Abstract:

An Artificial Chemistry (AChem) is a set of components and interactions that result in a composable system. Ideally, the system is rich, and results in rich higher-order emergent properties. We present a methodology for discovering interesting AChems through a series of tests that probe elementary lowlevel properties. In doing so, we assume that these elementary properties are a necessary, but not sufficient, basis for higher-order emergent properties, such as autocatalytic sets and hypercycles. The test strategy is applied to RBN-World, a sub-symbolic chemistry. This results in identifying a number of new and interesting RBN-World chemistries that appear richer than our original parameterisation.