Summary

The Zambezi Plateau region in Southern Africa experienced the formation and fall of archaeologically visible polities with different levels of sociopolitical complex­ity during many centuries, before the arrival of Europeans and the beginning of the region’s written history.

The agent-based model presented here, called ZambeziLand 1.0, provides support for a theoretically grounded explanation of settlement and abandonment based on the Canonical Theory. In this theory, a succession of opportunities to engage in collec­tive action by individuals and groups in society (iterative fast processes with canon­ical variations) strengthens or weakens the complexity of their respective polity (the singular slow process of each society). Iterations of this so-called “fast process” over time generate broader institutional changes whereby the effects of collective action within each fast process accumulate through a “slow process” resulting in a polity with variable and seemingly idiosyncratic but explainable levels of complexity over time.

The ZambeziLand model provides an explanation of how a society can change its complexity over time through decisions made by group members in fast processes. In the model, groups rose, declined, and disbanded as leadership and feelings of loyalty and group attachment rose and fell. Such feelings were affected by successes and failures in collective action, and the probability of success was dependent in part on the strength of group leaders. Comparable dynamics occur today in all societies.

The main finding presented here is that group dynamics, centered on collective feelings of loyalty to a group, can generate the macro-level behavior observed in the archaeological record of Southern Africa. This computational finding has implica­tions for further investigation into the role of ideologies and imagery, especially on views of group leadership and loyalty among the people that built the monumental zimbabwes of Southern Africa.

Acknowledgments An earlier version of this paper was prepared by the first author, who also wrote the original model code, for the Seminar on Origins of Social Complexity, Department of Computational Social Science, George Mason University, Fall 2012, taught by the second author. This version was extensively revised by both authors for presentation at the Conference on Simulat­ing the Past to Understand Human History, Barcelona, Spain, September 3-4, 2014, organized and chaired by Juan Antonio Barcelo, Autonomous University of Barcelona. All errors are the responsi­bility of the authors, who declare no conflicts of interest. Thanks to Jeff Bassett for introducing the authors to ShareLaTeX, the online system used to write and compile this paper while maintaining a single version, and to Henry Oswald and James Allen of the ShareLaTeX team in the UK for enabling this collaborative result (and others to follow).