This paper proposes that participatory models of engagement and associated design-based research methods can provide useful guidance in this regard. Such models grant primacy to collective participation in domain-specific discourse, while treating both the cognition and behavior of individuals as “special cases” of socially situated activity. Our own model is being refined and tested in design-based studies of Quest Atlantis (QA), an on-line multi-user academic/fantasy environment for 8-12 year olds.  Completing the various academic “quests” embedded in narrative “missions” gives students “cols” (used to purchase virtual or real commodities) and “lumens” (used to signify status within the narrative backstory). 

The most recent design cycle involved four classes of 5^th graders who devoted ten class periods to learning ecological science topics and scientific inquiry.  Two classes completed Taiga, a QA-based curriculum, while two other classes used a conventional textbook curriculum.  Analysis of discourse “in QA” (student submissions, feedback, chat) and “on-QA” (student & teacher conversations) revealed ways that the incentives both encouraged and discouraged valued discourse.  Conventional measures of individual cognition revealed comparable gains on near- and far-transfer measures of academic knowledge but a significantly greater improvement in domain interest in the QA classrooms.  These insights are being used to refine Taiga and QA more broadly to enhance engagement and the resulting individual learning and motivation.  In early 2007, multiple versions of the Taiga curriculum will again be implemented in the same teacher’s four classes to more formally test several key refinements.

Recent years have seen dramatic advances in video-gaming.  On-line virtual environments give millions of individuals a remarkably embodied experience with virtual and real others.  Pervasive use and scholarly evidence that gaming supports unprecedented levels and forms of learning have led to an explosion of interest in the potential of video games by learning theorists.  This interest is particularly evident among theorists who embrace contemporary sociocultural perspectives.  Leading theorists are advancing the use of games for learning academic and other formal content, and as a laboratory for understanding and improving learning more broadly.  Arguably, the synergy between gaming environments and sociocultural theories of learning will rival the synergy in the early ‘90s between multi-media technology and socio-cognitive theories.  This paper will examine one aspect of this current synergy that concerns the use of incentives, and engagement and motivation in general.

Current efforts suggest guidance is needed concerning incentives, which are essential in most non-academic games but have been shown repeatedly to undermine intrinsic motivation and (sometimes) undermine subsequent engagement in learning.  Competing meta-analyses of the hundreds of prior studies of incentives suggest that the continuing debate between behavioral and cognitive theorists reflects a basic philosophical difference that is unlikely to be resolved empirically.  

Certainly there are guidelines available for using incentives that many game developers and educators would do well to consider (e.g. from self-determination theory).  Nonetheless there is little consensus over the use of incentive systems and virtual economies that drive the most engaging games to motivate engagement in academic content.  For example, one published  study suggested that such incentives were useful in the multi-user virtual learning environment that is the focus of this paper, but other recently published papers have argued that the same sort of incentives distract from learning as players concentrate on completing, scoring, or winning.  New guidelines and models for their continued refinement are needed.

This paper suggests that emerging situative, sociocultural models of engagement might help guide the use of incentives, particularly in specific contexts such as gaming environments.  These models assume that all learning represents social change, and vice versa.  This leads to an approach to engagement based on the core sociocultural assumption that knowledge ultimately resides in the context of its use, in the “knowledge practices” that define “knowledge communities.”  In this view, the goals and values presumed to motivate learning of particular knowledge reside alongside that knowledge in the social context of use.  When knowledge (or more accurately—knowing) is viewed this way, engaged participation belongs at the core of a sociocultural theory of motivation.  From this perspective incentives are not inherently detrimental or beneficial for supporting engagement in knowledge practices associated with the target domain.  Rather, incentives should be refined primarily to maximize collective participation in those practices, with impacts on behavior and cognition considered secondarily.  Doing so provides a relatively neutral lens through which to judge consequences of incentives, potentially providing useful empirical guidance on the use of incentives in specific contexts.

As a caveat, the value of our approach is most apparent in the design-based refinement of “local” theories of incentive practices in particular contexts.  It remains to be seen whether or not this research trajectory will ever provide evidence that will resolve the more fundamental (and arguably philosophical) debate between cognitive and behavioral theorists over incentives.     Resolving this debate is not the goal of this research.  Rather, we take an agnostic view towards incentives, noting their seeming ubiquity in most educational contexts in the form of grades, and acknowledging the increased reliance on incentives within test-based accountability and behaviorally-oriented classroom management and school attendance programs.  With this in mind, motivational researchers are obliged to find ways to minimize negative consequences and maximize positive ones.

Research Context

The aforementioned approach to incentives is being refined and tested in a series of design-based studies using Quest Atlantis (QA), an on-line multi-user academic/ fantasy environment for 8-12 year olds (pictured).  Completing the various academic “quests” embedded in narrative “missions” gives students “cols” (used to purchase virtual or real commodities) and “lumens” (used to signify status within the narrative backstory). QA is an ideal laboratory for studying things like incentives, because the authoring software allows ready manipulation of features like incentives, while log files and tracked submissions make it possible to track activity and engagement. 

  The most recent design cycle involved the QA Taiga scenario, where students learn standard ecological science concepts like erosion, eutrophication, and scientific hypothesis testing, via interaction with virtual characters and data that relate to competing explanations for declining fish populations in the Taiga River (pictured).  This version of Taiga included several new incentives intended to motivate engagement.  Once each of the three assigned quests and two optional quests were accepted, students got virtual currency called ‘cols’ and virtual mp3 player components (earphones, batteries, and song.).  In this study, one teacher had two of his fifth-grade classes complete Taiga while he taught the same topics to his other two classes using a conventional textbook curriculum. 

Results

Analysis of discourse “in QA” (student submissions, feedback, chat) and “on-QA” (student & teacher conversations) revealed ways that the incentives both encouraged and discouraged valued discourse.   This revealed several clear instances where domain-specific discourse was discouraged by the new incentive, but encouraged by students’ desire to obtain cols and lumens.  More importantly, it revealed clear suggestions for improving the way that cols and lumens are given that seem likely to further encourage productive forms of scientific discourse.

 Conventional measures of individual cognition revealed comparable gains on knowledge assessments of near-transfer (topics targeted in both curricula) and far-transfer (aligned to the targeted content standards).  Self-report measures also revealed significantly improved domain interest in the QA classrooms.  These insights are being used to refine Taiga and QA more broadly to enhance motivation and engagement, and the resulting individual learning.  In early 2007, multiple versions of the Taiga curriculum will again be implemented in the same teacher’s four classes to more formally test several key refinements.