Score a Point for Seth Roberts and the Shangri-La Diet

Earlier this week, we linked to a news article about a medical study finding that rats gained about the same amount of weight (80 grams, versus 72 grams on average) when they ate saccharine sweetened yogurt as when they ate yogurt sweetened with glucose. In both cases, the rats ate the yogurt in addition to their regular food.

If I understand the study’s conclusions, the researchers interpreted their results as evidence that the artificial sweeteners weakened the rats’ ability to make the link between sweet taste and calories. However, I don’t really see how this analysis works, because I doubt the rats’ regular food was sweet to begin with. So why would they eat more unsweetened food because of a destroyed link between sweetness and calories? I must be missing the point.

It seems to me that there is another explanation that fits perfectly with the findings of both this study and the theory underlying Seth Roberts‘s Shangri-La diet. As we wrote in a past Times column, Roberts bases his diet on the idea that, when you get your calories from sugar-laden foods, they provide a signal to your body to eat more, because the presence of sugar means that it must be a time of plenty. On the other hand, when your calories come from relatively tasteless food sources, they tell your body not to work too hard to obtain food, because your energy will be better spent waiting until the tasty stuff is available again. In this new experiment, adding the flavorful yogurt to the rats’ diets may have told both types of rats to eat all they could. Also, I would imagine that the artificially sweetened yogurt was even sweeter than the glucose flavored yogurt, in which case the signal to eat more was stronger for the saccharine-fed rats.

One thing I found peculiar about the study is that some of the rats refused to eat the yogurt, and consequently were excluded from the analysis. The study writeup lists every detail about the temperature of the room, and the schedule in which the lights were turned on and off; but, strangely enough, I couldn’t find any discussion of exactly how many rats refused to eat the yogurt. Given that the randomization was done based on which rats would get the different kind of yogurt, I would think that the researchers would want to report results for all the rats who were offered yogurt, not simply the ones who ate it.

And since when do rats refuse any type of human food? I always thought rats would eat anything. Charlotte’s Web apparently lied to me.


I don't think that an intent to treat protocol would be more appropriate here. The researchers aren't studying rats, they're studying sweeteners. Making sure each rat gets the correct dose of sweetener is important; making sure that the results are generalizable to the entire rat population is not.

In an epidemiologic study it is important to maximize statistical resolution, but in a laboratory study, where manipulations are controlled and iterative, statistics are one evaluation tool rather than an outcome in and of itself. In this case intent to treat would result in an unnecessary dilution of the measurement. These data will be used to generate additional testable hypotheses, not make dietary recommendations across a human population.


Sorry #5, but that just isn't correct.
If you eat the same quantity of butter and meat (weight), the meat is the one that's gonna make you feel fuller. You may feel sick from eating butter (eek!), but that's not the point. Your body is completely unaware of the calories in each type of food - take yogurt and meat, for example. If you eat 4 ounces of meat you'll feel fuller than if you eat 4 ounces of yogurt, even if the yogurt is more caloric than the meat. Your body processes the quantity (=weight) of food you eat, not the calories. If the body got fuller on calories instead of weight, nobody would surpass the 'normal' count of calories their bodies need every day, and we know that just isn't true.

#1: Dieting in a time of plenty is not what worries me about society today. It's eating all you can eat at restaurants when you don't need to be eating all you can eat, since we don't have a need to store food (especially in our stomachs). To quote the marvelous Jerry Seinfeld, "who needs to be eating all they can eat? We're not bears!"


Steve Peterson

I think Ray and Carl @ 12 & 7 have it right.

The effect isn't about the rats eating more -- it's about how the rat physiology reacts to the taste of sweetener by adjusting body temperature -- and one's body temperature does a lot of the business of burning calories.

My understanding of the theory was that, normally, when a rat eats something sweet it triggers the body to expect high calories shortly, so the body temperature rises and more of those calories get burned off.

But artificial sweeteners trigger that without providing the calorie boost. Over time this "trains" the body that sweet things don't precede calorie uptakes and so those rats stop having body temperature increases.

That's also why they need to control environmental temperature so much -- because they don't want environmental temperature polluting their experiment.

Lyn LeJeune

Dieting during a time of plenty....this speaks volumes about our culture. And I have been looking for some good research or cultural analysis...not extrapolated from rats, because I've known too many of those in my lifetime and they always gave me boxes of icky chocolate when they pissed me off-sorry to all the guys in my life (sing that song, Willie!0)... about the glut of protein items on the market. Everything now has to have protein and I mean everything. Protein water, protein cereal, protein bars with as much as 32 grams. Now for people who are anxiety-riddenand those arrested for road rage, listen up. Eat too much sugar, your body eventually goes nuts processing the stuff and you need more, more. Then you end up with hypoglycemia and need to cut out the sugar and gorge on the protein every two hours or you go bonkers just about when Valentine's Day or Halloween comes around. And if you're not a conspiracy food theorist yet, just go to the grocery store today and walk up and down the aisles. Now I'm going to eat a candy bar because I'm finishing the second book in my trilogy and I need a rush, then later I'll have one of those protein bars and swing along the street like a freakin' monkey.

Lyn LeJeune- Rebuilding the Public Libraries of New Orleans, The New Orleans Trilogy - all royalties to NOLA to rebuild the libraries, "When Ignatius J. Reilly Worked at the New Orleans Public Library and I Went Crazy at the Port-O-Call," at



Glucose is sweet, just not as sweet as sucrose.

Jirka Lahvicka

[3] I have to disagree - I ate quite a lot of pure glucose when I was a child and it was the best sweet taste ever. I believe that it is commnonly consumed by athletes and easily available (at least over here in the Czech Republic) so try it for yourself :)


Re: Protein overdosing -- unfortunately, a lot of people don't seem to understand the difference between (a) substituting protein for the processed sugars that make up the bulk of your diet vs. (b) eating the same old crappy diet but with extra protein on the top of everything. We saw the same idiocy years back when people realized the importance of fiber -- so suddenly they were eating the same crap and taking a fiber drink on top, instead of eating all their food in more fibrous, less processed form.

The whole diet industry is built on fads and ignorance, unfortunately. Even respectable, qualified dietitians are locked into a paradigm that a century of evidence doesn't support that narrowly focuses on calorie balance.


The effect may be much simpler and more direct than either the paper's authors or this blog proposes. Sweetness in the mouth "primes" the body to digest sugars (through some complex hormonal cascade), causing the body to digest more of whatever sugar is already in the gut, even if the rats can't find anything additional to eat. Add that to the increasing evidence mentioned in #2 about the importance of glycemic index and the insulin cycle...


Hmmm, I would really like to learn more about this. Unfortunately, a single study with a few rats that doesn't seem to propose a mechanism to explain its results doesn't seem like the way to get more information. I'm a believer in the glycemic index theory of weight gain - it seems to make evolutionary sense. However, it's hard to connect that with a taste, but evolution works in strange ways. All told, I'm going to violently withold judgement on this study - far more work is required.


Concerning the Shangi-la diet, it seems to me that ingesting a couple hundred calories between meals would do exactly what this claims to do: spoil your appetite. And it would do that taste or no taste. I am a skeptic when it comes to the premise that taste (psychology) can affect metabolism (physiology). Placebo arguments aside, is this diet any different than the one that tells you instead of 2 or 3 large meals a day you should eat 4-6 small ones a day? It's just in this case a couple of small meals are taken as oils or sugar water.


In the clinical trials literature, there is a name for the analysis you describe: "per protocol". Remove all the rats who don't take the treatment they are supposed to take. The analysis you hint at in the last sentence of the next to last paragraph is "intent-to-treat". In intent-to-treat, we compare all rats randomized to one treatment against all those randomized to control, regardless of whether the rats on treatment actually ingest the treatment.

To massively oversimplify a complex area of statistical (and econometric, epidemiological, etc) research, the researchers should be doing an intent-to-treat analysis, not a per protocol analysis.

Consider offering people in a study a life-saving treatment that most people will not take. If most people will not take it, life-saving or not it doesn't do a whole lot of good. The per protocol analysis will show a wonderful drug, while the intent-to-treat analysis will show a much more watered down effect. If you start assigning the treatment in practice, the intent-to-treat analysis will have been the one that better identifies the results that will be seen in practice.


Steve Peterson

I am a skeptic when it comes to the premise that taste (psychology) can affect metabolism (physiology).

In the article it shows that they tested precisely for this and discovered that, after being acclimated to the diets: "Sweet predictive rats [those fed the glucose] showed significantly greater increases in core body temperature than did nonpredictive rats [those fed the low-cal sweetener]."

Moreover, it appears that taste, amongst other similar cues, is well-established as having physiological effects (from the article):

"It is well established that orosensory cues (e.g., the taste, flavor, texture of food) can be quickly and strongly associated with the postingestive consequences of eating."


"Sweet tasting substances have also been identified as strong elicitors of a number of preabsorptive (e.g., hormonal, thermo-genic, metabolic) "cephalic-phase" reflexes related to ingestion (Berthoud, Trimble, Siegel, Bereiter, & Jeanrenaud, 1980; Bruce, Storlien, Furler, & Chisholm, 1987; Teff, Devine, & Engelman, 1995; Tordoff, 1988)."



Dr. Levitt: I find it disappointing that you and Stephen Dubner keep promoting Seth Roberts and his diet. The book has been out for nearly two years now and, in spite of many promises by the author, there is still no valid scientific evidence supporting the majority of his claims.

Since there isn't real evidence refuting his basic premises (as far as I am aware of) it would be unfair as of yet to put Mr. Roberts in the realm of quackery. But a good dose of critical skepticism is vital whenever someone makes medical claims without real evidence to support them. This is especially true when they are publishing a book on a subject that has a long proud history of making people rich who can charismatically tout bogus theories.

Perhaps its easy to forget for an economist, but some fields have high standards of evidence requiring double-blind trials and active experimentation. Nutritional studies shouldn't need to rely on stretched arguments based on observations of rats.



I do think that Levitt has missed the author's point here, and agree with the other commenters that there is no reason to invoke flaky dietary theories as an alternative to the science. In addition to the physiologic response to sweet taste, also know that artifical sweeteners can interact with receptors in the gut to trigger glut2 (part of the insulin regulatory machinery). And diet sodas have been tied to an increased risk of metabolic disease in two good studies. We've got a long way to go but the data has been moving in this direction for a while now.


We don't understand the complex mechanisms involved in diet so we're dealing with consensus of opinion that agrees in some areas and disagrees in others. One competing theory is that sugar spikes insulin, which drives a host of bad results and included in that list is craving and over-eating, partly it seems because sugars convert very quickly. (Whether insulin resistance is real or even necessary in the theory is questionable.) Carbohydrates are sugar; the body treats them as sugars and the Glycemic Index gauges how quickly the body transforms a carb into glucose (sugar).

Rats eat tons of carbs, so their diet is "sweet" to their body, even if they don't eat candy. One speculation is that sweet, sugar free foods trigger some unknown mechanism which connects the taste to the same craving, eating mechanisms which real sugars drive. In other words, since sugar burns so fast, it drives you to constant replenishment (of similarly fast burning sugars) and the fake sugar triggers the same urges.


Jon Plummer

If I remember correctly, glucose doesn't actually taste sweet, further damaging the idea that the effect is due to any sweet flavor.


I have to agree with your assessment that the rats had a higher signal. The study clearly shows that the rats feed artificial sweetner consumed more calories/energy over the study period. Either they randomly got some rats that could eat more or the artificially sweetened yogurt just tasted better.

Phil Birnbaum

Um ... couldn't it just be that since saccharin doesn't make you feel full like sugar, you just keep eating until you're full?

Your body obviously knows how many calories you're eating. You'll be full after four ounces of butter, but not four ounces of steak.

It follows that zero calories of sweetener probably just make you feel zero full, no? What's with all the complicated theories?


As an owner of several pet rats, I can say they'll eat almost anything, but they definitely have preferred foods, and different rats have different preferences.

I'm sure if they're hungry enough they'll eat almost anything, but a rat who is otherwise sufficiently fed might just not like yogurt very much, and prefer not to eat it.

Chris Lawnsby

One of the best posts in a while. I'd like to see more posts analyzing possible problems/complications in recent studies.