Stigmergy in Comparative Settlement Choice and Palaeoenvironment Simulation

Decisions on settlement location in the face of climate change and coastal inundation may have resulted in success, survival or even catastrophic failure for early settlers in many parts of the world. In this study we investigate various questions related to how individuals respond to a palaeoenvironmental simulation, on an interactive tabletop device where participants have the opportunity to build a settlement on a coastal landscape, balancing safety and access to resources, including sea and terrestrial foodstuffs, whilst taking into consideration the threat of rising sea levels. The results of the study were analysed to consider whether decisions on settlement were predicated to be near to locations where previous structures were located, stigmergically, and whether later settler choice would fare better, and score higher, as time progressed. The proximity of settlements was investigated and the reasons for clustering were considered. The interactive simulation was exhibited to thousands of visitors at the 2012 Royal Society Summer Science Exhibition at the Europes Lost World exhibit. 347 participants contributed to the simulation, providing a sufficiently large sample of data for analysis.

💡 Research Summary

This paper investigates whether human settlement decisions exhibit stigmergy—indirect coordination through environmental traces—by using an interactive tabletop simulation of a Holocene coastal landscape. The authors reconstructed a virtual early‑Holocene (c. 8,000–7,500 BP) environment based on GIS, paleo‑environmental, and archaeological data, emphasizing resource‑rich bays, mixed forest, and marine food sources. During the 2012 Royal Society Summer Science Exhibition, 347 visitors each performed a single session in which they placed a settlement on the tabletop, balancing safety (elevation relative to projected sea‑level rise) against access to terrestrial and marine resources. The system automatically calculated a “survival score” that weighted safety and resource acquisition, and displayed previous participants’ settlements as 3‑D pyramids to serve as potential stigmergic cues.

All interaction data—coordinates, resource ratios, final scores, and distances to earlier settlements—were logged. The authors then applied spatial clustering analysis, correlation tests, and linear regression to answer three questions: (1) Do settlements cluster around particular landscape features? (2) Does the average score improve over time, suggesting learning? (3) Is proximity to earlier settlements associated with higher scores, indicating a stigmergic benefit?

Results show clear clustering: participants preferentially built near low‑gradient coastlines and resource‑dense bays, a pattern that is statistically distinct from random placement (p < 0.01). However, the hypothesised stigmergic effect was not supported. Distance to previous settlements showed a negligible negative correlation with score (r ≈ ‑0.08) and time‑based score trends were not significant (p = 0.27). The authors interpret this as either participants treating the simulation as a novel challenge, deliberately ignoring existing “traces,” or the exhibition context limiting the emergence of true social learning.

The study’s strengths lie in its integration of real paleo‑environmental data with a user‑friendly, high‑resolution interface, and in gathering a relatively large dataset from non‑expert participants. Limitations include the non‑representative sample (museum visitors), the single‑turn design that precludes simultaneous group dynamics, and the reliance on a researcher‑defined scoring system. The paper suggests that more sophisticated stigmergy metrics (e.g., trace visibility, cognitive load) and longitudinal, multi‑player experiments are needed to capture self‑organising processes observed in animal societies and, potentially, in past human groups.

In conclusion, the work demonstrates a novel methodological bridge between complexity science, archaeology, and human‑computer interaction, offering a proof‑of‑concept that virtual simulations can be used to test hypotheses about ancient settlement behaviour. Future research should expand the experimental design to include concurrent participants, varied climate scenarios, and cultural variables (ritual, tradition) to better assess how environmental cues may have guided real prehistoric colonisation and how such insights might inform modern responses to sea‑level rise.