Authors:
Michael S. Horn, David Weintrop, Elham Beheshti, & Izabel Duarte Olson
Abstract:
Computational models and simulations can be powerful tools to help learners understand a wide
variety of natural phenomena (National Research Council, 2011). However, in order to construct
understandings of target phenomena—especially those that emerge from the complex interactions of a large number of entities—it is important for learners to also understand a model’s simplifying assumptions, including the rules that govern the behavior of individual computational agents (Jacobson & Wilensky, 2006; Son & Goldstone, 2009; Wilensky & Reisman, 2006; Wilensky & Resnick, 1999). For example, in a model of a predator-prey ecosystem (e.g. Wilenksy & Reisman, 2006; Wilensky & Resnick, 1999) learners ideally understand that the model represent simplified ecosystems with two or three types
of organisms; that these organisms reproduce asexually based on some fixed probability; that encounters between predators and prey are determined by chance discrete movements on a two-dimensional plane; and that simulations advance in fixed step-by-step time intervals (ticks).
In this paper we consider the use of specially designed board games as a way to prepare elementary school students to explore computational agent-based models. We have designed these games to closely resemble their computational counterparts in certain key respects. For example, for a wolf-sheep predation model, we designed an activity in which game pieces represent wolves and sheep that move on a board based on the flick of a spinner (rather than a pseudo-random number generator). These games aren’t necessarily intended for playing more than a few times—their fun is quickly replaced by the tedium of moving many pieces and keeping track of things like energy levels on a score sheet. But then one of our learning objectives is to help children appreciate why computers are useful tools for scientific modeling—they can make thousands upon thousands of precise computations in the blink of an eye and keep track of vast arrays of data. Other goals include introducing learners to the world of complex
systems and the rules that govern their behavior, foregrounding the role of randomness, and promoting productive collaboration in investigations of scientific phenomena through computational modeling.
Our hope is to build on both conceptual and social resources of game play to scaffold learners’
understanding of agent-based computational models. In this paper we discuss both our learning objectives and game designs in more detail. We also elaborate on a theoretical framework based on the use of transitional forms (e.g. specially designed board games) to help learners make use of conceptual and social resources of board game play in the context of agent-based modeling. This research is preliminary. We present observations from a pilot test of our board games and computational modeling environment with eleven elementary school children in a summer workshop in a research laboratory.
Video Summary: Coming soon
Reference:
Horn, M. S., Weintrop, D., Beheshti, E., & Olson, I. D. (2012). Spinners, dice, and pawns: Using board games to prepare for agent-based modeling activities. Presented at the American Educational Research Association annual meeting, Vancouver, BC.