James Ashford, Berfin Sakallioglu, Marta Tataryn, Alessandro Valerio, Benedikt Hartl, Léo Pio-Lopez, Marko Cvjetko, Richard Löffler, Stefano Nichele
Preprint, Evolving self-organisation Workshop (EvoSelf@GECCO), 2026
Symbiogenesis—the formation of a new entity through the persistent association of previously independent replicators—was proposed by Barricelli in 1954 as the missing ingredient in any theory of the origin of life. While evolutionary algorithms encode mutation, inheritance, and selection, symbiogenesis as an emergent process remains largely absent from modern evolutionary computation (EC) and artificial life. This paper revisits Barricelli’s framework of numerical organisms evolving on a 1D cellular automaton (CA), replicates his core findings with information-theoretic metrics, and extends them in two directions: a 2D generalisation of his automaton, and a set of minimal DNA-inspired interaction norms. Our results confirm that symbiogenetic interaction, not replication or mutation alone, both kickstarts the emergence of self-replicating structures and sustains open-ended evolutionary dynamics. The 2D extension and Boolean CA experiments demonstrate that these properties are substrate-independent. The DNA-norm experiments show that complementary association and strand separation produce heritable spatial structure from a minimal four-symbol alphabet. Together, these results suggest that symbiogenesis is a general and underexplored mechanism for open-ended evolution, with implications for EC, artificial life, and collective intelligence.