J. E. Love and K. M. Johnson
Biomimetic computation seeks to develop artificial intelligence by methodologies inspired by natural information processing mechanisms. Genetic algorithms simulate biological evolution by 'operating on genotypes, while evolution strategies and evolutionary programming emphasize phenotypes. However, the ontogenetic mapping of genotypic space to phenotypic space in biological systems is not determini’stic, because ontogenesis is influenced by epigenetic factors,, acting at the microenvironment and macroenvironment levels. Hence, phenotype results from both genetic and epigenetic operators. Although genomes and phenomes are fundamental aspects of evolutionary theory, ontogenetic evolutionism requires integration into the neoDarwinian synthesis, and epigenetic operators should be included in evolutionary computation. In this context we propose a novel perspective for simulated evolution: ontogenctic programming, in which ontogenetie algorithms involving epigenetic operators shape the adaptive trajectories of massively parallel cellular tensor multidimensional manifolds during somatic development. We outline its theoretical rationale and its modular application to neurocomputational biocognitronics models, artificial life, and evolvable neuromorphic hardware.