I enjoy utilizing agent simulations in order to satisfy my curiosities in human behavior and cognition, particularly in the nuanced context of social human interaction.
I do so utilizing both simulated and virtual representations of agent interactions (either with other agents or humans).
An integral component to my agent modeling research is the simulation tool, PsychSim (Pynadath, Marsella, & Read, 2003). Agents in PsychSim are equipped with a “theory of mind” about the self and other agents in each resepective simulation, including the human user, and these intelligent agents try to pursue goals in the negotiation based on their different parameter settings.
The PsychSim system is the basis for a project both simulating and predicting causal models of evacuation during natural crises (e.g., hurricanes), where we examine the spread of scientific information across a complex information network to explore the conditions for successful communication of risk and science during natural disasters.
For instance, we can generate an agent-based simulation of the damage and casualties given certain parameters of a population, such as demographics (e.g., family size, wealth/employment), physical location (e.g., proximity/region), access (e.g., access to shelter, capacity to evacuate), psychological variables (e.g., attachment style, theory of mind), and communication source variables (e.g., federal evacuation notice, evacuation recommendation from friends/neighbors, social media posts).