Decisions on a platter: Food biases and arousal alter reward learning

Abstract

Food seeking and avoidance are powerful drivers of decision-making in healthy and clinical populations. The urge to eat engages primary reward systems in the brain, thought to bias preference for energy-dense or high-calorie food more likely to sate hunger. This process is however interrupted in eating disorders where high-calorie food is avoided. It is nevertheless unclear how innate or learned food biases may interact with general reward processing to predict learning and decisions. We developed a novel paradigm to investigate whether and how biases for high- and low-calorie food alter decision-making in a large sample of participants with typical eating (‘TE’) and disordered eating (‘DE’) behavior. Importantly, food characteristics (high- vs. low-calorie) were completely incidental to the task goal of maximizing monetary reward. An arousal manipulation involving a large monetary win or loss, examined whether heightened arousal, thought to influence goal-directed behavior, enhances the impact of food biases on reward learning. Consistent with prior notions of food-relevant biases, the TE group learned better (i.e., made more correct choices) when high-calorie foods were rewarding while the DE group learned better when low-calorie foods were rewarding. The arousal manipulation boosted this group-dependent bias. Fitting behavior to reinforcement learning models enabled us to identify distinct cognitive components underlying the effect. Specifically, the impact of food-related biases on learning was best explained when accounting for group differences in the initial values and learning rates (for positive prediction errors) for high- and low-calorie foods. In other words, typical and disordered eaters showed differences in their pre-experimental preference for a food type (high- or low-calorie food, respectively), as well as the extent to which they learned from positive outcomes associated with that preference. These findings provide a mechanistic account of how food biases alter reward-based choice, especially under heightened arousal. Our results suggest that interventions altering innate or learned food preference should target habit-directed mechanisms to help mitigate maladaptive eating behavior.