Investigating the cognitive, computational and neural underpinnings of food choice in healthy individuals

Choices are quintessential to our lives. Whether we realize it or not, we make thousands of them everyday, shaping our lives deeply. Among the choices we make, choosing what to eat represents one of the most important, as our lives depend on it. Food choice is a complex phenomenon, that is characterized by physiological, affective and cognitive determinants. These dimensions, however, have been studied either separately or jointly but without offering a comprehensive mechanistic approach that could pinpoint which cognitive aspects drive food choice. Most of the literature on the topic stems either from decision-making in cognitive neuroscience (Glimcher and Rustichini, 2004; Krajbich et al., 2012, 2015; Levy and Glimcher, 2011; Mormann et al., 2010; Pearson et al., 2014), considering food only as a special case of a general theory on value-based choice, or comes from experimental psychology and consumer research (Rozin, 1996; Shepherd and Raats, 2006), where it lacks a mechanistic approach to the neurocognitive determinants of food choice. In my thesis work, I aimed to bridge this gap by showing how we can study food choice combing the two lines of research. This could be done, I believe, by employing a set of different behavioral, neural and computational approaches. In my first Study I focused on the interaction of a cognitive on a physiological aspect (calories) in food choice, that is, how people understand and judge calories when it comes to choosing different food items. Calories are a fundamental element in driving choice about food, as they give us the energy we need to survive, grow and reproduce. But still little is known about the cognitive processes underlying their evaluation. In fact, it has been shown that the energy density (calorie content) of a food can bias the estimation of a portion size (Frobisher and Maxwell, 2003; Japur and Diez-Garcia, 2010), and vice versa (Wansink and Kim, 2005). However, we still do not know whether calories are considered as an absolute (total in a portion. Caloric Content, CC) or a relative (related to the type of food. Caloric Density, CD) quantity, which has implication as to the importance of controlling the size of portions in meals to contrast overeating (Rozin et al., 2011). I hypothesized that the type of food would be more important than the total amount of calories in a portion in determining how calories are understood, giving rise to known problems in estimating portion size. In my second Study, I wanted to investigate the neurocognitive mechanism of the interplay between a physiological (calories) and a cognitive-affective aspect (perception of risk). Safety concerns about food represent a powerful factor shaping food choice (Rosati and Saba, 2004), but little is known about how people choose food when it is at risk of being contaminated, with no knowledge about how different aspects of risk and reward (calories) interact with each other. I hypothesized that risk would drive the choice with participants choosing conservatively the least risky options, with a possible exception epresented by food with higher CD. What the results show is that overall participants were risk averse, while high calorie foods managed to partially counteract this tendency, i.e., some of them were chosen. Deactivation of the right anterior insula, a risk prediction error area (Preuschoff et al., 2008), as well as activation of the pre-SMA, implicated in working memory (d’Esposito et al., 1998; Pessoa et al., 2002; Petit et al., 1998), supported the idea that perceived differences in calories among the items made the task easier when it comes to risk evaluation. In my third Study, I wanted to investigate how a physiological element like hunger would impact on two components, one physiological (calorie) and one affective (preference) using a computational approach. While there is a wealth of research on the effects of hunger on food choice (Frank et al., 2010; Hoefling and Strack, 2010; Piech et al., 2010; Read and Van Leeuwen, 1998; Reisenman, 2014; Siep et al., 2009), we still lack a mechanistic approach that could help us examining the effects of hunger on calorie and preference at the same time. I hypothesized that hunger would make calories more important in determining choice, while satiety would, on the other hand, prioritize the role of preference in determining the pattern of choices. Using a drift-diffusion modeling approach I managed to show how hunger has opposite effects on calorie and preference, with participants choosing more high calorie and less preferred items when hungry and low calorie and high preferred items when satiated. Overall, with this work I aimed to show that the complexity of food choice can be stripped down to three basic dimensions (physiology, affect and cognition) and their interactions and that it is possible to study it, without too much sacrifice in terms of ecological validity, by using a template made of different analytic and quantitative approaches, from behavioral, to computational and neuroimaging.