Implementation of a New Food Picture Database in the Context of fMRI and Visual Cognitive Food-Choice Task in Healthy Volunteers

Abstract

This pilot study aimed at implementing a new food picture database in the context of functional magnetic resonance imaging (fMRI) cognitive food-choice task, with an internal conflict or not, in healthy normal-weight adults. The database contains 170 photographs including starters, main courses, and desserts; it presents a broad-spectrum of energy content and is provided with portion weight and nutritional information. It was tested in 16 participants who evaluated the energy density and gave a liking score for all food pictures via numerical scales. First, volunteers were segregated into two groups according to their eating habits according to a food consumption frequency questionnaire (FCFQ) to assess whether the database might elicit different appreciations according to individual eating habits. Second, participants underwent fMRI cognitive food-choice task (van der Laan et al., 2014), using our picture database, in which they had to choose between high-energy (HE) and low-energy (LE) foods, under a similar liking (SL, foods with similar hedonic appraisals) condition or a different liking (DL, foods with different hedonic appraisals) condition. Participants evaluated correctly the caloric content of dishes (from r = 0.72 to r = 0.79, P < 0.001), confirming a good perception of the caloric discrepancies between food pictures. Two subgroups based on FCFQ followed by a principal component analysis (PCA) and a hierarchical ascendant classification (HAC) were defined, that is, Prudent-type (PTc, N = 9) versus Western-type (WTc, N = 7) consumers, where the WTc group showed higher consumption of HE palatable foods than PTc (P < 0.05). The WTc group showed a higher correlation between liking and caloric evaluation of the food pictures as compared to PTc (r = 0.77 and r = 0.36, respectively, P < 0.001), confirming that food pictures elicited variable responses according to contrasted individual eating habits. The fMRI analyses showed that the DL condition elicited the activation of dorsal anterior cingulate cortex (dACC), involved in internal conflict monitoring, whereas SL condition did not, and that LE food choice involved high-level cognitive processes with higher activation of the hippocampus (HPC) and fusiform gyrus compared to HE food choice. Overall, this pilot study validated the use of the food picture database and fMRI-based procedure assessing decision-making processing during a food choice cognitive task with and without internal conflict.

Keywords: brain; decision making; eating habits; healthy subjects; internal conflict.

FIGURE 1

Example of food pictures pairs, extracted from our food picture database, according to categories (starters, main courses, and desserts) and energy content [high-energy (HE) food choice and low-energy (LE) food choice].

FIGURE 2

(A) Functional MRI (fMRI) task general work plan. SL, similar liking condition; DL, different liking condition; HE, high-energy food choice; LE, low-energy food choice. (B) The fMRI paradigm adapted from van der Laan et al. (2014).

FIGURE 3

(A) Energy content repartition (in kilocalories) among starters, main courses, and desserts among the 164 food pictures from the database. (B) Evaluated energy content (scored between 1 and 10) according to real energy content (in kilocalories) of starters (Spearman’s correlation r = 0.77, P < 0.001), (C) main courses (Spearman’s correlation r = 0.75, P < 0.001), and (D) desserts (Spearman’s correlation r = 0.83, P < 0.001).

FIGURE 4

(A) Variables factor map from principal component analysis (PCA) on Food Frequency Consumption Questionnaire. (B) Individuals’ graph from hierarchical ascendant classification (HAC) analysis discriminating both Western-type consumer (WTc, N = 7, black) and Prudent-type consumer (PTc, N = 9, gray) groups (number of clusters required: 2–6, with clustering consolidation).

FIGURE 5

Palatable food frequency consumption in both Prudent-type consumer (PTc) and Western-type consumer (WTc) groups for (A) Ice cream (W = 13, ^∗P < 0.05), (B) Meat (W = 4, ^∗∗P < 0.01), and (C) Offal, Cold meat, and Eggs (W = 8, ^∗P < 0.05) food category.

FIGURE 6

Top: Evaluated energy (scored between 1 and 10) according to real energy content of main courses (in kilocalories) pictures presented to participants in both (A) prudent-type consumers (PTc) group (Spearman’s correlation r = 0.73, P < 0.001) and (B) western-type consumers (WTc) group (Spearman’s correlation r = 0.73, P < 0.001). Bottom: (C) Liking evaluation (scored between 1 and 10) according to evaluated energy of main courses pictures (scored between 1 and 10) in PTc (Spearman’s correlation r = 0.36, P < 0.01) and (D) WTc (Spearman’s correlation r = 0.78, P < 0.001) groups.

FIGURE 7

Brain activations detected in functional MRI (fMRI) two-choice task viewed in coronal slices, in the posterior and anterior insula (Post. and Ant. Insula), dorsal anterior cingulate cortex (dACC), hippocampus (HPC), fusiform gyrus (FG), occipital-fusiform complex (LOC), in similar liking (SL) and different liking (DL) conditions and for high-energy (HE) and low-energy (LE) food choices. The bottom left panel indicates the location of the coronal slices on a sagittal slice (+9). MNI coordinates are indicated for each coronal slice.