Multivariate representation of food preferences in the human brain

- Investigating the decoding/prediction of food-preferences by multivariate fMRI.
- Category-dependent preference signals found in vmPFC, dACC, dlPFC.
- Category-independent preference signals found in dmPFC, dACC, dlPFC.
- Found activation in dlPFC is new in this context.
- dlPFC activity reflects food-domain specific aspects of preference processing.

One major goal in decision neuroscience is to investigate the neuronal mechanisms being responsible for the computation of product preferences. The aim of the present fMRI study was to investigate whether similar patterns of brain activity, reflecting category dependent and category independent preference signals, can be observed in case of different food product categories (i.e. chocolate bars and salty snacks). To that end we used a multivariate searchlight approach in which a linear support vector machine (l-SVM) was trained to distinguish preferred from non-preferred chocolate bars and subsequently tested its predictive power in case of chocolate bars (within category prediction) and salty snacks (across category prediction). Preferences were measured by a binary forced choice decision paradigm before the fMRI task. In the scanner, subjects saw only one product per trial which they had to rate after presentation. Consistent with previous multi voxel pattern analysis (MVPA) studies, we found category dependent preference signals in the ventral parts of medial prefrontal cortex (mPFC), but also in dorsal anterior cingulate cortex (dACC) and dorsolateral prefrontal cortex (dlPFC). Category independent preference signals were observed in the dorsal parts of mPFC, dACC, and dlPFC. While the first two results have also been reported in a closely related study, the activation in dlPFC is new in this context. We propose that the dlPFC activity does not reflect the products' value computation per se, but rather a modulatory signal which is computed in anticipation of the forthcoming product rating after stimulus presentation. Furthermore we postulate that this kind of dlPFC activation emerges only if the anticipated choices fall into the domain of primary rewards, such as foods. Thus, in contrast to previous studies which investigated preference decoding for stimuli from utterly different categories, the present study revealed some food domain specific aspects of preference processing in the human brain.