Reinforcement learning and food-based decision-making

In the modern food environment, we are exposed to multiple food cues that influence the way we learn about food, shaping our eating behavior. Key aspects of eating behavior are derived through reinforcement learning, where environmental cues become associated with foods, obtain motivational value, and influence eating choices. This process is supported by two types of food reinforcement learning: Pavlovian and instrumental. Supported by preclinical research, both types of learning create a theoretical framework of how food cues can lead to food cravings, overeating and weight gain. The degree to which these reinforcement processes influence eating behaviors, weight and their underlying neural mechanisms is less understood in humans. To better describe this framework in humans, we investigated the neural correlates of reinforcement learning, their association with adiposity and future weight. Data from healthy adults (n=47) were used to examine whether inter-individual differences in food-based Pavlovian learning were associated with adiposity and brain functional connectivity (Aim 1). Individuals with increased propensity to assign motivational value to food cues (stronger Pavlovian learning) had greater waist-to-hip ratio and showed increased engagement of visual, salience, subcortical and mediofrontal brain networks during the passive viewing of food videos.Data from healthy adults (n=56) were used to examine instrumental control over food choices (Aim 2). Instrumental learning is characterized by two behavioral control systems: habit-based and goal-directed. Habit-based behaviors are automatic, whereas goal-directed actions are more conscious and deliberate. Previous work shows that habitual learning is associated with increased substance use and obesity. Here, habitual vs. goal-directed learning was not associated with adiposity nor with differences in brain functional connectivity. Data from healthy adolescents (n=121) were used to examine whether brain functional connectivity during exposure to food commercials was associated with future BMI (Aim 3). We found that stronger connectivity between regions implicated in memory formation, learning, visual processing, and self-control in response to fast-food commercials predicts lower BMI at a 2-year-follow-up. Together, these results suggest that individual-level characteristics can explain how different people assign motivational value to environmental food cues and suggest that neural differences in response to those cues can successfully predict future weight.