Simple Evolving Connectionist Systems (SECoS)

During the introduction (cf. 2.1) it has been pointed out that biological systems can not only learn while preserving the structure of the brain and the body but they can also learn while they are changing the structure by adding and removing neurons and connections. To model this property of real brains within a formal theory one needs an appropriate formalism to enable this. Today several options are available to realize such an idea. In this chapter we will present the Simple Evolving Connectionist Systems (SECoS) introduced by Watts 2004 in his PhD-Thesis [131]:67ff. We will use the latest version of SECoSs as presented in Watts 2009 [129]:256ff.