<p>Most food preferences and aversions are acquired through experience. These learned behaviors arise from the positive or negative experiences that occur during (simultaneous taste–taste learning) or after (taste–consequence learning) consumption. Just as ingesting a food that causes nausea leads to a conditioned aversion, ingesting a highly hedonic or caloric food increases its preference. Appetitive preference learning has been extensively studied using the flavor preference conditioning paradigm. The latter involves presenting a preferred taste (i.e., sucrose; Unconditioned Stimulus [US]) with an initially neutral flavor cue (i.e., cinnamon; Conditioned Stimulus [CS+]), ultimately resulting in a conditioned preference for the CS+. Previous studies have shown how conditioned preferences are weakened after devaluing the US, suggesting that such preferences are based on flexible, goal-directed behavior guided by the activation of the US representation (Stimulus–Stimulus learning). The present study investigates whether the type of training (extensive or short) determines the learned content. Specifically, we explore whether initial learning involves flexible, goal-directed behavior that shifts to automatic, habitual behavior after extended training. Our findings reveal that while the minimally trained group showed a devaluation effect, the group given extended training continued responding even after the devaluation treatment. These findings are discussed in the context of modern obesogenic environments and in terms of their theoretical significance, given the limited evidence of habitual behavior in Pavlovian learning.</p>