Developing and testing new variants of cognitive bias modification for alcohol and tobacco use disorders

This dissertation examines whether, when, and for whom cognitive bias modification (CBM) can support recovery from alcohol and tobacco use disorders, two conditions marked by persistently high relapse rates despite available treatments. CBM is a family of brief computerized training programs designed to retrain the automatic mental tendencies (e.g., to approach substance-related cues) that may undermine quit attempts. Two Bayesian individual participant data (IPD) meta-analyses synthesized evidence across 23 alcohol trials (N = 8,297) and 19 smoking trials (N = 2,727). For alcohol, CBM produced small but meaningful reductions in cognitive bias and relapse risk, but mainly when delivered face-to-face alongside structured psychological treatment. For smoking, there was moderate-to-strong evidence that CBM did not help participants quit or cut down. Two randomized controlled trials then tested ABC-training, a new variant of CBM that links personal Antecedents, alternative Behaviors, and meaningful Consequences. Within the Dutch IkPas abstinence challenge, ABC-training did not outperform conventional CBM or placebo on pre-registered drinking outcomes, but consistently increased the likelihood of completing the abstinence challenge. A parallel trial within an online-delivered smoking cessation program showed no benefit of either training. Taken together, CBM is best understood as a context-dependent adjunct rather than a stand-alone intervention. The dissertation closes with a framework integrating ecological momentary assessment and person-specific modeling to guide a more personalized, context-sensitive next generation of CBM research.