Nudging with Attribution Bias: Promoting Healthy Over Unhealthy Drink Preferences

The intervention aims to exogenously manipulate subjects' thirst level prior to consumption of a target carbonated drink, with the aim to vary consumption experience across treatments. Participants will be asked to refrain from drinking water or any fluids for five hours before the experiment. They will then be randomly assigned to either the treatment or control group using computer-based randomization software. The treatment group will remain thirsty, while the control group will receive a bottle of water to offset their thirst. Following thirst manipulation, both groups will be provided with a healthy carbonated drink (e.g., Poppi drink) to consume, after which they will rate their consumption experience. The treatment group is expected to report higher ratings since they consume the drink under an arousal state (thirst), resulting in a better consumption experience than subjects in the control group, who had their thirst offset before consuming the healthy drink.

2026-04-01

2026-07-01

1. Subjects' choices between the healthy and equivalent, unhealthy carbonated drink in the choice task will be compared between the control and treatment.
2. Event-Related Potentials (ERPs) will be measured during the cue exposure task, and key components (e.g., P1, N1, N2, P3, LPP, RewP) will be compared between control and treatment.

Age, gender, ethnicity, income, education, exercise frequency and intensity, duration since last meal, current appetite, number of meals and snacks per day, expected time until next meal, consumption experience.

The experiment is conducted across two days. On the first day, subjects will consume a healthy carbonated drink under a control or treatment condition. On day two, they will make choices between the healthy drink and an equivalent, unhealthy alternative in a choice task. Neurophysiological indicators of preferences for the healthy and unhealthy drinks will also be collected during a cue exposure task.

The experiment is conducted across two days. The first day involves the treatment manipulation, where the consumption experience with a healthy carbonated drink will be exogenously varied between a control and treatment group. The manipulation will be thirst-based, where all subjects will be asked to refrain from drinking water or any fluids for 5 hours prior to arriving at their first-day session. Subjects in the control group will be provided with a bottle of water before starting to offset their thirst, while those in the treatment group will not. Subjects will then report their thirst levels, which will be used to confirm successful thirst manipulation, after which they will be asked to consume the healthy carbonated drink. During consumption of the healthy drink, all subjects will first consume one portion, after which they will answer a very short intermediate questionnaire before consuming the second portion. The intermediate questionnaire will consist of two questions about how enjoyable the product is and how excited they are to consume the second portion. After consuming both portions of the healthy drink, subjects will answer a short questionnaire to collect sociodemographic data, as well as behavioral information related to their beverage consumption, including how many glasses of water they typically have per day, how many glasses of liquids other than water they have per day, and how often they consume regular and enhanced carbonated drinks. The first day’s session will then conclude with measuring subjects’ weight and height to calculate BMI.

The second day’s session will start with a short questionnaire asking subjects to report how much they enjoyed consuming the healthy drink on day 1 and if they would consume the drink again if it were available to them. They will also report if they had consumed the healthy drink/unhealthy equivalent (e.g., Poppi drink and Sprite) between days 1 and 2. Subjects will then complete two tasks in a randomized order to elicit behavioral and psychophysiological indicators of their preferences for the healthy drink consumed on day 1 compared to an equivalent unhealthy alternative. The first is a choice task where subjects will be presented with an image of the healthy and unhealthy drinks and asked to make a choice between them in an incentive-compatible environment (i.e., real choices). The second is a cue exposure task, where subjects will be presented with a random sequence of 90 images, consisting of 30 non-food images (e.g., office items), 30 images of the healthy drink they consumed on day 1, and 30 images of the equivalent unhealthy alternative drink. The images will be presented in a quasi-random order that avoids more than 4 repetitions from the same category.

Participants will be assigned equally to Control and Treatment groups (60 in each). First, each of the 120 participants is labeled with a unique identifier. A random number is then generated for each ID from a uniform distribution. The numbers are ranked from lowest to highest, and the 60 IDs associated with the lowest numbers are placed in the Control group, while the remaining 60 are placed in the Treatment group. This ensures an exact 50–50 split while preserving random assignment.

Individual

No

Not clustered

120

60

Choice outcome minimum detectable effect size: 0.230 (m1=0.568, sd1=0.502, sd2=0.381)