Transferring knowledge from locational predictive models into spatial Agent Based Models: generilizing human perception of environment

Predictive models have been used for finding potential archaeological site locations already for more than 40 years. In addition to their extensively used application in cultural heritage management they can also offer insights into past human behaviour in relation to environment. The proposed paper discusses how the information gathered from predictive modelling of archeological settlements, especially the revealed from location determinants, can be transferred to spatial Agent Based Models.
We argue that spatial simulation models, especially those based on agents behaviour, above all aim to build theory and create generalisable knowledge. They therefore need to free themselves form local contexts and particular spatial heterogeneities, and automatically test the effect of agents behaviour on a wider range of spatial configurations. In cultural heritage management, models are usually created using geographical variables of existing settlement patterns (eg. water access and land formations). For finding archaeological sites the model needs to be calibrated for specific accuracy and precision. In this paper we focus on defining the potential residential space – an area that could be considered by people of the past as suitable for habitation or specific economic activities. We develop a framework that links geographical variables and behaviour as extracted from empirical studies to the development of a theoretical spatial agent-based model.