Representations of Taste and Health Information in Anorexia Nervosa and Healthy Controls

Description: Patients with anorexia nervosa (N=20) and healthy controls (N=21) viewed a range of different foods and rated how tasty and healthy they thought each food was on a Likert scale from 1 to 5. Median splits were performed on taste and health ratings for each participant and used to label each taste/health rating trial as a high or low taste/health trial. These labels were used in searchlight analyses, which were conducted to examine the distribution of taste and health information across the whole brain. The searchlight analyses were conducted with a searchlight diameter of 5 voxels (i.e. 15mm) and four-fold cross-validation using PyMVPA (Hanke et al., 2009). The resulting searchlight maps were spatially smoothed with a 6mm full width at half maximum (FWHM) Gaussian kernel. To assess the statistical significance of the searchlight maps and to compare the searchlight maps for HC and AN, we used a non-parametric two-sample unpaired t-test against zero (after subtracting 0.5, chance decoding performance, from cross-validation accuracy scores) and corrected for multiple comparisons using threshold-free cluster enhancement (TFCE) with 5000 permutations. These statistical tests were performed using FSL’s randomise (Winkler et al., 2014). More information about significant clusters is included in tables here: https://github.com/alicexue/FCT_MVPA/blob/main/FCT_MVPA_searchlight_clusters.xls

Private Collection: To share the link to this collection, please use the private url: https://neurovault.org/collections/MHPZTYJS/