A Memetic Approach to the Evolution of Rhythms in a Society of Software Agents Marcelo Gimenes, Eduardo Reck Miranda, Chris Johnson Computer Music Research, School of Computing, Communications and Electronics University of Plymouth, UK {marcelo.gimenes, eduardo.miranda, c.johnson}@plymouth.ac.uk Abstract. We are developing RGeme (Rhythmic Meme Generator), an artificial intelligence system for the composition of rhythmic streams inspired by Richard Dawkin's theory of memes. The system is based on intelligent agents that learn from examples and interact by generating rhythms. The system has two broad stages. In the first one, the learning stage, which is the main focus of this paper, Agents are trained with examples of musical pieces in order to evolve a “musical worldview” which consists of a “Style Matrix” of basic rhythmic elements (or "rhythmic memes”). In the next (production) stage Agents are able to learn from each other's "compositions" and capable of evolving new rhythmic styles by adapting to each other's rhythms. 1. Introduction Computers have long being used for aiding musical composition in a number of possible ways. Some composers ([Cope 1991], [Dodge 1985], [Worral 2001] and [Xenakis 1971]) use mathematical models such as combinatorial systems, grammars, probabilities and fractals to create new pieces of music. Other systems apply standard Genetic Algorithm procedures for evolving musical materials such as melodies, rhythms, chords, and so on. One such example is Vox Populi [Manzollli 2000], which evolves populations of chords of four notes through the operations of crossover and mutation. Evolutionary Computation models are also being used in many models. In one of them, CAMUS [Miranda 1993], the emergent behaviour of Cellular Automata (CA) is used to generate musical compositions in which case the co-ordinates of the cells are associated with the distances between the notes of a set of three musical notes. Impett (2001) uses an Agent system to generate musical compositions. Through the interaction of embodied behaviours that co-exist and interact in the same world, Agents are adaptive to the changing environment to which they belong. A growing number of researchers are developing computer models to study cultural evolution, including musical evolution [Blackmore 1999]. For instance, Miranda (1999) investigates how musical structures can originate and evolve in artificially created environments and inhabited by virtual communities of musicians and listeners. Some rhythmic generating systems have already been proposed [Horowitz 1994]. Pachet (2000) describes an evolutionary model where a group of agents play rhythms together in real time without prior knowledge about the music to play. Agents