The Unfinished Mystery of the Shangri-La Diet

Fol­lowup to: Be­ware of Other-Optimizing

Once upon a time, Seth Roberts (a pro­fes­sor of psy­chol­ogy at Berkeley, on the ed­i­to­rial board of Nutri­tion) no­ticed that he’d started los­ing weight while on va­ca­tion in Europe. For no ap­par­ent rea­son, he’d stopped want­ing to eat.

Some time later, The Shangri-La Diet swept… the econoblo­go­sphere, any­way. Peo­ple in­clud­ing some re­spectable economists tried it, found that it ac­tu­ally seemed to work, and told their friends.

The Shangri-La Diet is un­for­tu­nately named—I would have called it “the set-point diet”. And even worse, the ac­tual pro­ce­dure sounds like the wack­iest fad diet imag­in­able:

Just drink two ta­ble­spoons of ex­tra-light olive oil early in the morn­ing… don’t eat any­thing else for at least an hour af­ter­ward… and in a few days it will no longer take willpower to eat less; you’ll feel so full all the time, you’ll have to re­mind your­self to eat.

Why? I’m tempted to say “No one knows” just to see what kind of com­ments would show up, but that would be cheat­ing. Roberts does have a the­ory mo­ti­vat­ing the diet, an el­e­gant com­bi­na­tion of pieces in­di­vi­d­u­ally backed by pre­vi­ous ex­per­i­ments:

  • Your metabolism has a set point, like the set­ting on a ther­mo­stat: when your weight is be­low the set point, you feel hun­gry; when your weight is above the set point, you feel full.

  • But the set point is not a con­stant; it is raised and low­ered by what you eat.

  • This mechanism in turn seems to be reg­u­lated by a fla­vor-calorie as­so­ci­a­tion. (Pos­si­bly as a famine-stor­age mechanism that tries to store more re­sources when dense food sources are available.) If you eat some­thing with fla­vor X, which is fol­lowed by your metabolism de­tect­ing a large source of calories, fla­vor X will (a) seem more ap­peal­ing and taste bet­ter, and (b) will raise your set point when­ever you eat items with fla­vor X.

  • Your set point is always nat­u­rally drop­ping, but is raised by eat­ing; usu­ally these forces are in dy­namic bal­ance and your weight stays con­stant.

I’m not go­ing to go into all the ex­ist­ing ev­i­dence that backs up each step of this the­ory, but the the­ory is very beau­tiful and el­e­gant. The ac­tual Shangri-La Diet is painfully sim­ple by com­par­i­son: con­sume nearly taste­less ex­tra-light olive oil, be­ing care­ful not to as­so­ci­ate it with any fla­vors be­fore or af­ter, to raise your body weight a lit­tle with­out rais­ing your set point. Your body weight goes above your set point, and you stop feel­ing hun­gry. Then you eat less… and your weight drops… and your set point drops a lit­tle less than that… but then next morn­ing it’s time for your next dose of ex­tra-light olive oil, which once again puts your (de­creased) weight a bit above the set point. The reg­u­lar dose of al­most fla­vor­less calories tilts the dy­namic bal­ance down­ward. That’s the the­ory.

Many peo­ple, in­clud­ing some trust­wor­thy econ­blog­ger types, have re­ported los­ing 1-2 pounds/​week by im­ple­ment­ing the ac­tual ac­tions of the Shangri-La Diet, up to 30 pounds or even more in some cases. Without ex­pend­ing willpower.

I tried it. It didn’t work for me.

Now here’s the frus­trat­ing thing: The Shangri-La Diet does not con­tain an ob­vi­ous ex­cep­tion for Eliezer Yud­kowsky. On the the­ory as stated, it should just work. But I am not the only per­son who re­ports try­ing this diet (and a cou­ple of vari­a­tions that Roberts recom­mended) with­out any­thing hap­pen­ing, ex­cept pos­si­bly some weight gain due to the added calories.

And here’s the more frus­trat­ing thing: Roberts’s ex­pla­na­tion felt right. It’s one of those in­sights that you grasp and then so much else snaps into place.

It ex­plained that frus­trat­ing ex­pe­rience I’d of­ten had, wherein I would try a new food and it would fill me up for a whole day—and then, as I kept on eat­ing this amaz­ing food in an effort to keep my to­tal in­take down, the sa­ti­a­tion effect would go away.

It ex­plained why I’d lost on the or­der of 50-60 pounds—with what, in ret­ro­spect, was very lit­tle effort—when I first moved out of my par­ents’ house and to a new city and started eat­ing non-Jewish food. In ret­ro­spect, I was eat­ing an amaz­ingly lit­tle amount each day, like 1200 calories, but with­out any feel­ing of hunger. And then my weight slowly started creep­ing up again, and no amount of ex­er­cise—to which (ha!) I’d origi­nally at­tributed the weight loss—seemed able to stop it.

It’s always hard to pick re­al­ity out of the gi­gan­tic morass of com­pet­ing dietary the­o­ries. One of the el­e­gant charms of Robert’s hy­poth­e­sis is that it helps ex­plain why this is so—the mess of in­co­her­ent re­sults. Any new diet will seem to work for a few months or weeks, you’re los­ing weight and ev­ery­thing seems won­der­ful, you tell all your friends and they buy the same diet book, and then bam the fla­vor-calorie as­so­ci­a­tion kicks back in and you’re back to hell. The num­ber-one re­sult of weight-loss sci­ence is that 95% of peo­ple who lose weight re­gain it.

(I haven’t heard any com­plaints from peo­ple re­gain­ing weight they lost on the Shangri-La Diet, how­ever—if it works for you at all, it seems to go on work­ing. Most of the com­plaints on the fo­rums are from peo­ple who sud­denly plateau af­ter los­ing 30 pounds, but who want to lose more. Or peo­ple like me, who try it, and find that it doesn’t seem to do any­thing, or that we’re gain­ing weight with no ap­par­ent loss of ap­petite.)

I have a pretty strong feel­ing—I don’t know if I should trust it, since I’m not a dietary sci­en­tist—that Roberts’s hy­poth­e­sis is at least par­tially right. It makes a lot of data snap into fo­cus. The pieces are well-sup­ported in­di­vi­d­u­ally.

But I don’t think that Roberts has the whole story. There’s some­thing miss­ing—some­thing that would ex­plain why the Shangri-La Diet lets some peo­ple con­trol their weight as eas­ily as a ther­mo­stat set­ting, and why oth­ers lose 30 pounds and then plateau well short of their goal, and why oth­ers sim­ply find the Shangri-La diet in­effec­tive. The Mys­tery of Shangri-La is not how the diet works when it does work; Roberts has made an ex­cel­lent case for that. The ques­tion is why it some­times doesn’t work. There is a deeper law, I strongly sus­pect, that gov­erns both the rule and the ex­cep­tion.

The prob­lem is, though—and here’s the re­ally frus­trat­ing part—Roberts seems to think he does have the whole an­swer. If the diet doesn’t work at first, his an­swer is to try more oil… which is a pretty scary an­swer if you’re already gain­ing weight from the ex­tra calorie in­take! I de­cided not to go down this route be­cause it didn’t seem to work for the peo­ple on the fo­rums who were re­port­ing that the Shangri-La Diet didn’t work for them. They just gained even more weight.

And what re­ally makes this a catas­tro­phe is that this the­ory has never been an­a­lyzed by con­trol­led ex­per­i­ment, which drives me up the frickin’ WALL. Roberts him­self is a big ad­vo­cate of “self-ex­per­i­men­ta­tion”, which I sup­pose ex­plains why he’s not push­ing harder for test­ing. (Though it’s not like Roberts is a stan­dard pseu­do­scien­tist, he’s an aca­demic in good stand­ing.) But with re­ports of such dras­tic suc­cess from so many ob­servers, some of them re­li­able, out­side dietary sci­en­tists ought to be study­ing this. What the fsck, dietary sci­en­tists? Get off your butts and study this thing! NOW! Re­port these huge re­sults in a peer-re­viewed jour­nal so that ev­ery­one gets ex­cited and starts study­ing the ex­cep­tions to the rule!

It’s awful; it seems like Roberts has got­ten so close to bury­ing the scourge Obe­sity, but the the­ory is still miss­ing some fi­nal el­e­ment, some com­plet­ing piece that would ex­plain the rule and the ex­cep­tion, and with that last piece it might be pos­si­ble to make the diet work for ev­ery­one...

If we had a large-sized ra­tio­nal­ist com­mu­nity go­ing that had solved the group effort co­or­di­na­tion prob­lem, those of us who are metabol­i­cally dis­priv­ileged would be pool­ing re­sources and launch­ing our own con­trol­led study of this thing, and en­ter­ing ev­ery con­ceiv­able vari­able we could re­port into the ma­trix, and hiring a pro­fes­sional bio­chemist to an­a­lyze our metabolisms be­fore and af­ter­ward, and we would cry­op­re­serve any­one who got in our way. You have no idea.

(Warn­ing: Do not try the Shangri-La diet at home based on only the info here, there’s a cou­ple of caveats and I can’t think off­hand of a good com­plete de­scrip­tion on the ’Net. Also you might want to re­con­sider the recom­men­da­tion to use fruc­tose in the sugar wa­ter route, be­cause IIRC fruc­tose has been shown to con­tribute to in­sulin re­sis­tance or some­thing like that—su­crose may ac­tu­ally make more sense, de­spite the higher glycemic in­dex.)

Con­tinued in: Akra­sia and Shangri-La