Humans on a Cafeteria Diet

In the 1970s, as the modern obesity epidemic was just getting started, investigators were searching for new animal models of diet-induced obesity.  They tried all sorts of things, from sugar to various types of fats, but none of them caused obesity as rapidly and reproducibly as desired*.  1976, Anthony Sclafani tried something new, and disarmingly simple, which he called the "supermarket diet": he gave his rats access to a variety of palatable human foods, in addition to standard rodent chow.  They immediately ignored the chow, instead gorging on the palatable food and rapidly becoming obese (1).  Later renamed the "cafeteria diet", it remains the most rapid and effective way of producing dietary obesity and metabolic syndrome in rodents using solid food (2).


A few weeks ago, I was paging through Dr. David Kessler's book The End of Overeating, when I came across a citation for a paper with an extraordinary title: "Spontaneous Overfeeding with a 'Cafeteria Diet' in Men: Effects on 24-hour Energy Expenditure and Substrate Oxidation."  Further research revealed two more related papers, all from the research group of Dr. Eric Ravussin, who has dome some excellent work on energy balance, fat/carbohydrate/protein oxidation, and obesity (e.g, 3). 

The first study was published in 1992 (4), and seems initially to have simply been an attempt to design a novel way of accurately measuring food intake in free-living humans, which is notoriously difficult.  Investigators created an "automated food-selection system" consisting of two large vending machines filled with a variety of prepared foods of known calorie and nutrient compositions.  They recruited ten lean, healthy men.  At the beginning of the experiment, the investigators took four days to determine each volunteer's energy requirement for weight maintenance.  Then, in the setting of a metabolic ward where no other food was available, the volunteers were allowed to select and eat as much food as they pleased from the vending machines over seven days.  The foods available included English muffins, French toast, pancakes with syrup, scrambled eggs, chicken pie, cheeseburgers, margarine, white sugar, various cakes and puddings, apples, jelly beans, Doritos, M and M's, apple juice, 2% milk, sodas and several other foods.

I doubt the investigators were prepared for what they observed when they turned the ten men loose on those vending machines.  They immediately began consuming excessive calories, an average of 1,544 kcal per day in excess of their previously determined energy needs (with a fairly typical macronutrient composition by percentage).  That amounts to a roughly 60% increase in calorie intake over baseline, a striking change, particularly since it was completely voluntary.  Over the course of seven days, the volunteers gained an average of 5.1 lb (2.3 kg). 

The next two studies used a similar design and were both published in 1995.  One notable difference is that they were conducted in male and female Pima Indians, a population with a high susceptibility to obesity upon adopting an industrial diet/lifestyle.  Another difference is that these studies were conducted with a larger number of volunteers of various degrees of fatness.

The next study, whose title I mentioned above, was in Pima men (5).  After determining the volunteers' energy needs for weight maintenance, they set them loose on vending machines full of "familiar, palatable foods."  Over the course of the next five days, they consumed an excess of 1,637 kcal per day, a 56% increase over baseline.  33 volunteers out of 34 overate relative to baseline, including both lean and obese individuals, resulting in an average weight gain of 2 lbs (0.9 kg). 

The third and final study was in Pima women (6).  The design of this study was essentially the same as above.  The women overate by an average of 27% and gained 0.9 lbs (0.4 kg) over five days. The obese women consumed more calories, proportionally more fat, and gained more than twice as much body fat as the lean women.


Main Points

There are a few interesting things that can be gleaned from these studies:

1. Just as in rats, exposing humans to a variety of readily accessible, energy dense, palatable foods causes excessive food intake and rapid weight gain.  The degree of overeating varies by individual, but nearly everyone overeats to some degree.  Whatever the mechanism(s) underlying this may be, the phenomenon has important implications for the commercialization of food and the associated obesity epidemic in affluent nations.
2. In most cases, changes in body fatness are primarily, but not exclusively, the result of changes in energy intake. This is a consistent finding across many studies.  The obesity epidemic in the US has corresponded with a large increase in daily calorie intake, and also a substantial increase in energy expenditure, because larger bodies burn more energy.  Thus, energy "flux" has increased by roughly 400 kcal/day in the US since the 1970s (3).  
3. When energy intake is increased, energy expenditure also rapidly increases, although over the time scale of these studies it was not nearly sufficient to balance out the extra energy intake.
4. When a diet of mixed macronutrient composition is eaten to excess, the carbohydrate is preferentially burned off, while the fat is mostly shunted into fat tissue.  This makes sense, because why would the body go through the inefficient process of converting carbohydrate to fat for storage when it can just shunt dietary fat directly into fat tissue?  This does not imply that dietary fat is fattening under conditions of energy balance. 
5. Under the conditions of the experiment, obese Pima women spontaneously ate more calories and selected food with a higher fat and lower carbohydrate content than lean Pima women.  They also gained more than twice as much fat mass on average, perhaps pointing to differences underlying obesity susceptibility between individuals.  I believe this is similar to other observational findings on the subject.  

*High fat diets, particularly in combination with refined starches and sugars, were among the most effective.  The composition of these diets has been refined since then, and modern "purified" high-fat diets reliably induce obesity in susceptible strains of rodents.  The most commonly used diet is Research Diets D12492, which is 60% fat by calories, and composed mostly of lard, soybean oil, casein, maltodextrin, sucrose and cellulose (7).  It tastes kind of like raw cookie dough, and the rats are crazy about it.