<p>Habits are stimulus-driven responses that are produced independently of the current outcome value. They enable efficient actions in familiar contexts while freeing up cognitive resources. Habits are expected to influence behavior after substantial experience, whereas limited training leaves behavior under goal-directed control. This transition from goal-directed to habitual control has been documented in animal research but remains challenging to replicate in humans. Using a free-operant task, recent studies have suggested that human habits can be fully learned even after short training. This would produce a ceiling effect, explaining null results when the amount of training is manipulated. Here, we propose an alternative explanation: the devaluation protocol was ineffective for a subset of participants who appeared “habitual”. To test this hypothesis, we conducted a preregistered conceptual replication of the free-operant task with improved devaluation. As in previous research, we found no difference in habitual responses between short and extended training. In contrast to prior reports, most participants were sensitive to outcome devaluation. Habitual responding was strongly correlated with the effectiveness of the outcome devaluation protocol; habit-like responses were mostly produced by the few participants for whom the devaluation did not work. We further supported our hypothesis by reanalyzing previous datasets, finding that, across all studies, participants who showed more habit-like responses were those for whom the devaluation was less effective. Taken together, our findings suggest that suboptimal outcome devaluation protocols may have biased previous results, leading to habit-like goal-directed responses regardless of the amount of previous instrumental training.</p>