Towards Intervention and Counterfactual modelling in spatial agents: A simulation of constrained movement at the Observational level

Increased availability of spatiotemporal data has allowed movement studies to shift their objectives from descriptive models to explanations of the underlying “causes” of movement. This paper presents a model, consisting of three conceptual levels of space-time representation for framing movement analysis: ‘association’, ‘intervention’ and ‘counterfactual’. These are presented within a conceptual cube accommodating both absolute and relative measurements in space and/or time, relationships between observations and methods of enquiry. To ground these concepts, an agent-based football simulation model is presented as a demonstration of the observation level and the properties of the system related to the conceptual cube. This team sport domain was chosen as it represents constrained space and time as well as simplified spatiotemporal object behaviour that serves to effectively reflect on aspects of the conceptual model. Future theoretical developments in computational movement analysis are proposed in light of the model and the associated development of machine learning methods for simulation and analysis addressing the intervention and counterfactual levels of the conceptual cube.