As training progresses, animals show a transition from goal-dependent behavior to goal-independent behavior (habitual responses). Habit formation is influenced by several factors, including the amount of training and action-outcome contingency. However, it remains unknown whether and how discretion for behavioral selection influences habit formation. To this end, we trained male rats in two types of two-alternative forced-choice task: visual association and nonvisual association tasks. In the first type of task, rats learned the association between reward and a visual cue, the position of which was randomly changed per trial so that rats had to make a judgmental decision about which choice delivered the reward in each trial (discreet judgment group); in the second type of task, the rats learned that a reward was delivered after either choice following task initiation (uncontrolled judgment group). To test the sensitivity to contingency manipulation, the extinction tests were conducted in short- and long-term trained groups, with the result that the overtrained rats in the uncontrolled judgment group, but not the other three groups, showed less sensitivity. To further investigate the reward sensitivity in the long-term trained groups from another perspective, we continuously and periodically altered the reward size for each trial. The rats of the discreet judgment group changed intertrial intervals depending on reward size, while this tendency was weaker in the uncontrolled judgment group. These results suggest that discreet judgment maintained goal-directed rat behavior, whereas uncontrolled judgment led to the development of habit-like behavior.