We aim to show in this paper that a cortically-inspired control model applied to an autonomous robot can provide strong capabilities. First, such an approach enables a robot to extract spatio-temporal regularities during its ``life'' in an unknown environment, these regularities being stored in a more expressive way than the Q-values. Second, a compromise between perception and drives emerges from the distributed mechanism of spreading activities through the cortical net. Endly, the multi-modal management of information flows endows the robot with capabilities to generalise sequences of perceptions and use this knowledge to transfer learning from those learnt sequences to further behaviour.\\ Although our experiments with a simulated robot clearly exhibit these properties, they reveal a lack of planning capabilities that encourage us to extend our work toward a model of prefrontal cortex, grounded on the current results.