I'll let a press release from Science Daily do some of the explaining for me:
The key switch player in this is a transcription factor called Foxa2. Transcription factors are proteins that make sure other genes are activated and converted into proteins. Foxa2 is found in the liver, where it influences fatburning, but also in two important neuron populations in the hypothalamus -- the region of the brain that controls the daily rhythm, sleep, intake of food and sexual behavior. The control element for Foxa2 activity is insulin, in both the liver and the hypothalamus.
If a person or animal ingests food, the beta cells in the pancreas release insulin, which blocks Foxa2. When fasting, there is a lack of insulin and Foxa2 is active. In the brain, the scientists have discovered, Foxa2 assists the formation of two proteins: MCH and orexin. These two brain messenger substances trigger different behavior patterns: the intake of food and spontaneous movement. If mammals are hungry, they are more alert and physically active. In short, they hunt and look for food. "If you watch a cat or a dog before feeding it, you can see this very clearly," says [lead researcher Markus] Stoffel.
The researchers discovered a disorder in obese mice: in these animals, Foxa2 is permanently active, regardless of whether the animals are fasting or full. This explains a well-known but until now unaccountable phenomenon: the lack of movement in obese people and animals.
To prove this, the researchers used a genetic trick to breed mice, in the brains of which Foxa2 is always active, regardless of whether they have just eaten or are fasting. These mice produce more MCH and orexin and move five times more than normal animals, in which insulin deactivates Foxa2 after eating or which are obese. The genetically modified mice lose fatty tissue and form larger muscles. Their sugar and fat metabolism works flat out and their blood values are considerably improved.
To simplify even further: hungry mice were more active.
Starving people with eating disorders tend to be more active as well. Excessive exercise is very common in people with eating disorders, and is associated with higher levels of anxiety and somatization (that is, physical ailments brought about by psychological stress). Although most people with EDs cite exercise as a way to lose weight or otherwise self-regulate, it may be driven by other biological factors as well.
So why would biology be prodding an organism to get moving when common sense would indicate that they should be resting and conserving every last calorie? One explanation is that a more active animal will move further afield to seek out food. Sitting around won't get you fed; seeking out food just might. Short-term, this is a costly strategy, as there is no guarantee there will be food anywhere else, either. But long-term, you'll definitely starve if you stay in your den where there's no food, so it makes sense.
Of course, for people with eating disorders, the problem isn't the lack of food as much as it is an inability to eat the food that's already there. The body, however, doesn't really care why you're starving. It just knows you are and prods you to go get soemthing to eat, dammit!
The results also help explain how re-feeding, including regular meals and snacks (Stoffel and his snacks-are-bad schtick can go bite me), can help ED sufferers decrease excessive exercise.There are models of what is termed "activity-based anorexia" in rats, where an animal on a restricted feeding schedule ultimately runs itself to death on an exercise wheel (Epling, Pierce, and Stefan, 1983). Researchers have looked at the role of leptin (Hillebrand et al, 2005) and a-Melanocyte-Stimulating Hormone (Hillebrand et al, 2005b) in activity-based anorexia, with some very interesting and promising results. This latest research only adds to the hormones that may help regulate energy balance in people.