| dc.description.abstract | Alternative Reproductive Tactics (ARTs) are different reproductive strategies used by individuals of same sex to maximize the reproductive success, such as territorial defense, sneak mating, or cooperative breeding. In this study, the embodied evolution represents a real-life biologically inspired agent-based simulation where the agents move around freely in the simulated environment governed by the dynamic outputs from their neural network which has the inputs of the information of the environments such as food proximity, distance and orientation of other agents, their energy, age, and mating state which have weights which are the genome of the agents. The agents have no built-in genders but rather act as female or male when they choose the respected outputs, mate_as_female or mate_as_male respectively. Reproduction occurs probabilistically based on the energy thresholds and proximity with the female-role bearer agent taking the reproduction cost of energy deduction. 1000 generations are run, where agents evolve through crossover and mutation with probability 0.1. The simulation has no explicit fitness function rather the total reproductive success servers as the fitness. Natural selection plays a vital role to populate the fitted agents for the next generation, as randomly offspring are chosen. Our analysis showed few agents always acted as female, as male always whereas we had not hardcoded the genders initially and rest as hermaphrodites. The females and hermaphrodites even had two types of phenotypes that had evolved. | en_US |
