Scalable Models for Patterns of Life

Patterns of life (POL) are emergent properties of complex social systems such as neighborhoods or even cities (Schatz et al. 2012). In such systems, observable regularities manifest from the interactions of individuals’ behaviors and social norms. Simultaneously, these patterns of life impose structure on individual decisions. For example, a pattern of rush hour traffic arises from drivers’ decisions to commute at a certain time. Knowledge of rush hour influences individuals’ departure times. Accurately modeling POL is not only an academic pursuit; complex training and analysis efforts rely upon these models. As an example, social and cultural experts analyzed patterns of life in order to locate Osama bin Laden. These analysts, without any visual confirmation, cited the ways his household deviated from the “normal” pattern of life in the neighborhood, behaving much more secretively than his suburban neighbors (Preston 2012). Building on such social and cultural expertise, computational modeling of POL offers significant potential to continue extending POL’s practical application and theoretical study. In industry, POL models might support decisions about marketing, logistics, network security, or building design. Military and law-enforcement models could support observational and small-unit training, intrusion detection, emergency planning, or community relations. Computational POL models will benefit academic researchers by giving them new tools to generate and validate theories about complex sociocultural patterns. Finally, POL presents an interesting computational modeling challenge for research in many AI fields because of its complex, multi-level interaction. POL modeling differs from standard agent-based modeling in that not only individual behaviors, but also their collective emergent patterns, are targets of study. At the same time, the behavior of any single individual is potentially important. As a specific example, the Virtual Observation Platform (VOP) is an immersive simulation designed to train mili-