The capacity for autonomous behaviour is key to human intelligence, and fundamental to modern social life. However, experimental investigations of the cognitive bases of human autonomy are challenging, because experimental paradigms typically constrain behaviour using controlled contexts, and elicit behaviour by external triggers. In contrast, the sources of human autonomy and freedom are assumed to be endogenous. Here we propose a new theoretical construct of adaptive autonomy, meaning the capacity to make behavioural choices that are free from constraints of both immediate triggers and habitual responding. Participants played a competitive game in which they had to choose the right time to act, in the face of an opponent who punished (in separate blocks) either choice biases, habitual sequences of action timing across trials, or habitual responses to the effects of reinforcement. Adaptive autonomy with respect to each habit was measured by the ability to maintain performance against the opponent even when the corresponding habit was punished. We found that participants were able, under pressure from their opponent, to become free from habitual choices of when to act, but were not able to free themselves from win-stay, lose-shift patterns of reinforcement, even when these resulted in punishment. These results propose a new testing ground of autonomous behaviour as a flexible adaptation of more or less habitual behaviours that co-exist with different classes of external constraint.