Contemporary analyses of neurophysiological mechanisms of associative learning suggest that instrumental behavior can be controlled by separable action and habit processes. An increasingly broad range of human psychiatric and neurological disorders are now associated with maladaptive habit formation. The question of how the brain controls transitions into habit is thus relevant. Widely used training procedures that might differentially generate goal-directed actions or habits are variable schedules of reinforcement. Random interval schedules are known to generate habitual behavior compared with random ratio schedules Here, we report attempt to identify the behavioral characteristics of the bifurcation point of habitual and goal-directed behavior. We compared the time courses of learning in random ratio and random interval schedules with more common for neurophysiological researches parameters. Behavioral differences between schedules emerge early in learning. However, in outcome devaluation test we found that training in the random ratio schedule, but not in the random interval schedule, led to results interpreted as habitual behavior. This result is the opposite of what we expected based on previous research. We assume that the most commonly used variable schedules of reinforcement cannot serve as a reliable tool for analyzing neural mechanisms of habitual and goal-directed behavior.