#empirical #rate3 - [[brain valuation system]], [[orbitofrontal cortex]], [[lateral orbitofronal cortex]], [[ventromedial prefrontal cortex]], [[value-based decision making]] # Idea How value signals for food constructed by the brain? Given the distinct roles of the [[ventromedial prefrontal cortex]] and the [[lateral orbitofronal cortex]], the authors predicted that the lateral regions would encode food attributes (e.g., protein, fat, carbohydrates, vitamin, sodium, sugar, calories), whereas the medial regions would encode the overall subjective value. > p6. This result suggests that food value is computed at least in part through integrating information about elemental nutritive attributes. Leave-one-out cross-validation revealed that subjective value was best predicted by a model that included these features: **fat, carbohydrates, protein, and vitamin**. Note that sugar did not contribute to food valuation, likely because it highly correlated with carbohydrates. Subjective nutrient features were bettter predictors of subjective value than objective features. Used [[multivariate pattern analysis]] with leave-one-run-out [[cross-validation (in- vs out-of-sampling testing)]] to investigate the role of OFC in encoding the subjective value of food items. A [[linear classifiers]] was trained on patterns of [[functional magnetic resonance imaging]] response to categorize food items as being either high or low in subjective value based on each participant's ratings. ## Multivariate value representations in the brain Value representations could be decoded from medial parts of the [[orbitofrontal cortex]] at the time of valuation. Subjective value codes were also found in parts of the [[lateral orbitofronal cortex]]. Such representations were present during valuation, bidding, and feedback. ![[Pasted image 63.png]] ## Multivariate feature representations in the brain Subjective nutrient features could be decoded during valuation in the [[lateral orbitofronal cortex]], but not the [[ventromedial prefrontal cortex]]. These representations were absent during bidding and feedback. Thus, the [[lateral orbitofronal cortex]] represents information about nutrients only during valuation. ![[Pasted image 64.png]] ## Lateral OFC use distinct codes to represent subjective value and nutrient features Classifiers trained to decode each nutrient could not decode value information. Moreover, even after regressing out the effects of value from the nutrient ratings and fMRI responses to foods, the decoder could still decode nutrients in the [[lateral orbitofronal cortex]]. Distinct codes are also used by the [[lateral orbitofronal cortex]] to represent information about each nutrient. Subjective nutrient factors/ratings are more robustly represented in the OFC than objective factors/ratings. ## RSA analysis: lOFC but not mOFC Used [[representational similarity analysis]] to show that the lateral OFC but not the medial OFC neural responses reflected the similarity of the relative content of the nutrient factors. The lateral OFC specifically encodes information about value and nutrient factors but not low-level visual information about food images. ## Effective connectivity with psychophysiological interaction There was an increase in [[effective connectivity]] during valuation between the value-related mOFC subregion and the lOFC subregions representing nutrient features. ![[Pasted image 65.png]] ## Exploratory analyses Other regions of the [[brain valuation system]] also encode information about subjective value. However, only the [[posterior parietal cortex]] could decode all four nutrient features (but did not survive correction). Interestingly, even the V1 contained information about the four nutrient features. ![[Pasted image 67.png]] > p6. These results together may suggest a functional gradient from V1 to PPC and lateral OFC: V1 predominantly represents the visual information, lateral OFC predominantly represent the nutritive information, and PPC is the intermediate locus. However, because the PPC result did not survive correction for multiple comparisons across ROIs, this result should be treated with caution until it can be independently replicated. # References - [[Howard 2015 identity-specific coding of future rewards in the OFC]] # Quotes > p1. We hypothesized that the value of a food reward is at least in part computed by taking into account beliefs about the properties of the constituent nutritive attributes of a food item. We focused on beliefs about the amount of protein, carbohydrates and fat, and we also included beliefs about the specifically sweet carbohydrates (sugar), sodium and vitamin content contained in a food item. We further hypothesized that the OFC would play a role in representing these elemental attributes, which could thereby constitute precursor representations used to generate an integrated value signal. # Figures ![[Pasted image 62.png]]