<p>Habits are central to daily life but remain difficult to study in the lab. Outcome Devaluation is often considered a “gold standard” paradigm for habit measurement; yet human studies have produced mixed evidence: many fail to show the overtraining effects on devaluation sensitivity expected from a robust habit index. An alternative measure involves remapping stimuli to new responses rather than devaluing outcomes. Crucially, this procedure limits preparation time to reveal habits that may otherwise be masked by goal-directed processes. Although promising, it remains unclear whether these tasks assess the same behavioural phenomenon. To address this, we compared Outcome Devaluation and Response Remapping under matched preparation-time constraints after 6 days of training. Response Remapping showed the expected pattern: expression of the pretrained (putatively “habitual”) response was greatest at 200–500 ms and weakened as preparation time increased. In Outcome Devaluation, raw responses to devalued outcomes increased with preparation time, driven by a time-dependent withholding bias. Once controlled for, devalued responses decreased with preparation time. Crucially, this effect was larger in Remapping. Moreover, task comprehension problems may limit interpretation of the Devaluation task. These findings suggest that paradigms engage partially distinct processes—response inhibition versus automatic action selection—with one requiring inhibition of a learned response and the other relying on fast, automatic actions. Our findings reveal that behaviour on these tasks is sensitive to task structure, underscore the need to consider distinct cognitive and motivational factors driving action selection, and highlight that what researchers call “habits” is influenced by multiple processes.</p>