Computational protocol: How Does Food Taste in Anorexia and Bulimia Nervosa? A Protocol for a Quasi-Experimental, Cross-Sectional Design to Investigate Taste Aversion or Increased Hedonic Valence of Food in Eating Disorders

Similar protocols

Protocol publication

[…] The procedure is summarized in Figure . The participant recruitment process starts with a recruitment letter to inform female patients between 18 and 35 years of age, starting or finishing with treatment, about our study at the collaborating clinics or with advertising flyers in public places and at the University in the case of healthy control participants. If the patient agrees, and after obtaining informed consent and the authorization to release/exchange confidential information, they will be contacted by phone to set three appointments in weekly intervals. During the first week, a diagnostic interview will be carried out by telephone. Eligible participants will be required to fill out a set of online questionnaires (LimeSurvey;, including the General Health Questionnaire, EDE-Q, short form of the Beck Depression Inventory-II (BDI-II), Food Cravings Questionnaire—Trait (FCQ-T-reduced) and Thought-Shape Fusion Questionnaire (TSF-Trait) (see Table ) during the second week. Finally, the experimental session in the clinics or in the psychophysiological laboratory at the universities of Fribourg and Luxembourg are expected in the third week. In the case of healthy controls, the procedure will be the same except for the diagnostic interview.The experimental session will be composed of assessments of thirst and hunger levels, the Thought-Shape Fusion Body State Scale (TSF-B state) and Brief Mood Scale (BMS) questionnaire, a training session to habituate to the testing environment and apparatus, and either the discrimination of two-alternative forced-choice (2AFC) test in study 1 or multiple-sip temporal-liking (MSTL) with taste reactivity (TR) task in study 2. Each experimental session will take around 1.5 h. Concerning the first study (Table ), the 2AFC test with signal detection theory (SDT) analysis will be used as it provides a high level of power for small food differences and low levels of response bias (Hautus et al., ). On the one hand, the 2AFC test is a psychophysical method in which participants are required to select the one of two food samples, which represents the specified attribute best (e.g., the sweeter one). On the other hand, SDT is a sophisticated tool that permits to understand the information processing that takes place in the brain during sensory discrimination testing, providing an accurate index of perceived sensory differences and efficient characterization of cognitive strategy. Therefore, SDT is critical for investigating maladaptive cognitions involved in the processing of received taste information, which affects sensory testing (Smeets et al., ). Furthermore, whether and how much gustatory performance depends on low BMI or psychopathology will be examined including under- and normal-weight controls and patients at different stages of the disorder (currently-ill vs. recovered). Finally, in addition to the sip-and-spit condition, a swallow test condition will be included (Eiber et al., ) to enhance cognitive bias toward weigh gain underlying the fear of swallowing and quantify the impact of these biased cognitions on gustatory performance. The 2AFC task will be conducted using the SIPM™ software (Sussex Ingestion Pattern Monitoring, version 2.0.11), in which pairs of two samples (with varying levels of sweetness and fatness), presented simultaneously at each trial, are judged in terms of taste identity and intensity (see Table ).Concerning the second study (Table ), a full-scale hedonic taste methodology using MSTL with TR will be applied to provide self-report, facial expression, facial electromyographic and autonomic measures to investigate (1) whether restrictive-type AN individuals respond with taste aversion for high-calorie foods and (2) if individuals with BN respond with excessive pleasure to sweet tastes. The MSTL and TR methodologies have been successfully employed previously with healthy populations (Rocha-Parra et al., ), showing both to be highly sensitive to detect small emotional variations when hedonic reactions are monitored dynamically via cumulative measurements, multiples sips and measurements at different durations. Consequently, this approach will allow for addressing two important unresolved issues: the temporal aspects of the hedonic response while eating (Delarue and Blumenthal, ), which have received no attention in EDs despite clinically observable aberrant emotional processing, and the scarce data on taste reactivity in EDs, which is considered the gold standard measure of hedonic aspect of human gustation without being limited by cognitive appraisal confounders of self-reports. MSTL, in which participants report subjective hedonic evaluations with SensoMaker™ software (version 1.8; Pinheiro et al., ) during 60 s while drinking three sips of each sample, taking one sip every 20 s, and TR measures (facial expression, facial electromyographic, electrodermal and heart rate reactivity) (Table ) will be recorded while tasting food samples. The video files will be run through the FACET™ SDK (iMotions Inc., Cambridge Innovation Center, US), which is an automatic facial expression recognition software that tracks and analyzes frame-by-frame (1/25 s) valence, action units and emotions. Finally, the motivational component (i.e., desire to eat) of hedonic reward of each food sample will be also assessed at the end of the 60-s period, asking “How much do you want to eat this food?” Participants' responses will be monitored using a 20 cm unstructured line scale anchored at both extremes 0–100 on the monitor (0 = “not at all”; 100 = “extremely”). Finally, in order to even increase patients' safety and well-being the standardized procedure in clinics will be enlarged by a short muscle relaxation training session of 15 min offered to patients at the end of the experimental session for study 1/study 2 to reduce any anxiety raised by tasting chocolate ice cream samples. [...] Sample size calculations were based on findings of small to medium effects on eating behavior in patients with anorexia and bulimia by our group (e.g., Munsch, ) using the software G*Power (version 3.13). Assuming two-sided tests with Alpha = 0.05, Beta = 0.2, and effect size f = 0.20, the required sample size would be 80 to ensure a power >0.80 in each one of the two studies with a sample distribution of NC−AN = 15, NC−BN = 15, NR−AN = 10, NR−BN = 10, NU−CT = 15, and NN−CT = 15. To obtain the required number of 50 patients per study and taking into account a participation rate of 80% and a dropout rate of 20% in the patient population, a total of 60 patients per study will be needed. Based on our recruitment and testing experiences in Switzerland and Luxembourg (average annual recruitment of 35–40 AN and 35–40 BN patients), the access to a sufficiently powered sample of patients seems to be guaranteed. Finally, in order to achieve the sample size suggested by the power analysis, a sample of 30 healthy controls (university students) per study will be recruited in Fribourg. […]

Pipeline specifications

Software tools SDT, MUSCLE, G*Power
Applications Miscellaneous, Phylogenetics, Nucleotide sequence alignment
Organisms Homo sapiens