The purpose of this master project was to explore decision making process applied to a blackjack game and make the links with facets of impulsivity. The first part of this study goes through the mathematical of this game and presented the optimal policy, computed with reinforcement learning algorithm. We also computed with a Monte Carlo simulation, for each possible hand and for each possible decision, the 3 values ${Q}_{t}$, ${H}_{t}$ and ${h}_{t}$, value function, risk, one-step-ahead risk proposed by d'Acremont & Bossaerts, 2008. The second part of the study present the result of the behavioural study. Forty-seven undergraduate students perform 200 blackjack hands and completed a self-report questionnaire evaluating impulsivity and risk attitude. Our analysis reveals links with lack of perseverance and decision time. Furthermore percentage of optimal decision along the game reveals learning with novice player. The result of this study can be used as a basis for future research in neuroimaging on the risk, to try to answer if is possible to distinguish brain area associated with one-step ahead versus total risk