One fundamental and unanswered question in obesity research is what kind of foods contribute most to the condition. Experts variously blame, for example, fatty or sugary fare or foods that lack protein, which may prompt us, unconsciously, to overeat. Plenty of anecdotal evidence can be marshaled against any of the culprits, but there has been little long-term, large-scale experimental research on people’s comparative eating habits. It is neither ethical nor practical to have healthy subjects gorge themselves on one diet for years until they are obese.

It is possible, though, to conduct this sort of experiment on mice. For a diet study published this summer in Cell Metabolism, researchers randomly assigned one of 29 different diets to hundreds of adult male mice. (The scientists hope to include female mice in later experiments.) Some diets supplied up to 80 percent of their calories in the form of saturated and unsaturated fats, with few carbohydrates; others included little fat and consisted largely of refined carbohydrates, mostly from grains and corn syrup, although in some variations the carbs came from sugar. Yet other diets were characterized by extremely high or low percentages of protein. The mice stayed on the same diet for three months — estimated to be the equivalent of roughly nine human years — while being allowed to eat and move about their cages at will. The mice were then measured by weight and body composition, and their brain tissue was examined for evidence of altered gene activity.

Only some of the mice became obese — almost every one of which had been on a high-fat diet. These mice showed signs of changes in the activity of certain genes too, in areas of the brain related to processing rewards; fatty kibble made them happy, apparently. None of the other diets, including those rich in sugar, led to significant weight gain or changed gene expression in the same way. Even super-high-fat diets, consisting of more than 60 percent fat, did not lead to significant weight gains, and the mice on those diets consumed less food over all than their counterparts, presumably because they simply could not stomach so much fat. These findings were replicated in subsequent experiments with four other murine breeds. Male mice on relatively high fat diets became obese. The others did not.

"It looks like consuming high-fat diets, if they aren't extremely high fat, leads to weight gain, if you are a mouse," says John Speakman, a professor at the Chinese Academy of Sciences in Beijing and at the University of Aberdeen in Scotland, who oversaw the study. Speakman and his co-authors believe that the fatty meals stimulated and altered parts of the brains, causing the mice to want fatty food so much that they ignored other bodily signals indicating that they had already consumed enough energy.

The study was focused on weight gain, not loss, and its subjects were mice, of course, not humans. But the results are suggestive. Sugar did not make the mice fat, and neither did protein deficits. Only fat made them fat.