Knowledge or reactivity: where is the right trade-off for an artificial mind?

As opposed to the many robotic systems extensively built exclusively around the knowledge paradigm, examples taken from nature show a considerable level of adaptability to their environment without requiring a complete knowledge of their world. Extrapolating from such examples, the behavioural paradigm maintains that, since biological instincts are not processed at an intentional level, there is no need for a global knowledge. Even though the purely behavioural paradigm provides a possible answer to the issue of real-time reactivity, it is clearly not enough to allow an agent to scale up to an intentional level. Thus, although reactivity should be an integral part of an “intelligent robot system”, knowledge should also play an important role in it. Defining the right trade-off between the two of them is the topic of this paper