We hypothesize that humans tend to think in objects while nature can be interpreted as gradients of matter and processes. Such gradients can be steep, like the borders between water and land or between forest and pasture. But objects and decomposing complexity call for a scale. For place-based GIS, the scale issue and the ability of handling multiple scales are even more crucial than for classic GIS. We argue that the paradigm of near-decomposability of systems can play an important role in GIScience research and for the foundation of place-based GIS.