Sex with no regrets: How sexual reproduction uses a no regret learning algorithm for evolutionary advantage

Area of Science:

• Evolutionary biology
• Theoretical biology
• Genetics

Background:

• The evolutionary persistence of sexual reproduction remains a significant question in biology.
• The apparent disadvantage of recombination, which can produce less fit offspring, challenges traditional models of fitness optimization.
• Existing theories struggle to fully explain the prevalence of sex across diverse taxa.

Purpose of the Study:

• To investigate the evolutionary advantage of sexual reproduction using a novel theoretical framework.
• To explore the role of recombination in navigating complex genotypic landscapes.
• To provide a computational explanation for the 'why sex' problem.

Main Methods:

• Modeling evolution as a machine learning algorithm, specifically a no-regret algorithm.
• Simulating sexual and asexual populations within a vast genotype space.
• Comparing the efficiency of exploration and fitness gains between sexual and asexual reproduction strategies.

Main Results:

• Both sexual and asexual reproduction utilize no-regret learning algorithms for adaptation.
• Sexual reproduction, through recombination, performs a more efficient goal-directed search in genotype space.
• Sexual populations consistently achieve higher fitness than asexual populations, even in stable environments, by overcoming fitness plateaus.

Conclusions:

• Recombination in sexual reproduction is not a random hindrance but a critical component of an effective 'sampling for learning' algorithm.
• Evolutionary theory can be advanced by viewing populations as learning algorithms adapting to their environments.
• This framework offers a compelling explanation for the maintenance of sexual reproduction in nature.