The insulo-opercular cortex encodes food-specific content under controlled and naturalistic conditions

The human brain uses sensory information to make perceptual inferences and guide behavior. Food availability cues trigger food-seeking and eating behaviors. Highly palatable foods can stimulate eating even when energy needs are met, and dysregulation of this process contributes to eating disorders and obesity. Rodent studies show that the insular cortex integrates visceral and gustatory inputs to generate homeostatic responses to food cues, with inputs from the lateral hypothalamus, amygdala, and nucleus accumbens. Insular cortex inactivation abolishes cue-provoked food-seeking. Human neuroimaging suggests insular and frontal operculum involvement in taste evaluation and representation of food availability cues, processes dysregulated in obesity. However, current human neuroimaging has limited temporal resolution, hindering precise characterization of insulo-opercular dynamics during eating. This study aimed to measure human frontal opercular and insular cortex activity during food processing, focusing on real-time responses during food anticipation. The hypothesis was that both food-specific and topology-specific anticipatory responses exist within the insulo-opercular cortex. Intracranial recordings from epilepsy patients were used during a task eliciting anticipatory and consummatory food responses, and during *ad libitum* meal consumption to assess generalizability to naturalistic settings. The hypothesis was that insulo-opercular regions showing food-cue-specific activity would also be involved in food evaluation during regular meal consumption.

Prior research extensively explored the neural mechanisms underlying food intake and the role of the insula. Studies in rodents demonstrated the crucial role of the insular cortex in integrating various sensory inputs related to food, including taste, smell, and visceral signals. These studies utilized techniques like optogenetics to demonstrate a causal link between insular activity and food-seeking behavior. In humans, neuroimaging studies, primarily fMRI, implicated the insula and frontal operculum in processing taste information and responding to food cues. However, these studies often lack the temporal resolution needed to precisely delineate the dynamic interactions between different brain regions during various stages of food consumption, particularly the anticipatory phase. Moreover, most previous research has been conducted under strictly controlled laboratory conditions, raising questions about the ecological validity of findings in real-world eating scenarios. This study bridges this gap by employing high-temporal resolution intracranial EEG recordings to investigate the real-time neural dynamics in the insula and frontal operculum during both controlled experimental tasks and naturalistic eating.

Eleven epilepsy patients (two female) with depth electrodes implanted in the insulo-opercular cortex participated. Electrode locations were determined clinically for seizure monitoring and varied across participants. Participants performed a milkshake paradigm, a widely used task involving anticipation and receipt phases. During anticipation, cues indicated either a palatable (chocolate milkshake) or taste-neutral solution. The cue was followed by a fixation period, then delivery of 3 mL of the solution, followed by a swallow instruction. High-frequency broadband (HFB; 70-170 Hz) activity was analyzed using the FieldTrip toolbox. Data were epoched, preprocessed (notch filtering, Laplacian re-referencing), and analyzed for time-frequency characteristics. Cue-responsive and cue-specific sites were identified using cluster-based non-parametric testing. Classification analysis (weighted KNN) tested the ability of posterior insula HFB activity to classify anticipatory conditions. Response onset latency was also calculated. Naturalistic eating was examined by analyzing SEEG and video recordings during ad libitum meals, focusing on the time point immediately preceding food entering the mouth. Time-locked HFB activity was compared between entrée and non-entrée items. Spike detection and interpolation were performed to mitigate inter-ictal activity. Chi-square analysis tested for associations between task-based and ad libitum responses. Statistical analyses were performed using FieldTrip, MATLAB, and other standard toolboxes. All procedures followed ethical guidelines and had IRB approval.

During the anticipation phase of the milkshake paradigm, the left posterior insula showed greater HFB activity for taste-neutral cues, while the left anterior insula showed greater activity for palatable cues. This suggests a functional specialization within the insula for processing different types of food cues. The frontal operculum showed earlier non-discriminatory activity than the insula. A single-trial classification analysis using posterior insula HFB activity achieved 64% mean TPR and 69% AUC in classifying the anticipated solution. During *ad libitum* meal consumption, time-locked HFB activity at the moment of food intake discriminated between entrée and non-entrée items. Regions showing food-specific eating responses were significantly associated with regions demonstrating cue-specific responses during the task, particularly in the left posterior insula. This suggests that the neural activity observed in the controlled task generalizes to real-world eating behavior. Latency analysis revealed a sequential activation pattern: frontal operculum activation is followed by insular activation, with the left posterior insula showing later activation, but stronger food-type specificity. No significant correlations were found between neural responses and subject BMI, milkshake rating, or age.

This study provides strong evidence for the involvement of the insulo-opercular cortex in anticipatory food evaluation, both under controlled and naturalistic conditions. The findings support a model where the frontal operculum plays an initial role in processing food cues, followed by more specialized processing in the insula, with a distinction between taste-neutral and palatable cues. The consistent activity in the left posterior insula across task and naturalistic settings underscores its critical role in food expectancy. The ability to classify anticipation trials based on insula HFB activity suggests potential translational applications for neuro-modulatory approaches to treat eating disorders. The distributed pattern of responses during food receipt might reflect a broader sensory integration process involving various sensory modalities related to food consumption. Further studies could investigate the effects of hunger and satiety on insular responses to gain a more comprehensive understanding.

This study provides novel insights into the spatiotemporal dynamics of the human insulo-opercular network during food intake. The findings demonstrate that the left posterior insula plays a critical role in anticipatory food evaluation, displaying food-specific encoding that generalizes across both controlled and naturalistic conditions. The sequential activation pattern of the frontal operculum followed by the insula suggests a hierarchical processing system. Future research should investigate the role of other factors like hunger and satiety, explore potential clinical applications of these findings in eating disorders, and further refine our understanding of the complex neural circuitry underlying food-related behaviors.

The study sample size was relatively small and consisted of epilepsy patients, potentially limiting generalizability to the healthy population. The electrode placement was determined clinically, leading to anatomical variability in sampling. While efforts were made to control for hunger and satiety, these factors were not directly measured. The naturalistic eating analysis relied on a limited set of food categories. The reliance on visual cues in the task may have influenced the results.