When we learn something, we always learn it in a specific context. For example, we might learn to write a critical essay about Hamlet. The goal of the experience, though, is not only to write about Hamlet, or even about Shakespeare in general, but to be able to think and write critically about an arbitrary piece of fiction.
How do we generalize our knowledge and apply it in different circumstances? This is known as the problem of transfer. If we hope to create games for learning, we must demonstrate that students are capable of transferring knowledge acquired in the game into other contexts – including improved performance on academic tasks.
Before we answer this question, we must define more carefully what type of transfer we are looking for. Classical transfer, for example, requires students to produce precisely the parallels that an authority figure (a teacher or experimenter) expects to see. More recent research on transfer suggests that transfer is complicated, idiosyncratic and often only partial. In either case, transfer is improved when learners reflect on their choice of strategies and are explicitly encouraged to generalize.
To understand whether transfer occurs in games, we build on the concept of transfer as preparation for future learning. This theory proposes that games do not generally, by themselves, produce transferable learning outcomes. This is because game-play provides a set of experiences rather than general principles, and cannot be designed as explicitly instructional experiences without becoming bad games. However, the experiences provided in the game prepare students to take better advantage of explicit instruction by giving players a base of (virtual) experience to draw on. Our theory proposes, therefore, that students who play games connected to academic knowledge domains will learn those domains more quickly and effectively than their peers.
These effects are only likely to be visible for well-designed games that deeply engage players. Additionally, we must control for overall effects of game-play if we hope to understand the impact of specific games on specific academic domains. We therefore chose two excellent commercial games, Civilization and SimCity, to investigate our hypothesis. In addition to being deeply engaging, each game models a particular learning domain: history for Civilization and urban planning for SimCity.
We recruited two groups of players. One group was expert in Civilization but not in SimCity – the other group, vice versa. That let us compare the effects of each game on how players learned history and urban planning, while controlling for potential overall effects of being an expert game player. We gave our subjects a pre-test on their history and urban planning knowledge. We then asked them to read a textbook excerpt on each subject. Finally, we tested how much they learned from what they read.
We found, as expected, that players did not show a comparative advantage for “their” domain on the pre-test. Civilization players did no better in history than their SimCity-playing peers, nor did SimCity players excel at urban planning. In other words, games did not seem to evoke transfer effects by themselves.
After reading the text, however, that changed. Civilization experts outperformed SimCity experts on the post-test, suggesting that they learned more from the textbook excerpts provided. They did not show a similar benefit for urban planning, suggesting that the effect is limited to the domain in which they play.
Perhaps the most interesting result, however, is that SimCity players did not outperform Civilization players at learning about urban planning. In other words, Civilization prepared players for history learning, but SimCity did not help players learn about urban planning. Further research is necessary to understand why. It may relate to the design of the game or to the nature of the domain itself, which we hope to investigate empirically.
Our research suggests, therefore, that games with academic content require explicit educational support and scaffolding in order to be useful in the classroom. However, they may be of great value in preparing students to learn better from textbooks, lectures and other formal learning experiences. Finally, our research implies that not every game that claims to contain academic content is useful – even if it is extraordinarily well-designed. In the long term, we hope to identify the principles behind why some games provide useful preparation for future learning, and others do not.
Changing the Rules 