Work

Computer-based Impulsive Process Interventions in Dietary Behavior

Year: 2020
Type: dissertation
Abstract: Unhealthy foods are often appetizing and thus hard to resist, despite knowledge about their detrimental effects on health. The psychological processes underlying choices between immediately rewarding but health-harming and potentially less rewarding but health-promoting options have been described in dual-process models such as the Reflective-Impulsive Model. These models assume two interacting kinds of processes: one impulsive, association-based, fast, and effortless; the other reflective, based on syllogistic reasoning, slow, and dependent on cognitive resources. While most interventions on eating behavior aim to alter reflective reasoning and increase its influence on behavior, some relatively new computer-based interventions aim to directly change impulsive aspects of food intake. They do so by presenting users images of unhealthy food while asking them to show a behavior that is inconsistent with their impulse to approach, i.e. inhibit a response or show an avoidance reaction. This way, the association between unhealthy foods and approach behavior is broken up, reducing the strength of elicited impulses in the future. Evidence has mounted that particularly the Go/No-Go task can reduce intake of unhealthy foods. In this task, participants need to inhibit their impulse to react when seeing unhealthy food images, creating an “unhealthy-stop” association. Two important questions have remained unanswered in this research line: (1) What are the psychological mechanisms of behavioral effects in these interventions? (2) How well do laboratory-based demonstrations of efficacy translate to real-world behavior change? This thesis aims to produce evidence that helps answer those questions and enhance understanding of impulsive process interventions by (1) meta-analyzing the scientific literature with a specific focus on implicit bias change as a potential psychological mechanism (Study I), (2) examining a potential neural marker of the psychological mechanisms involved (Study II), and (3) investigating relevant aspects of such an intervention in the field (Study III). Study I thereby lays the foundation for the main research questions and study design in Studies II and III. The specific aims of Study I were to provide an overview of the literature on different kinds of computer-based impulsive process interventions by meta-analyzing their effects on eating behavior. In addition, it aimed to identify study-level moderators of effects, to investigate implicit bias change in the included studies as a potential mechanism of behavioral effects, and thus to identify ways forward for the research area. Study II followed up on the results of Study I and aimed to identify the role of a neural marker of the involved psychological processes, the N2 event-related potential (ERP), by measuring neural activity during the performance of an impulsive process intervention. Study III then used data from a field trial to investigate dose-response relationships in an impulsive process intervention in order to provide best-practice guidance for future intervention users. The studies in this thesis used a range of methods to answer the specific research questions. In Study I, I conducted a systematic literature review with meta-analysis. Meta-regressions with categorical and continuous predictors tested potential moderating variables and the relationship between behavioral effects and a hypothesized psychological mechanism of effects. In Study II, a specific neural marker, the N2 ERP , was analyzed during food Go/No-Go training and related to food intake both within and outside the laboratory. Study III used data from a pragmatic open trial in which participants used smartphone-based Go/No-Go training at their own discretion to determine dose-response relationships for intake of different food categories. Study I identified 30 randomized controlled trials with 47 effect sizes, delivering at least one of four computer-based impulsive process interventions, for inclusion in the analyses. The overall effect size for eating behavior was small (g = −0.17, CI95 = [−0.29; −0.05], p = .01) with only the Go/No-Go task showing significant effects (g = −0.39, CI95= [−0.57; −0.22], p < .001). The effects on implicit biases were also significantly different from zero (g = −0.46, CI95= [−0.68; −0.25], p < .001) and the effects on bias change and eating behavior were significantly related (B = .42, CI95= [.02; .81], z = 2.07, p = .04, k = 21). Against preregistered hypotheses, analyses in Study II showed no significant differences between healthy and unhealthy foods in N2 ERP strength (p = .18, ηp2 = 0.04), no effect of Go/No-Go training on immediate food intake (p = .50) or N2 amplitudes (p = .57, ηp2 = 0.0004), and no relationship between N2 and food intake both inside (B = -0.41, p = .82) and outside (B = 166.21, p = .53) the laboratory. In Study 3, analyses revealed significant relationships between the self-administered amount of conducted Go/No-Go training and self-reported food intake, such that conducting more blocks of training led to decreased unhealthy (β = -0.004, CI95 = [-0.006; -0.002]) and increased healthy food intake (b = 0.003, CI95 = [0; 0.005]). Separate analyses for different food types indicated variations between food categories in their dose-response relationship, indicating that completing about 8 to 24 4-minute sessions of Go/No-Go training over the course of one month leads to changes in intake by one point. The scientific contribution of this work is thus threefold. Firstly, it provides an overview of the research field, providing a clear recommendation for the precise task to use for intervention delivery, and indicates future directions for the field (Study I). Secondly, it contributes to the ongoing debate about psychological and neural mechanisms of impulsive process interventions in eating behavior and indicates changes in impulsive processes to be one driver of behavioral effects (Studies I and II). Lastly, it provides estimates for necessary dosage and usage recommendations based on real-world data from smartphone-delivered training (Study III), a promising future avenue for large-scale, self-administered training. Based on this thesis, future research should aim to further clarify psychological training mechanisms, establish ideal training design and training schedules, identify groups for whom training is most beneficial, and seek synergistic effects with other interventions targeting eating behavior.
Language: English
Field: Psychology
Open Access status: green
OpenAlex ID: W3092201778
Cites: 247
Cited by:
Related to: 10