In his Origin of Species, Charles Darwin summarizes his theory that in nature, living beings compete with each other for survival. Through natural selection, more fit individuals are more likely to inherit their genes. With the help of evolutionary game theory, we can study this process: is it enough to see/perceive what is needed for survival or do need more information to do so? Are fitness-based organisms better at surviving than those tuned to the objective truth? The nature of reality has been the subject of many scholarly debates and dialogues for millennia: the study of being (ontology) and the nature of knowledge (epistemology) have grown into philosophical disciplines. In his research on cognitive science, Donald Hoffman concluded that reality - independent of us - is far from what we think it is. Moreover, it is due to evolution itself that we have changed our perception of objective reality to the fitness-based truth. If they do not coincide, natural selection will favor living beings that are attuned to fitness and not to reality. The question whether evolution favours veridical perceptions was also investigated with the help of genetic algorithms (GAs). Building upon the problem first studied by Mitchell[1] and detailed in [2], the authors evolve with the help of a GA a perception-decision-action loop for an agent in a simple 2D 10x10 grid, surrounded by walls. The GA evolves a creature whose goal is to collect as many resources as possible within 200 moves. They find that besides it evolving effective foraging strategies, it also evolves interface perceptual strategies. We further analyze if the representation strategy used in the previous studies is scalable to a 3D grid and compare it to the performance of different control strategies obtained by neuroevolution and also compare its results with research involving human participants.
Keywords: neuroevolution, evolutionary game theory, fitness, perception, genetic algorithm
[1] https://melaniemitchell.me/ExplorationsContent/RobbyTheRobot/
[2] Melanie Mitchell. Complexity: A guided tour. Oxford University Press, 2009.