The global parameters to a large degree determine the effectiveness and characteristics of the action selection behavior. It is still an open question how the values for these parameters should be selected. They are very problem dependent, not only because every problem area requires different degrees of goal-orientedness, situation-orientedness, speed, adaptivity, etc. But also because the size and structure of the network also determines these characteristics. For example, in an application with a very big network, the threshold has to be put higher to obtain the same results. At the moment we tune the parameters by hand going through a series of experiments. We plan to build a second network of competence modules that would look at the results of the first one and tune its parameters so as to obtain the action selection characteristics specified by the user.