The paper presents a cooperative coevolutionary approach to function optimization, extending the traditional Genetic Algorithm (GA) model by explicitly modeling the coevolution of cooperating species. This approach aims to improve the performance of GA-based optimizers and evolve complex structures like neural networks and rule sets. The authors introduce the concept of *cooperative coevolutionary genetic algorithms* (CCGAs), where each species represents a subcomponent of a solution, and fitness is assigned based on the interactions among these subcomponents. The performance of CCGA-1, a simplified version of CCGA, is compared with a standard GA on several multimodal functions, showing significant improvements in both solution quality and convergence speed. However, CCGA-1 struggles with functions that have strong variable interactions, leading to the development of CCGA-2, which improves performance on such functions but slightly degrades on non-interacting variable problems. The paper concludes by discussing the potential of CCGAs for solving more complex problems and their suitability for parallel architectures.The paper presents a cooperative coevolutionary approach to function optimization, extending the traditional Genetic Algorithm (GA) model by explicitly modeling the coevolution of cooperating species. This approach aims to improve the performance of GA-based optimizers and evolve complex structures like neural networks and rule sets. The authors introduce the concept of *cooperative coevolutionary genetic algorithms* (CCGAs), where each species represents a subcomponent of a solution, and fitness is assigned based on the interactions among these subcomponents. The performance of CCGA-1, a simplified version of CCGA, is compared with a standard GA on several multimodal functions, showing significant improvements in both solution quality and convergence speed. However, CCGA-1 struggles with functions that have strong variable interactions, leading to the development of CCGA-2, which improves performance on such functions but slightly degrades on non-interacting variable problems. The paper concludes by discussing the potential of CCGAs for solving more complex problems and their suitability for parallel architectures.