Exploring behavioral regions in agents' mental maps

Authored by Paul M Torrens

Date Published: 2016

DOI: 10.1007/s00168-015-0682-0

Sponsors: United States National Science Foundation (NSF)

Platforms: No platforms listed

Model Documentation: Other Narrative

Model Code URLs: Model code not found

Abstract

Agent-based models continue to grow more sophisticated at the individual scale of regional science inquiry, but it remains difficult to ally the intricacies of individual behavior in those models to regional phenomena and processes in anything but a loose fashion, leaving explanatory pathways between the scales quite slack. In this paper, we present a mechanism for bridging the gap between individual agency and regional outcomes of that agency in simulation. We use agents to develop very rich behavioral understanding of their surroundings in simulation. We then sweep through the information that agents generate when determining how to execute their transition rules, using schemes that mine agent states to produce mental maps of varied aspects of their dynamics. From agents' mental maps, we define and visualize regions as geographies that are conjured from the unique, autonomous, local, and personal insight that agents can provide. We demonstrate the utility of the scheme, with application to indoor movement scenarios in which fleeting regions form amid agents interactions with each other and their built surroundings. Our approach is extensible beyond these applications and could be of broader use for other explorative scenarios in regional science.

Tags

models Dynamics time Space Geography Place World Residential preferences Plane