of freedom. There are plenty… multiple tests per day and multiple days.

  10. I find it rather remarkable that a number of smart nutrition experts (e.g., Joseph Mercola, The Weston Price Foundation) think that unfermented soy is bad, but fermented soy is good.

    The idea that fermentation gets rid of the badness—to the point where the good outweighs the bad—is tough to swallow (pun intended), and apparently not well-supported by the evidence.

    I haven’t done the sort of testing Seth did, but I have had blood and intestinal tests confirm that I am soy intolerant—and I feel much better, now that I’ve given soy, of all varieties, up.

    I don’t remember all of the biochemistry behind this, but the lectins apparently cause problems in the intestines and bloodstream (and brain?), while the indigestible, bacteria-feeding starches, cause digestive problems as well.

    1. “I find it rather remarkable that a number of smart nutrition experts (e.g., Joseph Mercola, The Weston Price Foundation) think that unfermented soy is bad, but fermented soy is good.

      The idea that fermentation gets rid of the badness—to the point where the good outweighs the bad—is tough to swallow (pun intended), and apparently not well-supported by the evidence.”

      I agree, that is remarkable. Unlike a lot of nutritionists, Mercola and WAPF could say whatever they wanted, no one would complain, no funding would be lost, no reputation would be damaged. And yet they applauded fermented soy in the face of evidence — lots of evidence, if you read the WAPF website — that soy is bad or at least dangerous. I cannot explain it. Maybe they are driven more by general principles (“fermentation is good”) than evidence. I really don’t know.

    1. I disagree that there is something “Western” about this. Epidemiology involving people in Indonesia found tofu consumption correlated with dementia (more tofu, more dementia). And my research assistant got similar results. She’s Chinese.

  11. Question: a good way to double-check the results would be to do this on another day after a couple of weeks of avoiding soy, then, maybe to wait a couple more weeks and then do it again. This would show reproducibility. Would also help show whether the first result was a fluke/coincidence.

    Seth: A research assistant got a similar result. It is no fluke.

  12. I like experimenting with different diets.
    Would be great to see their cognitive effects.
    I’d like to try your software if I can.


  13. Unless you’re a truly avid label watcher it’s very likely you eat soy every day, it’s in bread, condiments, and many meat products.

    Also studies need to control for how the soy is processed, particularly the use of hexane.

  14. Hi Seth,
    thanks a lot for this article, espescially for the way you pay attention to the effect of your environment and intake.
    Devising, testing, and following through with such a testing method, requires a lot of consequence. I admire that.

    As I experiment on optimal nutrition myself, and also optimal exercise/training regimes, your software approach to measure influence of substances to the body could help me a lot.
    Is this the way to contact you? Or did you have something else in mind?

    thx and bye

  15. Hey Seth, my girlfriend is an avid soy lover, I told her this and she didn’t take it too well, I did try to sugar coat it as much as possible…..and I didn’t call her stupid at all.

    Anyway her reply that she is following the “Blood Type Diet’ and she is Type A, which this book says her blood type gets the most benefit from a mostly vegetarian diet including soy. She is under the impression that other blood types will get little benefit or even negative effects from soy (and a vegetarian diet). Where types A’s process it well and benefit from it.

    Out of curiosity, what was your blood type in relation to this experiment. Also was the blood types recorded in any of the other experiments you looked into?


    Seth: I don’t know my blood type, sorry. The epidemiology I described involved people of all blood types.

  16. Or maybe it’s simply that Mercola’s in the fermented soy business (selling K2 from natto) but not in the unfermented soy business (tofu.)

    Occam’s Razor.

  17. Hi Seth, thanks for sharing your insights.

    I’m also interested in your software. Look forward to hearing from you how I can get access when you have a moment. Thanks!

  18. I would be delighted to test my diet, which I have recently changed considerably (mostly enriched in iron significantly), simply because my SO is anemic and I take most of my meals along with her. So if I could get access to your software, I’d like to see how I score.
    BTW – do you have any methodology to test the effects of different foods on other bodily functions? (immune system, general alertness, sleep cycle, tendency towards allergic nasal congestion, metabolism and attention span are the main factors that affect my life)

  19. My experience with soy stuff was always bad.

    I do not measure reaction times. But my stomach is usually quite clear. and i tried it in many forms before, processed soy was almost always significantly bad.

    I tried soy protein in its various forms. Tofu was bad enough to make me not try it . But not as bad as the meat-like protein things.

  20. Seth:

    How do you compensate for the psychological sub-conscience effect of your tests? You ate soy repetitively (as you indicate above) HOPING for a slower reaction to prove your inclination. You can say you were trying as hard as you could for the best time/score, but your sub-conscience knows different; almost like a like-detector. I predict that you could repeat this test 1000 times and still have the same slowing effect.

    Your initial observation, and then your dissecting of your diet was probably the accurate interpretation, but when you repeat it numerous times LOOKING for a negative correlation, you will find it. [Likewise, recent research uncovered on Global Warming data indicates a similar situation where scientist were looking for Global Warming and some went out of their way to misinterpret data at the fringes to find it. Now that average temperatures decreases from those peaks, they call it “Climate Change”.]

    Anyways, the more scientific and truer test would be to consume the soy, and then something that isn’t soy to determine if there is a placebo effect on your results. But it would take a third-party to execute that for you and not telling you what you’re consuming. Take a heavily flavored soy-mild versus a flavored rice milk. And repeat it numerous times then look for correlations only after it’s revealed what you consumed after the fact.

    Thanks for the links in your post; I think the outside evidence is there which you seemed to have researched after you experienced the initial delayed reaction (#4 above).

    Seth: No I didn’t hope for the result I obtained. Actually I always hate it when there is a drop. I wonder if I will recover. But you are certainly right that the question is far from settled. I don’t agree that later better evidence will be “more scientific” — science needs all sorts of evidence.

  21. I’d like to try your software please. I eat plenty of soy and feel like this would be important to know for the long term effects. Thanks in advance!

  22. Hi Seth,

    This sounds really interesting. I would love to do the experiment. I would really appreciate it if you could send me the software.

    Thank you!


  23. Your findings are absolutely correct. Within the paleo community it’s no secret that soy creates brain-fog.

    Here’s the missing element: Soy is estrogenic.

    Which means it acts as some weird estrogene surrogate in your body and messes with your hormonal balance.

    Men can feel impacts quicker than women, because of naturally higher testosterone levels. But the long term effects for women are worse. The little testosterone women have gets basically wiped out. It can’t counteract the natural estrogene levels AND soy estrogenes on top.
    That’s why soy-rich vegetarianism is the most common road to breast cancer. At least become a pescetarian and avoid soy like the plague.

    Dave Asprey from http://www.bulletproofexec.com has been negatively addressing soy and its effects for years. Check out his research. He’s a very innovative thinker similar to you.

    1. “Here’s the missing element: Soy is estrogenic.”

      That’s an interesting explanation. I have two questions: 1. Why should estrogens cause brain dysfunction? 2. What other foods are as estrogenic as soy?

  24. I’m not an expert on the topic, just a well-read and heavily researching self-experimenter.

    The estrogenic effects of soy kick off a chain reaction slowing down the ways the brain is being “fueled”, which results in fogginess. Phytoestrogens are quite controversial and as far as I know allowed in the US, but if you want to feed it to your baby in Germany for example, you need a prescription. If you mess negatively with the dominant sex hormone of each sex, you do no good.
    There’s no heavy research on it, but as a self-experimenter you can do the following: do your self-reaction test, but instead of eating soy beforehand like you described above, eat some brazil nuts. Those are the most potent natural source of testosterone equivalents. I did that and it sharpened my thinking tremendously. I guarantee you will find similar results.

    There are many nuts and oil seeds containing phytoestrogens, but in my own diet I have only found flax seed to be as harmful to my brain as soy.

  25. Thank you for this very interesting finding. I’m a string believer in the effects of food on our (imediate) well-being and a health geek. I’m especially fond of your rigurous self-testing. (I’d love a link to the software you use by the way).

    But I wondered if you could enhance your self-experiment to prevent a placebo-effect. Since you might have read all those articles about tofu being not good fo your health, there is a possibility that your unconciousness might slow you down during the tests, just to make sure you’re right.

    So you would need not not know in advance, if you’re eating tofu or not. In addition you might want to make it double blind. So when you’re eating it, the person/the cook doesn’t know what you’re eating either.

    I thought about cooking up maybe five different dishes, so heavily spiced for example that you can’t taste if it’s tofu or something else. Then somebody packs them for you in boxes labeled with numbers. Which is which, you can’t know. Then you eat on of those every day, and later check if it really was the tofu all by itself.

    It is certainly i bit overengineering, but might be worthwhile to make sure, we’re not fooling ourselfs.

    Thanks again and all the best


    Seth: I had forgotten about one of the epidemiology articles when I did the test. The other one I never knew about. The food I tested here I ate for years. One day my score was surprisingly low and the only unusual thing about that day is that I had eaten the tofu. That’s why I did this test. None of this supports the explanation you propose.

  26. Here are some observations from a long-term N=1 experiment:

    I used to eat a strict macrobiotic diet consisting of brown rice, vegetables (mostly brassica), beans, seaweed, and relatively high proportions of soy (tofu, tempeh, miso, soy sauce), nut butters, as well as occasional artificial garbage like “Nayanoise” and “Rice Dream”. I did this for years.

    At first, I experienced weight loss. Eventually (about 3 years later), I noticed that I often had mood swings (no doubt caused by blood sugar changes), cravings, and low energy. I had what today I call “brain fog” most of the time, and difficulty concentrating. Most interestingly, I noticed that I had an increasing tendency to be more sensitive and emotional rather than rational. I attribute this to phytoestrogens in the soy. I believe the effect is quite real and measurable, although the change was gradual. I also had a tendency to feel cold all of the time (probably caused by inhibition of thyroid by brassica and autoimmune thyroid disease cased by gluten).

    After about 12 years of this, I gradually changed my diet to LC/Paleo, by first getting rid of the soy, then the rice and beans, then the seed oils and incorporating fish, meat, eggs, dairy, coconut oil, etc. My energy level is now far higher, my blood sugar is normal and stable, and the brain fog has cleared up. I have a far easier time concentrating. Finally, I went strictly gluten-free. That made a tremendous difference.

    I now think that in many ways the macrobiotic diet (as practiced in the early- to mid-90s) is about the worst diet one can possibly consume. It is extremely pro-inflamatory, incorporates all the wrong kinds of fats, high in gluten and free glutamate, and prone to cause blood sugar swings (observe any macrobiot and you will see that many of them can’t stop eating carbs). I think that the only conditions in which macrobiotics makes any sense at all is if one is in a state of metabolic semi-starvation.

    Seth: That’s very interesting. Given how bad a macrobiotic diet actually is, why did some diet guru decide that such an unhealthy diet was so great?

  27. So you’re not going to reveal the software? We’re all waiting.

    Seth: “Reveal” the software? I’m not sure what you mean. If you have a Windows computer and contact me, I may give you the software to use.

  28. Hi Seth,

    Please include me for a copy of your software. I typically perform a variety of nutritional tests, mainly around migraines and your software sounds like it can add an interesting angle.

  29. “Seth: That’s very interesting. Given how bad a macrobiotic diet actually is, why did some diet guru decide that such an unhealthy diet was so great?”

    It was a series of gurus, starting in the 50s. If you look at the macrobiotic diet as basically a traditional Japanese peasant diet, it makes sense and I believe it to be healthy; i.e. it makes sense if you maintain a state of semi-starvation, constant exercise, and no modern junk foods, none of which are at all common among modern American practitioners of macrobiotics.

    Many have used a macrobiotic diet to treat serious, advanced disease. There is far too much anecdotal to dismiss macrobiotics. Also, the macrobiotic diet is certainly better than the SAD. Modern macrobiotics has made some unfortunate mistakes, however, like treating all grains as equal, treating soy as a kind of health food, ignoring gluten sensitivity, etc.

    I remember reading a passage somewhere in “Good Calories, Bad Calories” about blood triglycerides becoming visible in a vial of blood, but not visible if in a state of semi-starvation. That fascinated me, and I immediately thought of my experience with macrobiotics and the constant teaching that semi-starvation was something to aspire to.

    Seth: Interesting, thanks. I wonder how well people sleep on a macrobiotic diet.

  30. Great post!

    “every expert said sugar was fattening, I found it caused me to lose weight”

    Just wondering if you have a post on your experience on this, I’d be very interested, thank you.

    Seth: Thanks. See my book, The Shangri-La Diet.

  31. “Seth: Interesting, thanks. I wonder how well people sleep on a macrobiotic diet.”

    I didn’t note any significant effect on sleep while following a macrobiotic diet.

    However, I did begin to sleep better on an LC/Paleo diet, and on Dr. Kruse’s Leptin Reset diet; the latter emphasizes no food after about 6:00 PM. When combined with exercise in the late afternoon (anaerobic; not cardio), sleep becomes deep and restful. The most profound effect on sleep that I’ve noticed in my N=1 experimentation was achieved by wearing blue-blockers after sunset, especially if using a computer in the evening. Wearing the blue-blockers also seems to have a pronounced effect on waking blood sugar. Wearing blue blockers also seems to elevate my mood (I am, after all, seeing the world through rose-colored glasses).

Comments are closed.