Experimental habit formation using smartphone applications: challenges in Inducing devaluation with extended training

Understanding habit formation in humans remains a fundamental challenge, partly due to the constraints of laboratory-based studies. Thus, this study investigated whether prolonged, real-world engagement with a novel smartphone-based task could induce habitual behavior. Across five pilot experiments (n=27-43; total N = 177), participants were randomly assigned to either a short (ST) or extended training (ET) condition using a gamified button-press task delivered via a custom web application. Habit formation was assessed through three complementary tests targeting hallmark features of habitual behavior: automaticity, inflexibility, and insensitivity to outcome devaluation. While the ET group showed improved performance on dual-task measures (suggestive of increased automaticity), neither the devaluation nor the inflexibility tests yielded conclusive evidence of habit induction across experiments. Bayesian analysis supported the absence of robust group differences in insensitivity to outcome devaluation. Our findings highlight the difficulty of inducing measurable habits in humans, even under prolonged and ecologically valid training. Crucially, they also call into question the utility of outcome devaluation as a definitive measure of habit, regardless of training duration or setting. We argue that this widely used paradigm may miss important aspects of stimulus–response learning and suggest the need for alternative or refined tools to capture the complexity of human habit formation. Our null results, provide methodological insights for future research and suggest that smartphone-based paradigms, while scalable and flexible, require further refinement to elicit and detect true habitual behavior. All code and data are openly shared to promote replication and advancement in habit research.