Food Folio by Columbia Center for Eating Disorders: A Freely Available Food Image Database

Abstract

Food images are useful stimuli for the study of cognitive processes as well as eating behavior. To enhance rigor and reproducibility in task-based research, it is advantageous to have stimulus sets that are publicly available and well characterized. Food Folio by Columbia Center for Eating Disorders is a publicly available set of 138 images of Western food items. The set was developed for the study of eating disorders, particularly for use in tasks that capture eating behavior characteristic of these illnesses. It contains foods that are typically eaten, as well as those typically avoided, by individuals with eating disorders. Each image has now been rated across 17 different attributes by a large general United States population sample via Amazon’s Mechanical Turk (n = 1054). Ratings included subjective attributes (e.g., tastiness, healthiness, and favorable texture) as well as estimates of nutrient content (e.g., fat and carbohydrate). Each participant rated a subset of stimulus set food items (46 foods) on all 17 dimensions. Additional description of the image set is provided in terms of physical image information and accurate nutritional information. Correlations between subjective ratings were calculated and an exploratory factor analysis and exploratory cluster analysis completed. Outcomes of the factor analysis suggested foods may be described along three latent factors of healthiness, tastiness, and umami taste; the cluster analysis highlighted five distinct clusters of foods varying on these same dimensions. Descriptive outcomes indicated that the stimulus set includes a range of foods that vary along multiple dimensions and thus is likely to be useful in addressing various research questions surrounding eating behavior and cognition in healthy populations, as well as in those with eating disorders. The provision of comprehensive descriptive information allows for stimulus selection that is optimized for a given research question and promotes strong inference.

Neural Representations of Food-Related Attributes in the Human Orbitofrontal Cortex During Choice Deliberation in Anorexia Nervosa

“…The distribution of taste ratings among HC was skewed relative to AN, resulting in high/low-taste food labels that did not capture the well-characterized influence of taste on choices among HC that was also previously observed in this sample (Foerde et al, 2015). This did not appear to alter the decoding results presented here, but future studies that seek to employ classification methods could specifically select stimulus sets to address such concerns (Lloyd and Foerde, 2020; Lloyd et al, 2020; Schebendach et al, 2020). It should be noted that the magnitude of decoding accuracies should be interpreted with caution because these measures can be influenced by ROI size, degree of voxel smoothing, and the size of training and test data sets, among other factors (Haynes, 2015).…”

“…This reduced sensitivity to tastiness may explain why classifier evidence of taste among HC did not predict the magnitude of their choice preferences. Future studies that employ food stimulus sets with more normally distributed taste ratings among participants may have more success in characterizing the contribution of OFC representations of tastiness to choices in normative decision-making (Lloyd et al, 2020).…”

“…Studies of food choice that have assessed decision-making as a function of tastiness and healthiness (Hare et al, 2009, 2011; Foerde et al, 2015; Maier et al, 2015) generally assume that weighted sums of these attributes are computed to guide decisions. Here, the taste and health rating tasks were used to obtain subjective ratings of these attributes (Hare et al, 2009, 2011; Sullivan et al, 2015; Lloyd et al, 2020; Maier et al, 2020). Although participants were instructed in the taste rating task to only evaluate foods on taste, assessment of how good or bad different foods taste could potentially be thought of as similar to value assignment, during which features other than palatability are considered (Suzuki et al, 2017).…”

“…In each task, the food items were presented on white plates against a black background in high-resolution color photographs. These stimuli are included in the Food Folio by Columbia Center for Eating Disorders stimulus set (Lloyd and Foerde, 2020; Lloyd et al, 2020; Schebendach et al, 2020). The order of stimulus presentation was randomized in each task.…”

“…Previous studies have shown that the orbitofrontal cortex (OFC) plays a central role in representing the subjective value of individual foods and food choice (Padoa-Schioppa and Assad, 2006; Plassmann et al, 2007; Clithero and Rangel, 2014; Suzuki et al, 2017; Ballesta et al, 2020). Other studies have demonstrated that evaluations of tastiness and healthiness—two food attributes that tend to be unrelated among healthy individuals—interact to determine food choices (Hare et al, 2009, 2011; Maier et al, 2015; Lloyd et al, 2020).…”