Fig 5. Satiety renders fMRI ensemble patterns of meal-matched odor mixtures less “food-like”.

(A) Schematic of decoding analysis. An SVM classifier was trained to discriminate fMRI activity patterns evoked by food (average of cinnamon bun and pizza) vs. nonfood (average of cedar and pine) odors in the pre-meal session. The SVM classifier was tested on fMRI activity patterns evoked by each individual mixture (3 mixtures per odor pair), and then food-like values were averaged across mixtures for meal-matched and non-matched conditions separately (cinnamon bun/cedar mixtures and pizza/pine mixtures). (B) The SVM classifier identified fMRI patterns as food-like less often for meal-matched mixtures than non-matched mixtures in the left and right olfactory/limbic ROIs depicted in Fig 3B (left, t₍₂₉₎ = 2.57, p = 0.02; right, t₍₂₉₎ = 3.78, p < 0.001). (C) Individual participant data depicting food-like classification values in olfactory/limbic ROIs for meal-matched vs. non-matched odor mixtures. Error bars depict within-subject SEM for n = 30. Individual participant data summarized in these plots can be found in S1 Data. fMRI, functional magnetic resonance imaging; ROI, region of interest; SEM, standard error of the mean; SVM, support vector machine.