Research points to a potential new focus for treating eating disorders

While the decision-making process unfolds similarly in healthy individuals and patients with anorexia nervosa, the brain regions that support food deliberation differ.

By Sarah Steimer 

Individuals with anorexia nervosa restrict food intake, sometimes to the point of starvation. And yet, a new study published in The Journal of Neuroscience finds that the cognitive machinery they use to make decisions about what to eat is remarkably similar to that of healthy eaters. Instead, the difference appears to lie in the brain regions that supply evidence to the decision process: Deliberating food choices among healthy individuals was associated with neural activation in the hippocampus, whereas individuals with anorexia nervosa (AN) engaged both the hippocampus and the striatum. The findings suggest that those with AN rely on different inputs that lead to restrictive food choices, rather than a maladaptive decision process per se. The findings provide a new point of focus for novel eating disorder interventions. 

To separate valuation from other aspects of decision-making, the research team — which included Akram Bakkour, assistant professor in the Department of Psychology — had participants make food choices and also complete a perceptual decision-making task, in which they judged the predominant color in a display of flickering dots. This allowed the researchers to test whether any differences in the decision process were specific to food or reflected broader alterations in how decisions are made. They combined computational modeling (specifically, a drift-diffusion model that captures both choices and reaction times simultaneously) with fMRI to examine the underlying cognitive and neural mechanisms of food choice. 

Consistent with prior research, patients with AN consistently chose low-fat foods and avoided high-fat foods, whereas healthy controls (HC) neither preferred nor avoided fatty foods. But once those preference differences were accounted for, the decision-making process itself looked the same across groups. Both groups were more accurate and faster on easier decisions, and the drift-diffusion model fit both groups’ behavior equally well, with no differences in key parameters such as the rate of evidence accumulation or the amount of evidence required to commit to a choice. On the perceptual task, the groups were similarly indistinguishable. 

Contrary to the researchers’ expectations, the findings indicate that AN does not involve a breakdown in the cognitive process of deliberation for either food or non-food decisions. Instead, these results shift the focus to the input to the decision process — i.e., the valuation stage — that leads individuals to choose one food over another. 

Where the groups did diverge was in the brain. fMRI results showed that during food choice deliberation (specifically, as decision time increased) HC showed greater activation in the hippocampus when compared to perceptual decisions. Patients with AN showed a similar hippocampal effect but additionally recruited the striatum. The results show that individuals with AN differ from HC in what regions of the brain are engaged during deliberation of food choice when compared with perceptual decisions. The researchers interpret the results to suggest that individuals with AN recruit different neural systems to compute and generate internal evidence as inputs to the decision process. 

“The hippocampus is at the center of this flexible memory system,” Bakkour says. “We think about episodic memory as being very flexible, and it allows us to travel back in time, but we can also think about the future. It's important for planning. The striatum, on the other hand, really is at the heart of a pretty inflexible memory system. It's been implicated in habit learning, for example. It helps us learn habits, but also to sustain habits over time.” 

He says this means the source of evidence for decision-making in AN may be drawn from a more rigid system, whereas healthy individuals draw on a more flexible one. The direction of causality is still unknown — it remains unclear whether striatal involvement is a cause or consequence of the illness — but as Bakkour and his colleagues approach questions around AN from a behavioral decision-making angle, the findings suggest there may be an opportunity to consider new approaches to treatment for an illness with high relapse and death rates. 

“That might give us some hope that maybe we can move things around,” Bakkour says. “If we can just tip the balance a little bit more toward the hippocampus in the patients, maybe we would normalize decisions.” 

He says the next step in their research is to focus on the input stage. He and his colleagues are looking to explore the attributes that contribute most strongly to choice and valuation, and how those might differ in patients with AN versus healthy individuals.