Our experiments demonstrated that online, incremental, supervised learning can be effective at learning functions with continuous domains. We found that the degree of generalization in memory affected the speed of learning. Using a memory size appropriate to our task, we then saw that adaptive memory made it possible to learn both time-varying and nondeterministic concepts. Finally, we demonstrated that short-term performance was better when acting with a memory trained on a concept related to but different from the testing concept, than when starting from scratch. This paper reports extensive experimental results on our work towards multiple learning agents, both cooperative and adversarial, in a continuous environment.