A recent review article titled " Theoretical perspective on anorexia nervosa: The conﬂict of reward," has to be one of the most fascinating scientific reads I've had for a long time (and my Facebook friends can confirm that I read a lot!). The gist of the paper is that many of the behaviors of AN, such as food restriction and excessive exercise, are initially rewarding, they eventually become punishing. An overlap in the neural circuits that process reward and punishment enables these two factors to become all knotted up, or "contaminated."
The author, Charlotte Keating, begins her argument with the concept of anhedonia, or an inability to experience pleasure, which is central to both major depression and a clinical feature of AN. Moreover, excessive exercisers tend to report greater levels of anhedonia, perhaps because exercise is being misused as a mood elevator. Initially, exercise and food restriction are very rewarding, which may be partly why people with AN become entrenched in these behaviors in the first place. Not eating feels better. Exercising feels better. Continued food restriction and excessive exercise only reinforces the reward, leading to the expectation that not eating and over-exercising will make the person with AN feel better.
The problem, says Keating, is that food restriction and excessive exercise are ultimately rather punishing behaviors. So how can punishing behaviors simultaneously be rewarding? The answer appears to lay in the anterior cingulate cortex, which (among many other things) is involved in the processing of reward, punishment, conflict, empathy, and other rational cognitive behaviors. In people with AN, the ACC doesn't process reward the same way; whether ultimately derived from dopamine circuits, reward is blunted in people with AN.
"...it may be that hypoactivity in ACC (which reﬂects the bulk of literature investigating this region in AN) reﬂects an impaired ability to adjust maladaptive behaviors which may also lead to illness maintenance."
Thus reward-punishment contamination means that the AN sufferer has a greatly reduced capacity for motivation to change, and to regulate his/her pathological behaviors. Furthermore, a low motivation for change only increases the neural "blurring" between reward and punishment.
The ultimate goal is not only to improve motivation to change by decreasing the blurring between reward and punishment in AN sufferers, but also to target "the mechanisms that may be responsible for bringing about behavior modiﬁcation."