Simulating food environment- and individual-level interventions to shift ultra-processed food preferences

Food environments are increasingly filled with ultra-processed foods influencing food preferences. Extinction - the process by which learned food preferences may decline when foods are encountered but not eaten - can be leveraged in interventions. Most interventions that leverage extinction solely focus on the role of the individual, neglecting the role of the food environment. Moreover, computational models that seek to understand how food preferences are learned have not yet modeled extinction, limiting the ability to compare individual- and food environment-based interventions. Hence, we integrated extinction in a previous agent-based model in which agents learned food preferences based on rewards experienced after eating. Our results showed that extinction increased preferences for ultra-processed foods by suppressing learning for (un)processed foods, particularly in food environments where ultra-processed foods were abundant. Intervening in the food environment was more effective in shifting food preferences than intervening in individual behavior. Overall, the model suggests that shifting focus from individual- to food environment-level interventions may be helpful to mitigate health risks associated with ultra-processed foods, especially during formative periods of food preference learning. This work is supported by ZonMW (projectnumber: 05550032110022)