Because clustering of organizational activities in space induces - and at the same time emerges from patterns of imperfect connectivity among interacting agents, the study of geography and strategy necessarily hinges on assumptions about how agents are linked. Spatial structure matters for the evolutionary dynamics of organizations because social systems are prime examples of connected systems, i.e. systems whose collective properties emerge from interaction among a large number of component micro-elements. Starting from this proposition, in this paper we explore the value of the claim that a wide range of interesting organizational phenomena can be represented as the outcome of processes that occur in overlapping local neighborhoods embedded in more general network structures. We document how patterns of spatial organization are sensitive to assumptions about the range of local interaction and about expectation formation mechanisms that induce temporal interdependence in agents' choice. Within the lattice world that we define we discover a concave relation between the sensitivity of individual agents to new information (cognitive inertia) and system-level performance. These results provide experimental evidence in favor of the general claim that the evolutionary dynamics of social systems are directly affected by patterns of spatial organization induced by network-based activities.