It started with an experiment on locusts in 1991. David Raubenheimer and Stephen Simpson, two zoologists who were at the time doing research at Oxford University, wanted to know what would happen to locusts if they varied the relative proportions of protein and carbohydrate in their diets. They therefore conducted an experiment in which they fed locusts pellets containing varying proportions of protein and carbohydrate, and the results astounded them so much that they ended up determining the course of their research over the next thirty years, which they’ve chronicled in their book, Eat like the animals.
What Raubenheimer and Simpson found was that the locusts were not eating until they’d satisfied their overall need for calories. Rather they ate until they’d satisfied their need for protein, so that overall, all the locusts were consuming the same total amount of protein. This meant that the locusts on the high protein diet were consuming much less food overall than the locusts on the low protein diet. Consequently, the locusts on the high protein diet became extremely lean, while the locusts on the low protein diet became fat (which they describe in their book as equivalent to an overweight knight squeezing in to a suit of armour that is a few sizes too small).
This led Raubenheimer and Simpson to conclude that protein is the dominant macronutrient in terms of determining how much we eat – At least if we’re locusts. They wanted to see if the same pattern would be seen in other species. They started off with flies, and the results were similar, which was encouraging. But flies and locusts are relatively closely related, at least in the sense that they’re both insects. What Raubenheimer and Simpson really wanted to know was whether they’d stumbled on a general dietary principle, that could be applied to all animals.
For reasons of practicality, they next chose mice. Unlike locusts and flies, which subsist pretty much entirely on protein and carbs, mice also eat fat, so in order to get a full understanding of how macronutrients impact body composition, this variable also needed to be part of the experiment. Additionally, Raubenheimer and Simpson wanted to increase the scope of their research, to look not just at the effect of various macronutrient combinations on body composition, but also on longevity. They were also curious to see what effect differing levels of dietary fibre would have on the mice.
The experiment took five years to carry out. 856 mice were sorted in to 25 different groups, that were fed identical pellets but with varying compositions of protein, fat, carbs, and fibre. They were followed from birth to death. In terms of body composition, the results were largely as expected. The mice fed a high protein diet all became lean and muscular. When it came to the mice fed a high carb diet, however, there was more variation. Those on a high carb diet that was low in fibre grew fat, while those on a high carb diet that was high in fibre remained slim.
The fact that fibre mattered so much to the body composition of the mice on a high carb diet is interesting. It provides a reasonable explanation for why people in traditional agrarian societies usually aren’t fat, even though their diets are very high in carbohydrates, and for why the current obesity epidemic coincided with a massive increase in intake of processed foods that were rich in carbs but lacking in fibre. It also provides an explanation for why people are able to lose weight both on a paleo/carnivore/keto diet that is low in carbs, and on a vegan diet that is high in carbs but also high in fibre. Fibre appears to provide a kind of “get out of jail free” card that lets you consume lots of carbs without becoming fat.
What about fat? Fat was found to be neutral in terms of it’s effect on how much the mice ate. In other words, fat intake didn’t have any limiting effect on appetite, so the mice on a high fat low protein diet grew fat, just like the mice on a high carb low protein diet that was low in fibre. If this result were to apply also to humans (which is, of course, not necessarily the case), it would suggest that LCHF/keto diets don’t work because people are replacing carbs with fat, but rather because they’re replacing carbs with protein.
Ok, so we know how the various macronutrient combinations affected body composition. What about the effect on life span? Here, the results as presented in Eat like the animals surprised me. Alot. The longest lived mice, according to Raubenheimer and Simpson, were the ones following a high carb low protein diet. Whether they ate a high or low fibre diet didn’t seem to matter. So the fat high carb mice were actually living longer than the lean, muscular high protein mice!
Baffled by these results, I decided to go and take a look at the data, to confirm that they weren’t just trying to pull a fast one, as nutrition researchers so often do when presenting their research. Hidden away in the supplement to the published study, is this table:
Two things immediately jump out at me. The first is that the group with the longest median lifespan was on a 42% protein diet. Hardly low protein!
If instead of looking at the median lifespan, we look at the maximum, we get a different picture. We see that the extremely low protein mice did best. But their median lifespans were far more average. The authors have obviously based the claims in their book, and in their published research article, on the maximum lifespan, rather than the median. That is something I find very odd.
Personally, I assume I’m going to live an average amount of time for people like me, following my type of lifestyle. I don’t assume I’m going to be the outlier who lives to 120! The median provides a much better picture of the effect of a diet on a group than the maximum lifespan seen in a few individuals.
Apart from that, they’ve chosen an odd definition of maximum life span. They’ve defined it as the top 10% with the longest life span in each group. Which is suspicious. Why the top 10% rather than just the top individual, which would be the more common way to define “maximum”? And why not the top 20%? Or top 30%? The definition really seems to have been chosen specifically because it gave the desired result, which is what is usually referred to as “torturing the data”.
I can only imagine that they chose to base their claims on their odd definition of the maximum rather than on the more appropriate median because the maximum showed a picture more in line with their own biases, possibly shaped by an environmental or animal rights agenda, or by the fact that it’s easier to get research published if it feeds in to the dominant dogmas.
The second thing that jumps out from the table is that the mice eating a high fibre diet (i.e. with a low energy density) lived much shorter lives than the other mice. That is by far the biggest difference, much bigger than any difference induced by varying protein or carb concentrations. Does this mean fibre is deadly and should be avoided it like the plague?
Well, no. The pellets that the mice were fed only contained one fibre, cellulose, which is hardly representative of the full spectrum of fibres that exist in real food. So it’s impossible to draw any conclusions from this about the effects of fibre on longevity. What we can say is that cellulose appears to be toxic to mice.
Next, I took the data from the table and re-tabulated it in a form that would allow for easier analysis of the data, which you can see here:
So what we see is that the low protein mice do appear to live the longest, but the differences between the groups are small and hardly linear. The difference between the 5% protein mice and the 42% protein mice is only 2 weeks, equivalent to about a year and a half if translated to a human lifetime. Since there’s no evidence of a linear relationship between protein intake and life expectancy, it’s hard to say that that result isn’t just caused by chance.
If we move on to carbs, then it again isn’t clear that the high carb diet leads to a longer life. The longest lived group is actually the one consuming a moderate 29% carbs, and again, there is no evidence of a linear relationship. The same is also true for fats.
So overall, the claims the authors make about a high carb low protein diet resulting in the longest life expectancy don’t hold up to close inspection. They’ve tortured the data until they’ve gotten the result they want.
What can we conclude?
If you want to be lean, muscular, and beautiful, then you should eat a high protein diet. If you just want to lose weight and be slim, then you can either go high protein or high fiber, or do a combination of both.
Well, as long as you’re a lab mouse, that is. Whether all of this also applies to humans is harder to say for certain. The results from the experiments mentioned here and others have led Raubenheimer and Simpson to develop the “protein leverage hypothesis” of obesity, which basically states that the modern obesity epidemic is due to the fact that modern diets are lacking in protein and fibre. This has come to be one of three main hypotheses that try to explain the rise in obesity. The other two are the “carbohydrate-insulin model”, which argues that the rise in obesity is due to the high consumption of carbohydrates and their downstream effects on insulin levels and thus body fat storage, and the traditional “calories in vs calories out model”, which argues that the rise of obesity is due to the fact that modern foods taste too good and are too readily available while our lifestyles have become too sedentary. From my perspective, Raubenheimers and Simpson’s hypothesis is the one of the three that fits the known facts the best. Their book, Eat like the animals, is therefore well worth a read, even though the claims they make about diet and longevity are unsupported by the evidence they present.