Scalable Representation Structures for Visuo-Spatial Reasoning — Dynamic Explorations into Knowledge Types