: Mon travail de thèse a porté sur l’étude neurocomportementale des actions et des habitudes chez le Rat. En effet, lors d’un apprentissage opérant la réponse peut être acquise selon deux systèmes associatifs. Le premier dépend d’une représentation des conséquences de la réponse, le second d’une association plus « simple » entre le stimulus et la réponse. Un premier axe de recherche a consisté à étudier plusieurs paramètres du conditionnement, afin de déterminer leur influence sur le contrôle de la réponse instrumentale par un système plutôt que l’autre. Le deuxième axe de recherche a porté sur l’étude des substrats neuronaux impliqués dans l’acquisition et l’expression d’une action, par l’intermédiaire de techniques d’inactivation cérébrale et d’étude immuno-histochimique de l’expression génique de la protéine Fos. / Previous research has established that instrumental conditioning, in both primates and rats, is mediated by two concurrent associative systems. In early stages of training, instrumental response is thought to be mediated by an association between the action and the outcome (A-O). While training proceeds however, as the response becomes less sensible to the outcome value, it is conceived as being mediated by an association between thestimulus and the response (S-R). Recent evidences suggest that the both systems operate in tandem and/or competition from the beginning of training. This work aimed at studying the mechanisms that coordinate the control of the instrumental response by the goal-directedsystem or the habit system. A first batch of results indicates no effect of the amount of training sessions on the goal-directed nature of the conditioned instrumental response. Indeed,the outcome devaluations by CTA or selective satiety reduced the instrumental performances,independently of the training procedure applied. The instrumental responses resulting from our 3 training procedures depend of an actualized representation of their outcomes. A secondbatch of results indicates that information about the context of instrumental conditioning isincluded in the incentive representation of the outcome. Indeed, we observed no sensitivity tooutcome devaluation when devaluation occurred outside the training context. These results offer new original hypotheses about context encoding and the nature of instrumental responding. A third batch of experiments investigates the role of the prelimbic cortex in acquisition vs. expression of goal-directed instrumental behaviour, using reversible neuronal inactivation. The results show that the prelimbic cortex plays a transient but crucial role in theacquisition of goal-directed responding and that the A-O and S-R systems can operate in a competitive fashion early in training. Using ex-vivo imaging, a last batch of experiments aimed to study the temporal cerebral activation throughout instrumental training with a focuson prefrontal and striatal regions. Results show levels of Fos expression that vary with regions. At the beginning of conditioning, the density of Fos positive nuclei is high in the prefrontal regions. It decreases with training. Labelling is denser in the dorsomedial striatumthan in the dorsolateral striatum. The weak activation in the dorsolateral striatum appears consistent with the absence of habit. These data are in accordance with data of the literature concerning dynamics of activation in cortico-striatal circuits. Furthermore, they are in agreement with the suggestion that activity in the prelimbic cortex could promote the acquisition of goal-directed action by the induction of neuronal plasticity in the dorsal striatum.