Learning the Value of Food: Mechanisms of Decision Making in Appetitive Behavior

A key debate in nutrition research is whether highly processed foods uniquely reinforce consumption and if, in being uniquely reinforcing, highly processed foods cause overeating and obesity. To that end, overarching goal of my dissertation was to examine the mechanisms through which food motivates appetitive behavior and how relative differences in these processes predict differences in eating behavior.

Here we cover three studies where participants are asked to make choices to earn food rewards. In each study, we examine how participants form action-reward contingency estimations and reward magnitude estimations through learning and how their choices reflect differences in the underlying decision processes. In chapter two, we show that when you reinforce behavior with a food rather than money, differences in reward magnitude estimation predict differences in behavior. Our results demonstrate that a key feature of natural reinforcers like food – that reward magnitude diminishes as a function of consumption due to changes in internal state – is observable in participants’ decision-making. In chapter three, we find that a large subset of children, whose choices appear as if they are trying to lose, were using a choice strategy that would be nearly optimal given a plausible (albeit false) assumption about the experimental environment. Our results indicate that children can form computationally complex models of their environment; however, their choices can be maladaptive if those models are incorrect. In chapter four, we examined how individual differences in action-reward contingency estimation and estimated reward magnitude for food rewards predicted differences on validated clinical measures associated with weight gain in children. We show preliminary evidence that overall reward magnitude for snack foods, as well as change in reward magnitude, are associated with outcomes on the Child Eating Behavior Questionnaire and the Eating in the Absence of Hunger Task.

Taken together, our results demonstrate that highly processed foods could affect how people make reward magnitude estimations and ultimately make decisions about what and how much to eat. However, the extent to which decision processes affect clinical outcomes like risk for weight gain is unclear. We show that while differences in reward magnitude estimation may not affect cognitive processes underlying the choice of when to stop eating, it may affect choices about portion. Highly processed foods could meaningfully affect portion size selection by altering expectations about how consumption affects internal state. Given the strong association between portion sizes and overall consumption, this suggests a that highly processed foods could trigger meaningful differences in caloric intake by changing what and how much people put on their plates.