Repetitive habitual behaviour can persist even when it conflicts with goals. This is termed an action slip. Wood et al. (2023) demonstrated this effect using a novel procedure, where participants’ performance was strong on congruent trials but comparatively poor on incongruent trials. The current study extends their work by exploring a devaluation version of the experiment, followed by a further experiment that manipulates training duration. The devaluation approach eliminated the congruency effect, while reinstating standard testing with varied training led to its reappearance—even after a short amount of training. These findings challenge the Stimulus-Response (S-R) account and dual-process theory. We discuss the question of whether the congruency effect seen in these experiments is evidence to support a dual-process account of habitual behaviour.