Adaptive knowledge dynamics and emergent artificial societies: ethnographically based multi-agent simulations of behavioural adaptation in agro-climatic systems

The goal of this research is to enhance an ethnographic understanding of agricultural adaptation to environmental change, within the context of an anthropological theory of 'adaptive dynamics' [Bennett, 1976], using computer-based techniques. An agent-based model was developed to investigate the 'transitional' adaptive strategies of farmers in south-east England based on data collected during fieldwork. Using ethnographic evidence, the model included the interactions of differing structures of knowledge relating to possible environmental change. This resulted in a variety of adaptive, non-adaptive and indeed mal-adaptive responses by agents in the system and thus, differing degrees of success for individual actors and the group as a whole. The choices made in response to change and their consequences were analysed. Success was measured in terms of minimising vulnerability, achieving sustainable adaptation and meeting economic objectives. Adaptive responses classified using criteria proposed by John W. Bennett [1976], under the heading of adaptive dynamics and incorporated within an agent-based model, allowed a refined understanding of the ethnographic data that was collected exposing new insights and areas for further investigation. The agent-based model illustrated the importance of Bennett's model in illuminating the benefits of indigenous strategies for successful adaptation and sustainability. The broad scope of this research means that it is aimed at an interdisciplinary audience. It is organised such that each chapter will contain a general introductory overview which requires little specialist knowledge, while further reading will entail greater technical detail, which will assume some specialist knowledge. Specialist areas covered include simulation toolkits, declarative programming software and social science research methodologies amongst others. This research is intended to be a guide for non-computing anthropologists to understand the potential in developing simple computer models to complement and support their traditional research methods. Experienced modellers may also benefit from specific techniques described from the social sciences domain, such as the participatory knowledge engineering process developed using ethnographic techniques, which may enhance conventional modelling approaches.