Australasian Science: Australia's authority on science since 1938

What Two Experiments that Could Never Be Repeated Tell Us about Weight Loss

By Tim Olds

Starvation and overfeeding studies reveal extreme differences in how we gain and lose weight.

Seventy years ago, 36 men gathered at the University of Minnesota to begin one of the most famous experiments of all time – one that for ethical reasons could not be repeated today. Most of them were Quakers, all were pacifists and conscientious objectors.

The Minnesota Starvation Experiment was led by the charismatic Ancel Keys, a brilliant polymath with PhDs in oceanography and physiology, and undergraduate degrees in political science and zoology. As a child, Key himself became a subject in another famous experiment, Louis Terman’s 35-year study of extremely gifted children.

The aim of the Minnesota experiment was to explore the biology of human starvation, and to prepare for the feeding of tens of millions of displaced Europeans in the aftermath of World War II. The participants were underfed by 4200 kJ/day for 6 months, eating mainly turnips and potatoes, a regimen that would see them lose 25% of their body weight.

Forty-four years later, Canadian obesity researcher Claude Bouchard undertook an almost equally famous experiment that would also be very difficult to repeat ethically: he overfed 12 pairs of identical twin males for 22 days by 4200 kJ/day.

There had been other overfeeding experiments. In 1964 the Vermont Prison Overfeeding Experiment had five “volunteers” consume massive amounts of food (32–42 MJ/day), gaining on average 16.2 kg, of which 10.4 kg was fat. But what made the Quebec study unique was the use of identical twins, enabling the researchers to work out to what degree the responses to overfeeding could be ascribed to genetic factors.

These two experiments tell us a lot about how people respond to food intake, and why some people find it so much easier to lose weight than others. In the Quebec experiment, despite being overfed by exactly the same amount above what was required to keep their weight stable, individuals gained between 4.3 kg and 13.3 kg, with varying mixes of fat and fat-free mass. But within twin pairs, gains and losses were strikingly similar. If one twin gained 4 kg, the other also gained about 4 kg; if one gained 13 kg, the other gained about the same. Overall, 40–80% of individual responses could be ascribed to heredity.

The Minnesota experiment found a similarly wide variation in weight loss, and also differences in the amount of fat and fat-free mass lost. Some participants lost relatively large amounts of fat-free mass, some mainly fat.

How is it that two people can have the same excess or deficit in food intake and do the same exercise, yet changes in weight are so different?

First, even when physical activity is controlled – the original plan in the Quebec experiment was to have no exercise at all, but the participants rebelled so Bouchard took them for a walk once a day – there are various forms of “subliminal” exercise known as non-exercise activity thermogenesis (NEAT). This is basically fidgeting, which varies greatly among individuals. One study published in Science in 1999 ( overfed volunteers by 4200 kJ/day above weight maintenance requirements for 8 weeks. The differences in fat gain were ten-fold, 50% of which could be accounted for by differences in NEAT.

Second, there are different responses to feeding. When we eat, metabolic rate increases. This is known as the thermic effect of feeding, which varies with diet composition (it’s higher for high-protein diets) but also shows individual variation.

Third, there’s also an increase in metabolic rate after we exercise. This is greater if the exercise has been more intense, but there is also very large variation across individuals.

Fourth, there is individual variation in how much fat and fat-free mass (such as muscle) we gain or lose when we are overfed or underfed. In general, fatter people gain and lose more fat, while leaner people mainly experience shifts in fat-free mass.

There is, however, a lower limit to human body fat. In the Minnesota experiment, when the subjects reached about 5% body fat, only lean tissue was lost, a finding reinforced in a 1994 study of marines enduring semi-starvation on a survival course (

There are also genetic controls governing fat patterning – where we gain and lose fat from. Other factors potentially under genetic control, such as the amount of energy we retain as food passes through us, and changes in basal metabolic rate in response to diet may also play a role.

So a lot depends on the genetic lottery: if you are someone who fidgets a lot, feels the heat after having a meal and glows after exercise, and find that when you gain weight you get more muscular too, then you’re a genetic winner in the weight control stakes. If you’re the opposite, it doesn’t mean that diet and exercise are useless; it just means you have to try so much harder.

Life is unfair, so just suck it up. But the next time you’re with an overweight person, don’t be too quick to brand them as lazy and ill-disciplined because they may be swimming against a much, much stronger genetic tide than you are.

Professor Tim Olds leads the Health and Use of Time Group at the Sansom Institute for Health Research, University of South Australia.