The remarkable scale and complexity of modern human societies stand in stark contrast to the small-scale groups characterizing much of human history. The emergence of large-scale societies (“macro-states” or “empires”) involving millions of individuals is a recent phenomenon, primarily observed within a specific band of the Afro-Eurasian landmass. This geographical concentration suggests a significant role for ecological factors, but several other hypotheses have been proposed, including agricultural productivity, information processing capabilities, warfare intensity, continental geography, technological advancements, and religious factors. This study uses cultural evolutionary theory (CET) to systematically test these competing hypotheses within a unified framework, focusing on factors influencing the geographical and temporal distribution of large-scale societies.

Existing literature highlights various factors potentially driving the evolution of social complexity. Diamond (1997) and Nielsen (2004) emphasize the role of agriculture, both its duration and productivity. Morris (2013) focuses on information processing, while Turchin et al. (2013) highlight the impact of warfare. Other factors proposed include continental geography (Diamond, 1997), technology (Morris, 2013), and religion (Norenzayan et al., 2014). However, comprehensive empirical tests comparing these hypotheses have been lacking. This research addresses this gap by applying CET to systematically examine and test the relative importance of various socio-ecological factors.

To test the hypotheses, the researchers compiled a spatially explicit dataset using Geographic Information Systems (GIS). Data were organized into grid cells (10,000 km²) under an equal-area projection. The dependent variable was “imperial density,” representing the frequency of large polities (>100,000 km²) occupying each cell over the period 1500 BCE–1500 CE. Data on polity extent were compiled from historical sources and sampled at 100-year intervals. The study focused on cells where agriculture was practiced by 1500 CE and where agricultural production was at least potentially possible. Predictor variables included distance from the Eurasian steppe (a proxy for warfare intensity), duration of agriculture, potential agricultural productivity, terrain ruggedness, and distance from the first empires (1500 BCE). Multiple versions of the distance-from-steppe and duration-of-agriculture variables were created to account for uncertainty in their measurement. Generalized least squares (GLS) analyses were employed to account for the spatial autocorrelation in the data, and model selection was based on Akaike Information Criteria (AIC). Analyses were conducted across multiple random subsamples of the data to ensure robustness. Time series analyses examined the predictive strength of different factors over sliding 1000-year windows.

Correlational analyses revealed positive relationships between imperial density and distance from the steppe, duration of agriculture, and distance from the first empires. However, potential agricultural productivity and terrain ruggedness showed weak or no relationship with imperial density. Spatially explicit GLS analyses confirmed that duration of agriculture and distance from the steppe were the strongest predictors of imperial density, with a significant interaction effect between these two variables. The model including these factors explained 58% of the variation in imperial density. The inclusion of “distance from first empires” as a control variable increased AIC scores, suggesting that the other factors captured additional explanatory power beyond the initial locations of complex societies. Time-series analyses indicated that the importance of distance from the steppe increased over time, possibly reflecting intensifying warfare or the expansion of European states. Duration of agriculture remained a strong predictor throughout the period, although its relative importance decreased slightly over time. In contrast, agricultural productivity and terrain ruggedness remained consistently weak predictors.

The study's findings strongly support the hypotheses that longer durations of agriculture and proximity to the Eurasian steppe (high warfare intensity) are crucial factors in the emergence of large-scale societies. The interaction effect suggests that the selective pressure for scaling up is amplified when the necessary cultural variations (norms and institutions) have had sufficient time to develop. The robustness of these findings is strengthened by the use of multiple measures of the key variables and the inclusion of distance from first empires as a control variable. The lack of support for the agricultural productivity hypothesis challenges the notion of simple environmental determinism. Instead, the results emphasize the importance of human interaction with and adaptation to their environments.

This study demonstrates the importance of considering multiple interacting factors in explaining the emergence of complex societies. The strong support for the duration of agriculture and steppe warfare hypotheses highlights the role of both cultural evolution (institution building) and selection pressure (intense warfare) in shaping the distribution of large-scale societies. Future research should explore additional factors, such as resource control and societal connectivity, and extend these analyses to other regions and time periods to assess the generalizability of these findings. More fine-grained data and analysis could further refine our understanding of the complex interplay between ecological conditions, cultural evolution, and the development of large-scale human societies.

The study's limitations include the coarse resolution of the spatial data and reliance on proxy variables for warfare intensity and agricultural productivity. Some residual variation in imperial density may be due to measurement error or the influence of factors not included in the analysis. The focus on Afro-Eurasia and the period 1500 BCE–1500 CE limits the generalizability of the findings to other geographical regions or historical periods. Future work could also benefit from improved measures of cultural similarity and incorporating measures of societal connectivity to address any potential bias or omitted variables.