Phylogeny Recapitulates Learning: Self-Optimization of Genetic Code

Learning algorithms have been proposed as a non-selective mechanism capable of creating complex adaptive systems in life. Evolutionary learning however has not been demonstrated to be a plausible cause for the origin of a specific molecular system. Here we show that genetic codes as optimal as the Standard Genetic Code (SGC) emerge readily by following a molecular analog of the Hebb’s rule (“neurons fire together, wire together”). Specifically, error-minimizing genetic codes are obtained by maximizing the number of physio-chemically similar amino acids assigned to evolutionarily similar codons. Formulating genetic code as a Traveling Salesman Problem (TSP) with amino acids as “cities” and codons as “tour positions” and implemented with a Hopfield neural network, the unsupervised learning algorithm efficiently finds an abundance of genetic codes that are more error-minimizing than SGC. Drawing evidence from molecular phylogenies of contemporary tRNAs and aminoacyl-tRNA synthetases, we show that co-diversification between gene sequences and gene functions, which cumulatively captures functional differences with sequence differences and creates a genomic “memory” of the living environment, provides the biological basis for the Hebbian learning algorithm. Like the Hebb’s rule, the locally acting phylogenetic learning rule, which may simply be stated as increasing phylogenetic divergence for increasing functional difference, could lead to complex and robust life systems. Natural selection, while essential for maintaining gene function, is not necessary to act at system levels. For molecular systems that are self-organizing through phylogenetic learning, the TSP model and its Hopfield network solution offer a promising framework for simulating emerging behavior, forecasting evolutionary trajectories, and designing optimal synthetic systems.

Reflections on what timescale?

Recent developments in both evolutionary theory and in our ideas about development suggest that genetic assimilation of environmental regularities may occur on shorter timescales than those considered by Shepard. The nervous system is more plastic and for longer periods than previously thought. Hence, the internal basis of cognitive-perceptual skills is likely to blend ontogenetic and phylogenetic learning. This blend is made more rich and interactive by the special cultural scaffolding that surrounds human development. This being so, the regularities of the environment which have been genetically assimilated during the emergence of modern human beings may themselves be the products of human action. [Shepard]