BOXES is a well known methodology that learns to perform control maneuvers for dynamic systems with only cursory a priori knowledge of the mathematics of the system model. A limiting factor in the BOXES algorithm has always been the assignment of appropriate boundaries to subdivide each state variable into regions. In addition to suggesting a method of alleviating this weakness, the paper shows that the accumulated statistical data in near neighboring states may be a powerful agent in accelerating learning, and may eventually provide a possible evolution to self-organization. A heuristic process is postulated that may allow strong areas of knowledge in the system domain to create larger cellular structures, while causing the more delicate, uncertain, or untested regions to take on a finer granularity. The paper theorizes that these untrainable states may contain valuable and sufficient information about the location of switching surfaces. The paper concludes with some future research questions.