Solomou, S., 2024.
Incorporating alternative rational behaviour theories into urban simulative models
Output Type: | Thesis or dissertation |
Within urban design and planning, computational models help understand and explore future demand for space in specific locations and situations. In the literature these are a category of urban simulative models classed by this research as real-estate demand models. These models incorporate modules that deal specifically with market forces of demand in an attempt to anticipate human behaviour and response to future intervention and policy change. The current trend of increasingly disaggregated urban scale operational planning support models revolves around microsimulation and agent-based models (ABM) that incorporate individual behaviour. The individual behaviour represented in these real-estate demand models lends heavily on economics theory. The thesis aims to enhance the evaluation and prediction of demand and value associated with specific designs or spaces within the urban environment. It addresses the gap in current urban simulative models, which primarily rely on objective rationality theories to predict individual decision-making regarding land, space, and design choices. The research investigates the potential of new disaggregated behavioural techniques, focusing on improving agent decision-making mechanisms to allow for subjective rationality. This involves exploring innovative theoretical bases and agent architectures to better represent individual behaviours within computational models.
The thesis objectives include a classification of real-estate demand urban simulation models, a systematic review of these models that evaluates current agent architecture usage, the construction of novel computational models for real-estate demand, their validation through human role-playing simulations, and the drawing of conclusions on their capacity to represent subjective rationality. The findings of the initial review indicate that current real-estate models lack certain agent attributes, such as cognitive abilities, collaboration, belief-desire-intention structures, and pro-activeness traits. The research creates three distinct ABMs incorporating different theoretical foundations and agent architectures: Theory of Planned Behaviour with Belief Desire Intention (BDI) agent architecture, Case-Based Decision theory with cognitive agent architecture, and a classic utility maximization model with logic-based architecture. The suitability, strength and weakness as well as external validity for each type of model is analysed against the results of an active role-playing simulation that substitutes the decision-making of computational agents with those of real people in a virtual environment that mimics the one present in the aforementioned three different computer agent simulations.
The research demonstrates the potential of introducing currently lacking agent attributes in the field through innovative theoretical bases and agent architectures, as evidenced by the performance of the created models when compared to the human role-playing experiment. Key findings include the demonstration of the limitations of traditional neo-classical economic theories in urban modelling, and the potential of alternative decision-making theories. Moreover, it emphasises the potential of cognitive agents to improve decision-making processes and design outcomes in urban planning. Additionally, it comments on the need for future research to overcome limitations, such as the lack of real-world representation and the impact of scale on agent decision-making. Overall, this work underscores the importance of considering subjective factors in urban modelling and design to better reflect real-world complexities.