The current issue of ORMS Today is now available to INFORMS members in "digital magazine" format. Log in to the Digital Editions Archive and enjoy a new online reading experience.

Software Surveys

Biennial survey of statistical analysis software...
more »

The Pedagogy of Zombies

Given their cult-like following, many educators have turned to zombies to act as attention-grabbing pedagogical tools.

By Brant M. Horio and Nathan Arrowsmith

Professor Scott Grasman, head of the Industrial & Systems Engineering Department at RIT, gets “zombified”

Professor Scott Grasman, head of the Industrial & Systems Engineering Department at RIT, gets “zombified” in keeping with the spirit of the modeling competition.

The zombie apocalypse is quite real and over the years a great diversity of these aberrations have invaded deep into our human world. The business of zombies and their role as cultural objects in today’s society is conservatively estimated to contribute $5.74 billion to the global economy per annum [1]. Zombies seem to be everywhere, from movies and television shows to video games, organized events and merchandise. Given their cult-like following and entrenchment in the social fabric of today’s culture, it is no wonder that many have turned to zombies to act as attention-grabbing pedagogical tools, where an apocalyptic zombie invasion is used as an educational metaphor; examples include disaster planning [2] and learning how diseases spread through a population [3, 4].

Complex topics are easier to introduce when you can use engaging scenarios like zombie apocalypses because such scenarios – and the work done within them – are memorable and allow you to more easily adapt the material to real-world problems. Complex adaptive systems (CASs) are well suited for educational exploration within a zombie apocalypse framework. CASs are composed of many component parts – often referred to as agents – that interact and adapt or learn [5]. They frame many of our contemporary problems, including the spread of infectious diseases, diffusion of radicalism over networks and financial markets.

What makes them complex, however, is that the behavior of the system itself is less about individual agent behaviors and more about the behaviors that arise, or emerge, from the network interactions of the collectives that these agents form, e.g., coalitions, crowds, firms, markets and countries [6]. Neurons have relatively simple behaviors in how they pass signals throughout the brain, but from the collective ensemble of these cells emerges phenomena such as consciousness, cognition and personality. And these phenomena are constantly in flux as the inputs to the system are constantly changing with time, driven by factors such as past experiences and environmental conditions.

Because of this constant evolution, CASs tend to operate far from an equilibrium state [7]. This makes the use of common modeling approaches founded on linear techniques and assumptions of equilibria ineffective to approximate complex agent behaviors. Agent-based modeling (ABM) is one way to explore CASs and mitigate these shortcomings [5].

Agent-Based Modeling

As an example from research at LMI – a not-for-profit strategic consulting firm serving the federal government – ABM techniques were leveraged for work with the National Aeronautical and Space Administration (NASA). NASA was interested in modeling how U.S. commercial airlines would respond to potential policy scenarios and any consequential effects of airline decisions on consumers and the airline industry [8]. This example is a CAS because policies that impact airline operating costs may influence airfare pricing, potentially changing consumer purchasing behavior and transportation modal choice. Those changes in turn influence how airlines respond in adjusting their pricing strategies, service network designs and assigning specific aircraft to specific routes.

These changes at the agent level impact the collective behaviors of the consumer markets along airline routes and the service industry formed by the airlines, all of which impact the larger emergent system behavior, described by metrics for airspace congestion, delay and emissions impacts. These system-level outcomes feed back into subsequent airline and consumer decisions, enabling airlines to respond to the constant evolution of the system.

ABM is a valuable tool that if appropriately applied, allows insights into how a system works and explores how interventions change the system and in what ways. What might be driving factors for change? What are the conditions for specific system phenomena to emerge? Do “lever points” exist such that small change interventions might propagate to larger system change [7]?

LMI has recently sought to enrich the learning experience for its cooperative education (co-op) students by introducing them to this rapidly expanding field and giving them hands-on experience with agent-based models. LMI’s co-op cohorts are usually operations researchers, industrial engineers and applied mathematicians in training. They typically are somewhat familiar with simulation modeling and its general concepts, but most do not have any awareness of ABM and how it may be applied to problems and questions of complex systems.

As a case study for how the context of zombies was used for teaching a complex topic, the remainder of this article briefly describes the program at LMI for introducing ABM and an initiative currently underway in which LMI scaled up the program for the Rochester Institute of Technology’s Industrial & Systems Engineering Department (RIT).

For the LMI program, an informal instructional session was held focusing on the following:

  • Introductory concepts of complex systems
  • ABM as one method for researching complex systems
  • Model documentation using the ODD protocol [9], a standard increasingly being adopted by the ABM community
  • Demonstration of NetLogo [10] (a freely available ABM platform) and basic programming
  • Discussion of how zombie modeling is a direct metaphor for many other contemporary problems (e.g., infectious diseases, computer viruses on a network)
The model simulated populations of humans and zombies that were randomly generated on a floor plan map of the office.

The model simulated populations of humans and zombies that were randomly generated on a floor plan map of the office.

Simulation Modeling Competition

LMI followed this with a simulation modeling competition, dubbed Contingency-Z@LMI, to allow participants to expand upon the introductory material. The model simulated populations of humans and zombies that were randomly generated on a floor plan map of the office. Participants were only allowed to modify the logic for the humans in an effort to enable effective zombie evasion.

Constrained by office walls, the humans have limited maneuverability and must find random offices designated as “safe rooms” in which they could barricade the door. Zombies pursue humans only if they detect them within a “smell radius,” else they randomly wander around. Zombies can smell through doors and if they detect a human inside, they will approach the door and try to push it in. Zombies indirectly cooperate, and if others notice a zombie trying to push in a door, they will move to investigate. This usually results in a congregating mass of zombies that collectively try to overpower any humans inside. The challenge therefore is about getting humans to first identify nearby safe rooms and then to get there as quickly as possible with as many people as possible to ensure safety since the assumption is five or more humans in a room can indefinitely hold the door shut.

LMI provides a complete model but the human agent logic is simplistic and ineffective; competition participants have an opportunity to implement new evasion behaviors and other modeling features to improve upon the basic model. In this way, they may explore learned techniques and develop ABM programming skills. At the end of the competition, LMI runs the entries for a fixed number of replications using predetermined random seed values and identifies the model that results with the highest median survival after 1,000 time units of the model.

At RIT, LMI is sponsoring a similar event with support from INFORMS and IIE – codenamed Contingency-Z@RIT – with RIT faculty briefing contestants on the basics of ABM, followed by an RIT-customized model. Instead of the LMI floor plan, human agents are fleeing zombies in their engineering building.

While this year’s Contingency-Z@RIT competition is being held as you might be reading this (Oct. 1-Oct. 15), there are high hopes for zombies leading the way to a rewarding experience. Former co-ops have reached back to LMI after returning to school to describe how they have the edge on classmates in modeling courses when the topic of agent-based modeling comes up, not just with respect to familiarity with the terms and concepts, but because of hands-on practical experience leading humanity to survival.

Brant M. Horio ( is a consultant with LMI, in Tysons, Va. Nathan Arrowsmith ( is the student president of the INFORMS chapter at the Rochester Institute of Technology and the winner of the 2014 Contingency-Z@LMI modeling competition. Horio and Arrowsmith are modelers by day, but zombie hunters by night.

  1. Ogg, J.C., 2011, “Zombies Worth over $5 Billion to Economy,” available from: Last accessed Sept. 1, 2015.
  2. Centers for Disease Control ( Last accessed Sept. 1, 2015.
  3. Alemi, A.A.; Bierbaum, M.; Myers, C.R.; and Sethna, J.P., 2015, “You Can Run, You Can Hide: The Epidemiology and Statistical Mechanics of Zombies,” Quantitative Biology.
  4. Zombietown ( Last accessed Sept. 1, 2015.
  5. Holland, J.H., 2006, “Studying Complex Adaptive Systems,” Journal of Systems Science and Complexity, Vol. 19, No. 1, pp. 1-8.
  6. Watts, D., 2013, “Computational Social Science: Exciting Progress and Future Directions,” The Bridge, Winter 2013, pp. 5-10.
  7. Holland, J.H., 1995, “Hidden Order: How Adaptation Builds Complexity,” Addison Wesley Longman Publishing Co., Inc., Redwood City, Calif.
  8. Horio, B.H.; Kumar, V.; and DeCicco, A., 2015, “An Agent-Based Approach to Modeling Airlines, Passengers and Policy in the U.S. Air Transportation System,” Proceedings of the Winter Simulation Conference 2015, Huntington Beach, Calif.
  9. Grimm, V.; Berger, U.; DeAngelis, D.L.; Polhill, J.G.; Giske, J.; and Railsback, S.F., 2010, “The ODD Protocol: A Review and First Update,” Ecological Modelling, Vol. 221, No. 23, pp. 2,760-2,768.
  10. Wilensky, U., 1999, NetLogo, Northwestern University, Center for Connected Learning and Computer-Based Modeling, Evanston, Ill.