Psychologically Realistic Decision-Making in Multiagent Simulations

This dissertation demonstrates formalizing psychological phenomena is essential for advancing both Computational Social Science and Psychological Science. To achieve this, I examine heuristic decision-making at three levels: the individual level, the group level, and the city level. In the second chapter, I formalize confidence by modifying an equation from ACT-R (Anderson & Lebiere, 2014) and replicating some findings from the literature. The results also highlight an adaptive role of confidence in heuristic decision-making. In the third chapter, I investigate the interaction between heuristics, confidence, and decision aggregation methods in groups. The results align with Conte and Giardini’s (2016) perspective, which emphasizes both computational science and social science benefits from computational modeling. The insights generated demonstrate the adaptive role of confidence in group decision-making. In the fourth chapter, I compare an agent-based model (ABM) where agent decision-making is based on heuristics to other ABMs, with the purpose of explaining solar-panel adoption trends. The results in terms of error measures are lower or comparable to other approaches such as neural networks or The Theory of Planned Behavior (Ajzen, 1991), while using less data than these other approaches.

Author

• Aloraini, Fahad Abdullah H.

• ACT-R
• Agent-Based Modeling
• Confidence
• Fast and Frugal Heuristics
• Fluency heuristic
• Group decision-making

Identifier

hdl:1920/14722

OAI identifier oai:identifier

oai:MARS:1920/14722