Can People Accurately Estimate the Calories in Food Images? An Optimised Set of Low- and High- Calorie Images from the food-pics database. (1st August 2019)

• Record Type: Journal Article
• Authors: Horne, Dielle; Palermo, Romina; Neumann, Markus F.; Housley, Regan; Bell, Jason
• Is Part Of: Appetite. Volume 139(2019)
• Page Start: 189
• Page End: 196
• Publication Date: 2019-08-01
• Subjects: Food habits -- Periodicals; Appetite -- Periodicals; Appetite disorders -- Periodicals; Electronic journals; 306.4613
• DOI: 10.1016/j.appet.2019.04.017
• Languages: English
• ISSNs: 0195-6663

Abstract

Abstract: Calorie intake plays an important role in maintaining a healthy weight. As such, researchers often use the calorie content of food as a distinction when investigating appetite related brain processes and eating behaviour. This distinction assumes that observers accurately perceive caloric content. However, there is evidence suggesting this is not always the case. The current study examined how accurately observers could estimate the caloric content of food images from the widely used "Food-pics" database. Eight hundred and forty psychology undergraduate students (aged 16–60, 64% female) estimated the caloric value of 178 high and 182 low calorie foods. Calorie content of food from both categories was significantly overestimated. Additionally, 7.7% of low calorie images were misperceived as being high calorie images and 35% of high calorie images were misperceived as being low calorie foods. Neither participants' gender, nor the recognisability and likability of the food images, influenced calorie estimation. Our findings show that most people are unable to accurately estimate caloric content of most food. Despite this, a selection of food images were judged accurately, and we advocate the use of these in research where it is important to have low- and high-calorie food images. Specifically, we propose an optimised stimulus set of 25 high and 25 low calorie food images that are accurately judged by adult participants. In addition, we provide the open source dataset of our ratings of Food-pics images which, when added to the existing Food-pics attributes, creates an enhanced tool for researchers selecting food stimuli.