Implicit process interventions in eating behaviour: A meta-analysis examining mediators and moderators.

Dual-process models integrate deliberative and impulsive mental systems and predict dietary behaviours better than deliberative processes alone. Computerized tasks such as the Go/No-Go, Stop-Signal, Approach-Avoidance, and Evaluative Conditioning have been used as interventions to directly alter implicit biases. This meta-analysis examines the effects of these tasks on dietary behaviours, explores potential moderators of effectiveness, and examines implicit bias change as a proposed mechanism. Thirty randomized controlled trials testing implicit bias interventions (47 comparisons) were included in a random-effects meta-analysis, which indicated small cumulative effects on eating-related behavioural outcomes (g = -0.17, CI95 = [-0.29; -0.05], p = 0.01) and implicit biases (g = -0.18, CI95 = [-0.34; -0.02], p = 0.02). Task type moderated these effects, with Go/No-Go tasks producing larger effects than other tasks. Effects of interventions on implicit biases were positively related to effects on eating behaviour (B = 0.42, CI95 = [0.02; 0.81], p = 0.03). Go/No-Go tasks seem to have most potential for altering dietary behaviours through implicit processes. While changes in implicit biases seem related to the effects of these interventions on dietary outcomes, more research should explore whether repeated exposure to implicit bias interventions may have any practical intervention value in real world settings.