Decoding The Electrophysiological Response To High Calorie Food

Introduction: Eating is a complex behavior and disruption in eating disorders is not well understood. Previous studies have identified the nucleus accumbens, hypothalamus, and amygdala as regions involved during eating. However, further study is needed to understand the electrophysiological activity that underlying eating. Machine learning is a method that can decode stereoelectroencephalography (sEEG) activity and implicate brain regions involved in food-related behaviors. The objective of this study is to determine brain areas and their electrophysiological activity associated with attention to high calorie foods.

Methods: Participants were shown images of food (high calorie and low calorie), and images of non-food objects in random order lasting between 0-0.5 seconds. Task sEEG recordings were extracted and dimensionally reduced using principal component analysis (PCA). A supervised support vector machines (SVM) model was trained on 80% of the recordings to differentiate food and objects and high calorie and low calorie foods. The model was tested on the remaining 20% for accuracy. Patient's body mass index (BMI), hunger level, eating patterns were collected prior to testing.

Results: Our model successfully identified food from objects with up to 75% accuracy. The pars triangularis, hippocampus, and middle temporal gyrus were the most accurate in classifying food and objects. Our model classified high calorie and low-calorie food cues with up to 79% accuracy. The middle temporal and amygdala were the most accurate. The model did not identify an obvious electrophysiologic frequency to classify between different cues, though it seems beta and gamma waves were most accurate.

Conclusion : Our study establishes a successful model to classify brain response to food cues using sEEG, which validates previous fmri paradigms. We identified the middle temporal gyrus and amygdala as significant locations involved in food cues. Specifically, the amygdala was only significant in classifying low calorie and high calorie foods, which suggests it may not be involved in identifying food from other stimuli but important for associating reward with high calorie foods. This supports existing literature linking these regions to food addiction and obesity.