THE JOURNAL OF THE LEARNING SCIENCES, 16(3), 371–413 Copyright © 2007, Lawrence Erlbaum Associates, Inc.

Augmented Reality Simulations on Handheld Computers Kurt Squire School of Education, University of Wisconsin–Madison

Eric Klopfer AUGMENTED SQUIRE AND KLOPFER REALITY SIMULATION ON HANDHELD COMPUTERS

Teacher Education, Massachusetts Institute of Technology

Advancements in handheld computing, particularly its portability, social interactivity, context sensitivity, connectivity, and individuality, open new opportunities for immersive learning environments. This article articulates the pedagogical potential of augmented reality simulations in environmental engineering education by immersing students in the roles of scientists conducting investigations. This design experiment examined if augmented reality simulation games can be used to help students understand science as a social practice, whereby inquiry is a process of balancing and managing resources, combining multiple data sources, and forming and revising hypotheses in situ. We provide 4 case studies of secondary environmental science students participating in the program. Positioning students in virtual investigations made apparent their beliefs about science and confronted simplistic beliefs about the nature of science. Playing the game in “real” space also triggered students’ preexisting knowledge, suggesting that a powerful potential of augmented reality simulation games can be in their ability to connect academic content and practices with students’ physical, lived worlds. The game structure provided students a narrative to think with, although students differed in their ability to create a coherent narrative of events. We argue that Environmental Detectives is 1 model for helping students understand the socially situated nature of scientific practice.

This research was supported with a grant from Microsoft–MIT (Massachusetts Institute of Technology) iCampus as a part of the Games-to-Teach Project. We would like to thank Henry Jenkins of MIT and Randy Hinrichs at Microsoft Research, co-principal investigators of this project, for their support, as well as Kodjo Hesse, Gunnar Harboe, and Walter Holland for their hard work in the development of Environmental Detectives. Thanks to Susan Yoon for her helpful feedback on an earlier draft of this article. Correspondence should be addressed to Kurt Squire, School of Education, University of Wisconsin–Madison, 5446 Teacher Education Building, 225 N. Mills St., Madison, WI 53706. E-mail: [email protected]

372

SQUIRE AND KLOPFER

INTRODUCTION The use of computer simulations is changing the very nature of scientific investigation (Casti, 1998) and providing unique insights into the way the world works (Wolfram, 2002). Scientists can now experiment in a virtual world of complex, dynamic systems in a way that was impossible just years ago. These tools have led to discoveries on topics ranging from the origins of planets to the spread of diseases through human populations. In an effort to engage students in the authentic making of science, many science educators (e.g., Feurzeig & Roberts, 1999) have begun using models and simulations in classrooms as well (cf. Colella, Klopfer, & Resnick, 2001; Friedman & diSessa, 1999; Stratford, Krajcik, & Soloway, 1998). To date, most computer simulations have been tethered to the desktop, as they have relied on the processing power of desktop computers, but more ubiquitous and increasingly powerful portable devices have made entirely new kinds of simulation experiences possible (Holland, Jenkins, & Squire 2003). Handheld computers’ portability, social interactivity, context sensitivity, connectivity, and individuality open new opportunities for creating participatory and augmented reality simulations wherein players play a part in a simulated system, coming to understand its properties through social interactions (Colella, 2000). One possible genre of applications is augmented reality simulations, simulations in which virtual data are connected to real-world locations and contexts (Klopfer et al., 2001). In fields such as environmental science, where investigations are profoundly rooted in the particulars of local context, augmented reality applications invite science educators to bring the environment into the investigation process while exploring phenomena impossible to produce in the real world, such as diseases or toxic chemicals flowing through watersheds. By simulating a virtual investigation, educators can potentially initiate students into environmental science as a coherent social practice, as opposed to a set of disconnected procedures or body of facts. Investigating how a toxin such as trichloroethylene (TCE) spreads through a watershed might be educationally valuable (particularly for environmental engineering students who might eventually conduct such investigations), but it is obviously too dangerous to pursue. In this article, we explore how augmented reality applications might play a role in environmental science education as they allow curriculum developers to design game trade-offs around core disciplinary dilemmas (Cobb et al., 2000), nonlinear open-ended dilemmas with no clear boundaries, that are central to a field. This allows students to learn through failure, by intellectual play with robust disciplinary problems. Students’ reflections on their successes and failure combined with carefully crafted collaboration allow them to explore difficult and complex tasks while building expertise in the field. This research study examined the potential for creating an augmented reality application around the core of environmental science practice, as defined by faculty in a leading environmental research department. Specifically, we wanted stu-

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

373

dents to understand (a) trade-offs between efficiency and quality of data in conducting an investigation, (b) the importance of synthesizing background “desktop research” with secondary sources and primary data collected in the field, and (c) the necessity of continuously refining hypotheses in response to emerging data. In short, a struggle for students studying environmental science (particularly engineering students) is in understanding that research programs are situated in social contexts in which access to resources, affordances and constraints of tools, and, perhaps most importantly, time shape inquiry (Bhandari & Erickson, 2005; Latour, 1987). Emerging pedagogies such as case studies are increasingly used to help environmental engineering students understand the socially situated nature of engineering as a practice and see the interrelationships among variables in conducting an investigation. Within high school science curricula, these same educational goals align with most state earth science inquiry standards.

RESEARCH QUESTIONS Specifically, this research study investigated the following: How can handheld augmented reality technologies and game play be used to enrich inquiry and provide a new pedagogical paradigm for environmental science education? We hypothesized that an augmented reality game that positions players as environmental scientists conducting a virtual investigation of a hypothetical toxic spill (modeled on a similar case study) might help participants learn to see investigations as socially situated enterprises. As such, this research study also investigated the potential of designing learning environments using digital gaming conventions and aesthetics (e.g., character conventions) to enlist and mobilize game players’ identities and aesthetic considerations (Games-to-Teach Team, 2003; Gee, 2003). Working with environmental science faculty at the Massachusetts Institute of Technology, we developed augmented reality simulations of a carcinogenic toxin (TCE) flowing through an urban watershed, known collectively as Environmental Detectives. In a series of four case studies with approximately 75 students, we examined the following: (a) what practices students engaged in while participating in Environmental Detectives, and specifically how they integrated real and virtual data in problem solving and conducting their scientific investigations; (b) how students constructed the problem (e.g., as well defined or open ended, authentic or inauthentic); (c) how field investigation in the physical environment mediated students’ inquiry; and (d) what instructional supports were useful in supporting learning. We explored how augmented reality simulations could be used as learning tools for understanding the socially situated nature of science, specifically in situations in which educators want the physical environment to be a part of students’ thinking and scientific reasoning. Through presenting a series of case studies, we

374

SQUIRE AND KLOPFER

attempt to articulate how this pedagogical model can work, while also suggesting where there are limitations in our current understandings of how it functions.

THEORETICAL APPROACH: AUGMENTED REALITY AND SITUATED COGNITION Over the past decades, a growing number of educational theorists and researchers in the learning sciences have argued for the importance of understanding cognition in context (e.g., Barab & Kirshner, 2001; Brown, Collins, & Duguid, 1989; Cognition and Technology Group at Vanderbilt, 1990; Greeno, 1998; Kirshner & Whitson, 1997). Whereas traditional cognitive models treat the workings of the mind as somewhat independent, a host of emerging, complementary approaches to cognition treat cognition and context as inextricably linked. How these different approaches construct the notion of context depends on their underlying theoretical framework. In this article, we use this situated model of cognition as the basis for designing a curriculum around conducting investigations in environmental science. Specifically, we try to use augmented realities to situate learners in emotionally compelling, cognitively complex problem-solving contexts. Learning as Doing Greeno (1998) introduced the notion of situativity as a way of understanding the problem space of a learning episode. Greeno described problem space as “the understanding of a problem by a problem solver, including a representation of the situation, the main goal, and operators for changing situations, and strategies, plans, and knowledge of general properties and relations in the domain” (p. 7). Whereas traditional psychological models consider the individual learner operating without regard to context, situativity theorists argue that there is no such thing as context-independent thought and behavior. The central goal of educational psychology from this perspective is to understand performance as it occurs in socially meaningful situations, accounting for multiperson communal structures, individuals’ goals and intentions, and tools and resources that mediate action. Learning is always fundamentally about doing something for some purpose in a social context equipped with tools and resources, making the minimal meaningful ontology the “who, what, where, and whys” of a situation (Wertsch, 1998). Because learning is a process of creating meaning in situ, the environment plays an important role in the processes of knowing and learning; the environment constrains activity, affords particular types of activity or performance, and supports performance (Dewey, 1938; Peirce, 1868/1992; Salomon, 1993). Effective action is always situated within environmental constraints and affordances, and a mark of expertise is one’s ability to see the environment in particular ways (cf. Glenberg,

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

375

1997; Goodwin, 1994). If one is to take a situated view of environmental engineering, then a primary goal is to help students learn to see the environment as an environmental engineer might. Educators need to help students become attuned to the affordances and limitations of doing in environmental science, particularly navigating complex problem spaces with multiple variables and solutions. From this perspective, it is not enough for students to know a list of facts or procedures about environmental engineering. They need robust experiences in environmental engineering that can be the basis for future action. Indeed, from the situated perspective, an indictment of most school-based learning is the way that information is cleaved from direct experience in the physical world, processed and digested for learners (Barab, Cherkes-Julkowski, Swenson, Garrett, & Shaw, 1999). In the case of environmental science, this means being handed prepackaged research techniques (such as sampling strategies) or investigative design heuristics (e.g., investigations as social processes that involve managing budgets and constraints) without having opportunities to develop such understandings through action and to appreciate their practical importance. Results and procedures are handed to students ready made, divorced from the social contexts that produce them. Designing Learning Environments Based on Situated Learning Theory Apprenticeships have been posited as one model for education as they situate learners in complex tasks whereby they have access to expert cognition—including the social context of activity (Collins, Brown, & Newman, 1989). Unfortunately, apprenticing students to experts is not always feasible, particularly for secondary students or postsecondary students in an early stage of career development, as studied here. Apprenticeships are also often long, difficult, even exploitive. As Shaffer (2004) argued, a challenge facing contemporary learning scientists is how to recreate the most robust learning moments of apprenticeships (which often occur in the practicum), but in ways that are most efficacious for long-term learning. We hypothesized that augmented reality simulations are one possible way to engage learners in complex investigations within a context that is socially safe and feasible. Augmented reality approaches draw from earlier situated approaches, ranging from problem-based learning to case-based scenarios to anchored instruction, which Barab and Duffy (2000) called practice field approaches. In the context of environmental science, handheld computers allow students to collect data while conducting complex field investigations, access authentic tools and resources, and participate in collaborative learning practices while in the field. Whereas traditional desktop virtual reality applications or three-dimensional gaming technologies such as MUVEs burden the computer with reproducing reality in three dimensions, augmented realities exploit the affordances of the real world, providing

376

SQUIRE AND KLOPFER

users layers of data that augment their experience of reality. As a result, simulations are untethered from the desktop and learners can participate in technology-enhanced investigations, location-based games, or participatory simulations. Because players are free to move throughout the world, novel opportunities exist for learners to interact with the physical environment, literally reading the landscape as they conduct environmental investigations or historical studies. Inquiry and Environmental Investigations Augmented realities attempt to build on earlier work with digital tools that attempt to use technologies to mediate students’ interactions with science. Tools such as Model-It (Spitulnik et al., 1995) or Climate Watcher (Edelson, Pea, & Gomez, 1996) have been used to help learners engage in scientific modeling processes that enable students to build understandings of their environment or that mediate how students encounter dilemmas, collaborate in solving problems, and represent problem solutions (Salomon, 1993). Handheld augmented realities borrow much from the tradition of handheld probeware and the inquiry-based labs that have been designed around this technology (Staudt, 2001). Using handhelds with data logging probes attached, students can collect real-time data out in the field, and either analyze them on site or download them back in the classroom for further analysis and comparison. Many teachers have created environmental science labs that use these tools to collect information about the pH of soil in the schoolyard as a proximate measure of acid rain or about the dissolved oxygen in a nearby pond as a measure of the pond’s health. These labs provide authentic inquiry through the use of real data, tools, and locations that matter to the students. Students explore science in a physical and geographical context and experience connections between science and their world. However, the questions students might investigate in these activities are constrained by the actual environmental health in the area. The actual inquiry may be somewhat trivial and limited if there is nothing anomalous to explore. From the perspective of designing a learning activity it may be highly desirable to subtly (or dramatically) disturb the environment to provide the students with a rich situation to investigate. For a variety of obvious reasons, such real perturbations are not practical, except in rare cases. Augmented reality simulations can combine the positive features of simulations and probeware to nearly approximate the case of being able to explore experimentally perturbed environments. Such simulations could provide the benefits of explorations in context, linking science and the students’ surroundings, with the nearly limitless inquiry potential of simulations by (a) tying a more broadly applicable intellectual experience to a core disciplinary dilemma and scientific practice, and (b) using computational media to help students appropriate their real surroundings for authentic simulated investigations.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

377

In particular, we tried to use the Pocket PCs’ multimedia and simulation capacities for interactive storytelling, creating contexts wherein learners would experience a story that could become a narrative to think with in the study of science (cf. Schank, 1994). Pocket PCs, which can display video, text, and host webs of information in intranets, can create virtual worlds that go beyond just presenting data by providing narrative context similar to problem-based learning or anchored instruction environments. Leveraging design techniques from role-playing games (cf. Gee, 2003), we investigated if augmented reality simulations could entice learners into complex scientific practices through adopting the personae of scientists. We hypothesized that opportunities existed for immersive gaming environments to recruit players into assuming new identities as environmental investigators, scientists, and environmental activists, thereby encouraging students to adopt ways of thinking that might be ideal preparation for future learning. Augmented reality applications hold particular promise in disciplines such as environmental engineering, in which spatial and contextual information are core components of professional practice. In authentic field studies, such as investigating and remediating toxic spills, spatial information about the distribution of the spill and location-sensitive information about the spill’s proximity to other parts of the environment are central to conducting an investigation. However, the investigative process, sampling strategies, and remediation strategies are all mediated by social factors (cf. Dorweiler & Yakhou, 1998).1 Students often have difficulty recognizing the situated nature of environmental engineering investigations and learning to act effectively within the many constraints (Nepf, 2002). Yet these constraints and the ability to adapt to them are key disciplinary practices that are manifest in several distinct ways. First, environmental investigations are affected by resource constraints. The amount of time, money, equipment, and human power available affects what strategies are feasible in any given context. Second, the physical particulars of the research context drive an investigation, and research goals are often reprioritized in relation to local context. For example, discovering a lethal toxin in groundwater in close proximity to a major source of drinking water might be cause for reevaluating a research approach, whereas a similar toxin in another location that does not use groundwater for drinking would not be. Third, there is an interplay between desktop research and collecting field data. In some cases, a knowledgeable informant can save investigators time and money by pointing investigators to probable culprits. Finally, social constraints affect both the investigative process and remediation strategies, as investigators need to manage how their work is perceived by others (particularly the press). Investigators need to avoid generating unwarranted public alarm or, in some cases, generating bad press for clients. A few environmental educators have begun exploring Thanks to Heidi Nepf, hydrologist and toxicologist at the Massachusetts Institute of Technology, for helping us understand these factors.

378

SQUIRE AND KLOPFER

how immersing students in problems based on current events might serve as useful pedagogical models in environmental and chemical engineering to address some of these issues (cf. Dorland & Baria, 1995; Patterson, 1980).

CONTEXT This study examined the implementation of a particular augmented reality simulation, Environmental Detectives, in three different university classes and one high school class. We deliberately chose a wide range of classes in order to see how learners with different backgrounds and affiliations toward science would react to this experimental program. As such, the study was designed to illustrate the range of possible enactments of the program, rather than generate strict comparisons. This study was a part of a larger design research agenda (Collins, 1992) exploring the potential of augmented reality for supporting learning in environmental education. Environmental Detectives is an augmented reality simulation game for the Pocket PC developed by the investigating team using the Microsoft .NET compact framework. Environmental Detectives was designed in consultation with environmental engineering faculty and is matched to scientific inquiry learning goals in advanced-placement-level science, making it possible for use across high school and college courses (with teachers choosing to appropriate it in different ways according to the contexts). Curricular Goals and Framework The curricular goal of Environmental Detectives is to give students an experience of leading a complex environmental science investigation so that they can understand the socially situated nature of scientific investigations. The game scenario was designed in consultation with two environmental science faculty and designed around a core dilemma of environmental science: how to conduct effective environmental investigations within social, geographic, and temporal constraints. This scenario requires students to (a) develop sampling strategies, (b) analyze and interpret data, (c) read and interpret scientific texts to understand the problem, and (d) ultimately design a viable remediation plan for core constituents. Scientific investigations are frequently presented to students as closed-ended problems with one right answer that can be solved linearly (cf. Zolin, Fruchter, & Levitt, 2003). Conversely, scientists in the field continuously frame and reframe the problem in response to budgetary and time constraints, local conditions, and what is known about the problem. As an example, researchers design sampling strategies in relation to the chemical and physical properties of a toxin, its potential health and environmental effects, legal issues surrounding its spread, and local conditions, such as nearby waterways and impediments to sampling (i.e., human-made physical struc-

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

379

tures or waterways). Consistent with efforts such as the Problem- Project- ProductProcess- People-Based Learning Laboratory at Stanford University (Fruchter, in press; ), our goal was to immerse students in complex problem spaces where they would draw on diverse resources, design creative solutions, and work across complex distributed environments in solving problems.

Environmental Detectives In Environmental Detectives, participants work in teams of 2 to 3 students playing the role of environmental engineers investigating a simulated chemical spill within a watershed. In the university implementations, the watershed is surrounding the students’ university, including a nearby river, whereas for the high school students the watershed is associated with a working farm located within a nature center. The high school class regularly takes field trips to the nature center, thus making it the best proxy environment comparable to the university campus. Both real-world watersheds include streams, trees, and other natural elements that are then augmented by a simulation of an environmental disaster: in this case, a toxic spill of TCE that can potentially contaminate ground and surface water. In the university case, further context was added concerning a recent construction project on campus, whereas in the high school case additional information was added concerning a possible state buyout of the farm at the nature center. Each of these additions was done to provide locally topical information, a hallmark of augmented realities. Moving about in the real world, the handheld computers (Pocket PCs) provide a simulation whereby students can take simulated sample readings, interview virtual people, and get local geographical information (see Figure 1). The spread of TCE is simulated on a location-aware Pocket PC, which, equipped with a global positioning system (GPS) device, allows players to sample chemical concentrations in the groundwater depending on their location. For example, a player standing at Point a, near the source of the spill (see Figure 1), might take a reading of 85 parts per billion, whereas a student standing on the opposite end of campus (Point b) might take a reading of 10 points per billion. Players are given three reusable virtual drilling apparatuses that they can use to drill for water samples. After drilling for a sample, players must wait 3 min for the drilling to complete and an additional 1 to 3 min for a sample to be processed. These waiting periods were designed into the game to simulate actual temporal constraints. This limits students to collecting only three samples at a time, driving them to develop sampling strategies to optimize the amount of territory that they can cover within their limited time. Environmental Detectives contains a multimedia database of resources that students can access to learn more about the chemical makeup of TCE, where TCE is found on campus, the health risks associated with exposure to TCE, how TCE flows through groundwater, relevant Environmental Protection Agency regula-

380

SQUIRE AND KLOPFER

FIGURE 1 (Left) A screenshot from Environmental Detectives. The red dot indicates the player’s current location and is guided by real-world position as supplied by a global positioning system device. The pink markers represent locations of interviews, whereas the blue markers show where the player has already sampled the water. (Right) Some of the textual resources that players can uncover.

tions regarding TCE, remediation strategies for cleaning up TCE, and the political and economic consequences of Environmental Protection Agency violations on campus. Students access these resources by obtaining interviews from virtual experts located at various points around the campus in locations roughly corresponding with actual operations. That is, an expert on hydrology would be near a building where that topic is studied, and a character with records of where chemicals are used would be located near an office that performs these functions. Because there is not enough time to interview everyone or to drill more than a handful of wells, students must make choices between collecting interviews, gathering background information, and drilling wells, adjusting and reprioritizing goals as new information becomes available. In addition to simulating an environmental investigation within complex socially situated settings, Environmental Detectives is designed to leverage the affordances and conventions of computer gaming to intellectually engage students in complex problem solving by providing a safe realm for experimenting with new ideas and new identities. Whereas in authentic environmental engineering investigations (or learning-by-apprenticeship models) students’ failure might result in damaged professional reputation, a waste of public resources, or, in a worst case

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

381

scenario, human illness or death, games and simulations allow students to enact strategies in a pedagogically safe space where failure is possible, if not expected, and players are encouraged to experiment with new ideas and identities. To be successful in Environmental Detectives, students must combine both the real-world and virtual-world data to get to the bottom of the problem. The precise location of the spill is unknowable to students, and there is no one perfect solution to remediating the problem; each solution involves political, financial, and practical trade-offs that must be considered. Consistent with problem-based learning frameworks (e.g., Barron et al., 1998), students use their handheld computers as tools for gathering firsthand data about the location and severity of the spill, and as a resource for accessing archives of information about toxicology, hydrology, similar cases, and local environmental conditions. Although each participant chooses his or her own path through the informational and geographic landscape of the game, the following describes what a typical player might experience. By design, Environmental Detectives starts with a statement of the problem (the potential contamination of a local water supply with a chemical) that should provoke questions about the geographic extent and intensity of the problem (determined by collecting primary quantitative data) and the history and future ramifications of the problem (determined through interviews with experts). A team of players (2–3) in the game might start by walking from the initial briefing location (where all players receive an orientation) to the site of the initial reported measurement (which may take 5–10 min) and taking a measure there by drilling a virtual sampling well. After getting that reading back (reported as a unitless number, e.g., “40” rather than “40 parts per billion”), they may seek an expert who could help explain that reading. Along the way to that location, the players might take additional samples (by drilling wells) along some transect to try to determine a trend in the samples. After getting information on what units the readings are reported in and thus their significance, the players could decide to seek information from other experts on health or legal ramifications of the toxin, or perhaps investigate from where the toxin may have come. They would also need to return to the geographic site of their sampling wells to retrieve the readings from those locations. This process ideally would be iterated, taking a planned array of samples and interviewing the experts to determine a course of action. This plan is complicated by the physical barriers (bodies of water, fences, etc.) and geographic information (terrain, tree cover, etc.) that the players gather as they experience the real environment around them. This version of Environmental Detectives takes 2 to 3 hr to complete, including introduction, game play, and debriefing, although a teacher might extend or shorten the game in order to meet his or her classroom needs. This time period was chosen for a combination of pedagogical and practical reasons. We wanted the simulation to place realistic constraints on the activity for the students so that they would have to make decisions about their actions. Students often approach such

382

SQUIRE AND KLOPFER

situations with the idea that they need to get all of the information in order to come up with a reasonable solution. This is enforced by many school-based activities in which it is the expectation that students will know all of the information. However, in real situations it is not practical or perhaps even desirable to spend the effort to know everything about a system. It is too resource intensive, or sometimes scientifically impossible, to obtain complete information. Thus, constraining time forces the students to make hard decisions about what they can find out and challenges them to analyze and judge the information that they have to come up with what they define as the best solution. There were practical considerations as well. This was an activity that needed to be planned out of normal classroom time (for both of the audiences that we address below). In surveying audiences with which we might work, the 2- to 3-hr window was what most professors and teachers advised would be the right time that they could arrange outside of the normal classroom. The simulation is designed to be flexibly adaptive so that teachers might easily add extension activities (such as exploring the properties of TCE, the health effects of TCE, hydrology, water treatment plans, or similar cases) or remove activities as local conditions suggest (see Squire, Makinster, Barnett, Barab, & Barab, 2003). For example, some of the university classes drew parallels to similar engineering studies done on toxins in the area or further analyzed the research methodology applied during the investigation. Similarly, the high school class engaged in further reflection on chemical properties of the toxin and further analysis of the watershed in which the investigation took place. Participants In the first phase of the project, we examined Environmental Detectives in three courses at a private technical university in the eastern United States. One course was a freshmen environmental engineering course; the other two were sections from an undergraduate scientific research and writing course, each with 18 to 20 students. In both contexts, the game was used to introduce students to issues around conducting real-world environmental investigations and was used as a prelude for a larger research project. All three classes were 2 hr in length. This article reports findings synthesized from these classes, with the focus on a small number of teams from two of the classes. These teams were intended to be representative of the range of student experiences (including those who successfully engaged in the necessary practices and those who struggled). Findings from the other course are reported elsewhere (Klopfer & Squire, in press; Klopfer, Squire, & Jenkins, 2004). The second phase of the project took place at a nature center in an East Coast metropolitan area and involved an environmental science class of 18 high school students. The session involved roughly 20 min of introduction time, 90 min of game play, and 20 min of debriefing. The pedagogical goals of the game were developed with nature center educators interested in engaging students in more ro-

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

383

bust activities than traditional field trip scavenger hunt exercises. They hoped that Environmental Detectives would encourage students to interact with the environment, geography, and history of the site as well as participate in domain-based problem solving. We chose this group because we wanted to see how students from a nontechnical background would respond to the activity. In particular, we were interested in examining how nonengineering students would use the technology, balance the driving problem behind the curriculum, and construct the problem of understanding toxic flows. Here we primarily focus on two groups as case studies but also include information from other groups and the entire class debrief. The teams we chose to focus on again represent the range of experiences demonstrating more and less successful problem-solving strategies. Although the specifics of the problem were adapted for the nature center site, the scenario was essentially the same and involved the same information and subject matter, making the scenario and experience comparable to those of the university classes. An overview of the participant populations are shown in Table 1. Although a small number of teams were selected for case studies, all students participated in the pre-/postsurveys and presentation of cases.

METHODOLOGY In this study, we used a naturalistic case study methodology (Stake, 1995) to gain a holistic view of the activity that unfolded during game play, understand how learning occurred through participation in these activities, and remain responsive to unanticipated issues that might arise during the research. Because we were interested in accounting for student–computer, student–student, and culture–student interactions, we employed quasi-ethnographic techniques designed to capture student actions at the molar level (Goodwin, 1994). Capturing an ecology, including the many tools, resources, and social structures that characterize any particular context of activity, is challenging and is still being negotiated in educational research (Engestrom & Cole, 1997). In describing a situation as a unit of analysis, Cole (1995) concentrated on practice, activity, contexts, situations, and events. We used narrative case studies to provide a broad flow of events that take each of these factors into consideration (cf. Hoadley, 2002). We also used discourse analysis (Gee, TABLE 1 Distribution of Study Participants Variable Number of classes Total participants Team case studies (students) selected

University

High School

3 58 3 (8)

1 18 3 (7)

384

SQUIRE AND KLOPFER

1992) to examine more closely how students constructed and framed problems and to study relations between class discourses and students’ scientific investigations. Specifically, we investigated (a) the practices students engaged in while participating in Environmental Detectives (i.e., how they integrated real and virtual data in problem solving and conducting their scientific investigations), (b) how students constructed the problem (e.g., as well defined or ill defined, authentic or inauthentic), (c) how investigation in the physical environment mediated students’ inquiry, and (d) what instructional supports were useful in supporting learning. Data Sources

Observations. Four trained researchers attended each session, and a trained researcher followed each student team during the game, videotaping a subset of the teams and documenting student practices in field notes. Consistent with other researchers studying problem-based learning environments (e.g., Barron et al., 1998; Nelson, 1999), we paid special attention to student discourse, examining how students framed the initial problem, constructed goals of the activity, negotiated information in groups, planned activities, and developed shared understandings. The text selected here for analysis was chosen because it was representative of typical dialogue across a range of responses. We used informal, nonstructured interview questions during the exercise to confirm observations, clarify students’ goals and intentions, and learn more about students’ handheld-mediated activities. Although the researchers were clearly participant observers in the activity, they attempted to remain unobtrusive whenever possible. Interviews and artifacts. We also conducted a 20-min focus group and exit survey to probe students’ experiences in depth to document their thoughts, feelings, and attitudes toward the experience. We also recorded students’ inscriptions, physical gestures, and interactions with the Pocket PC. Additionally, we gathered and analyzed data emerging from students’ off-computer activity, including written inscriptions some teams used to plan their investigation (cf. Roth, 1996). Data Analysis Two researchers viewed and analyzed all researcher field notes, videotapes, and students’ projects using the constant comparative method (Glaser & Strauss, 1967) to generate relevant themes from the data. Consistent with Stake’s (1995) responsive method, we paid special attention to unexpected and unintended consequences, given the exploratory nature of this research. After each round of videotape viewing, we developed emergent hypotheses, reexamining and refining these hypotheses as we watched subsequent tapes looking for disconfirming evidence or counterhypotheses. We then wrote several case studies from both the university

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

385

and high school parts of the project to capture the key events or turning points in students’ thinking. Two of the university case studies are included here (although we also include short excerpts from and mention of other teams, as well as information from an additional case study in Klopfer, Squire, & Jenkins, 2004). The cases are intended as a means of conveying a flavor of activity and providing the reader with a basis for generating contrary interpretations of the activity. Two case studies from the high school participants are also included (a third additional case is reported in Klopfer, Squire, & Jenkins, 2004). In the high school cases, we focused specifically on the discourse of the teams, as well as the presentations and debriefs in order to understand how students framed the problem and generated meaning in situ. Given our observations with the university students in this study—that a driving contradiction existed between the dual needs of doing desktop research and collecting samples—we decided to focus on this issue in greater depth in this part. For each of the case studies we provide the synthesis of a discourse analysis (roughly 15 pages per team and not all included here), an analysis of how language “enacts activities, perspectives, and identities” (Gee, 1999, p. 4–5). Researchers transcribed the interactions of teams that were representative (typical) of talk across the range of successful and unsuccessful teams. Consistent with Gee (1999), we focused on how language—specifically word choice, cues, syntactic and prosodic markers, cohesion devices, discourse organization, contextualization signals, and thematic organization in language—created the activity. Essentially, this analysis is toward understanding meaning, how it is made, enacted, and represented in situ. We specifically looked for moments when meaning was negotiated and shared understandings were mobilized to solve problems, and when meanings generated further action. Specifically, this methodology allowed us to gain insight into how participants framed the problem, constructed the reason behind the activity, and negotiated problem-solving strategies in situ (e.g., Barab & Kirshner, 2001).

RESULTS The following case studies describe the results of our design experiment. We start by describing an illustrative example, a case study of a typical team. In this first case we outline the process of their investigation as they notably engaged in (a) privileging quantitative data; (b) framing the problem as a unidimensional one of “tracking down the toxin to its source” as opposed to a multidimensional problem involving probable cause, potential health and legal effects, and suggested remediation strategies; (c) integrating prior knowledge of the environment with students’ reasoning; (d) creating emergent sampling strategies, such as triangulation; and (e) “voting with their feet” as they decided which problem-solving path to pursue. We

386

SQUIRE AND KLOPFER

then present contrasting case studies of Environmental Detectives in action and focus on how the activity unfolded across teams. In particular, we examine how students constructed the activity and then use a discourse analysis method as a basis for showing how the activity was constructed in different settings. We hypothesized that augmented reality simulation games would be a potentially powerful emerging medium for education in contextual settings.

University Case Studies After a classroom briefing introducing the problem, students met at the center of campus and learned to use their GPS and Pocket PC. Most teams immediately drilled a sample and then picked a direction to move to, based on their theories of where toxins might have originated, concern for downstream consequences of the toxin’s spread, or, in some cases, just random guessing. Teams generally negotiated where to take the second sample from; in some cases, a team leader, usually the person with the Pocket PC, would lead the way. Across all teams, participants frequently negotiated and debated where to go (as evinced through their talk below). One team of three students, tailed by a researcher, headed up away from the river and toward campus. One of the students inquired, “How many samples do we need?” It was not clear whether the question was addressed to the researcher or the rest of the team, but no one responded. The student holding the Pocket PC had previous GPS experience and started to guide the team. He drilled for one sample and then walked to nearby locations to take two more samples, the maximum number of concurrent samples permitted. He chose a triangular configuration, though when another student asked why he chose this arrangement, he cited no particular reason. Students retraced their steps as they waited for the required 3 min between sample drilling and reading. Finally the sample was retrieved. The reading was 88. Another student asked if 88 was good or bad. One student hypothesized that the number could be a percentage, but no one could answer definitively. They decided to collect more data. As they walked to collect their two drill rigs (used to take samples), a student not holding the Pocket PC asked what the data looked like. The student with the handheld described their current readings by pointing to three locations in physical space (as opposed to showing on the handheld) and citing the readings. Students again debated the meanings of these readings. One student hypothesized that the readings were in parts per million. The student holding the Pocket PC suggested that they should go toward the “higher numbers,” pointing into the distance. They walked several hundred yards through several buildings toward the higher number and placed more drills.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

387

This pattern of drilling to find the source before considering the meaning of numbers was similar across teams as suggested by this exchange, taken from another team: Lisa: Ben: Lisa: Ben: Lisa: Ben: Mel: Ben: Lisa: Mel: Ben:

The reading is 4. It’s obviously good. Come on now. I don’t think it is good. It’s obviously good. Four. Like 4 is a bad reading. Like 4 on a scale of 1 to 5. Four is real bad. On a scale of 1 to 50 though, 4 is pretty damn good. True, but what is this scale? We don’t know that. We don’t know that. We have no idea. It could even be that the top one is the best. Ok. So we need to dig another well. Let’s get this one first [referring to an already dug well].

Most teams initially constructed the activity as a pattern recognition search for the source of the toxin, opting to drill more samples to define some pattern rather than consult documents or experts who could definitively tell them what levels were of concern, as they were informed at the onset of the activity. They avoided conducting the desktop research that environmental scientists describe as critical to these investigations. This exchange, which was typical of most teams, also reflects the amount of negotiation and debate behind sampling strategies. Most teams (typical of prominent discourse patterns in the class and in the institution) were argumentative in thinking through results (e.g., “Come on now”). After several minutes, the readings from this second round of drill placements returned from the lab. One student noted that the new readings were very high in one direction. They walked in the direction of the higher readings, as if following a trail or scent, pausing briefly to interview a virtual staff member in environmental policy, who happened to be nearby. The interview yielded little information, but it did reveal that they could conduct a second interview with a TCE supplier from facilities at a new location across campus, which they needed to visit within the next half hour because the informant was leaving for another meeting (this event was then “triggered” on their Pocket PC). They decided to immediately go to the new building although there was no discussion about what information they hoped to find, or hypothesizing its anticipated value. Along the way they looked at the emerging gradient and one student hypothesized that the concentration was likely to be higher on the other side of the building (the one they hadn’t visited yet). The second interview revealed where TCE was used on campus, and the student holding the Pocket PC summarized the information for the others. Meanwhile, the team took another reading. One student (not holding the Pocket PC) realized that the highest concentration appeared to be surrounding one building and suggested

388

SQUIRE AND KLOPFER

that they should drill more wells there. Another student dismissed this idea, assuring them that they had already sufficiently pinpointed the source of the leakage to that building. Using the information from the toxin supplier combined with preexisting knowledge of the activities near that building (the university machine shop is located there), he correctly identified the source of the toxin and suggested that they obtain interviews to help interpret their data. It is worth noting that although they had spent nearly 50% of their time already, the team did not know what units the readings were in (and indeed, one student hypothesized incorrectly that they were a percentage), what levels of TCE were dangerous, how likely the TCE was to spread throughout the environment (including into a nearby river), or what legal repercussions the university might face if the TCE were to leak off of university property. Most teams (all but 1 or 2of the approximately 12 teams that we studied) had similar problem-solving strategies, although 1 team, notably, stopped at a computer and used Google to find a good deal of information on TCE (which was applicable in this simulation that used realistic data). Seeing another interview nearby, they headed in that direction. One student noticed the time and paused, causing the team to stop. He suggested that they use their last 15 min wisely. The Pocket PC changed hands briefly to a different team member but was quickly returned to the student who had held it most of the time because there was some confusion as to where they were headed next. After several minutes of circling the building, they finally accessed the interview, which explained how groundwater flows through campus. Here they learned that the groundwater was not used for drinking. As the students headed back to class, they discussed the implications of their findings. Reviewing their documents, they learned that planting trees could mitigate some of the effects of TCE. One student looked at the building where they hypothesized the toxin had originated and then back at the river, declaring that by the time the pollution gets to the river the pollution is likely to be highly reduced (although they had no concrete evidence on which to base this assertion).

Debriefing. Each team presented their findings before the class. This team, like most, had pinned the location of the spill down to a particular building based on following a gradient that they had observed (which was correct) and theorizing that the spill had come from the machine shop. They argued that the spill was not a problem because the groundwater was not a source for drinking water, and the river was too far from the source of the pollution to be a problem. They recommended planting trees to mitigate the problem and monitoring the situation over time. They noted that this solution would cause little alarm in the community and would not destroy the only grassy area on campus. Cross-team discussions. Most of the 12 teams that we studied made similar findings. Most relied heavily on sampling, and roughly 75% of the teams accurately

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

389

determined the location in the time allotted. Most teams also suggested the politically expedient answer of planting trees and monitoring the situation because they saw no immediate legal or health threat. Only 3 teams, which all focused on collecting interview data, correctly surmised that regardless of whether the spill was an immediate health hazard it was a legal threat and should be cleaned up to avoid Environmental Protection Agency fines. Students across all the teams were very sensitive of the political ramifications of falsely calling too much attention to the problem, given the fact that Building 3 was centrally located on campus. This concern about unduly drawing negative attention to the university was introduced in the cover story but eagerly taken up by players as a driving factor behind any solutions. Successful teams gathered both samples and interview data, recursively examining what was known, reframing the research questions, and gathering new data. For example, the following description from another team began as they were taking their second reading. Instead of immediately trying to pinpoint the precise location of the spill, they located an interview with a faculty member to tell them more about TCE while they waited for lab results: Jenny: It just said that the results of the lab said “30,” so it might be 30 parts per cubic feet. Steve: That is not as bad as the military base in Cape Cod, so just remember that it can be nasty or something. [summarizing the text from the interview to the team]: So what do you want to know … TCE is found all over place … a spill in Illinois … So how fast does it move? Depends on the soil and whatnot, 1.5 to 7 feet per day, ooooh… [Repeating aloud, 1.5 to 7 feet per day. Bill writes down the numbers.] Jenny: That’s not a concentration. Bill: It doesn’t sound like a rate of flow; it’s a rate of spill. Steve: Well it just said the result from the lab is 30 so it might be … 30 cubic feet … I don’t know. Bill: Cubic feet per day doesn’t make sense either. It ought to be a … rate of spill. Steve: [continues summarizing] We need to build a model of how TCE moves through the groundwater … lots of things to take into account … You have a certain mass of stuff that’s been spilled, and it’s covering a larger and larger region every day because of spread…. As a rule of thumb you might assume that it spreads at a rate of 150 feet per year. Bill: Whoa, whoa, whoa per year? Steve: Per year. Bill: Ok, 150 feet per year. [Writes the numbers down] Ok. So [pausing to think], decaying at about half of its concentration. So if you start with 100 parts per billion that’s per ... 50 parts per billion at 150 feet per year. The 30 and 70 could be possible.

390

SQUIRE AND KLOPFER

This interaction, although less representative of what occurred, shows a more productive intellectual interplay between primary and secondary data sources. Right away they queried the meaning of the readings, speculating that it could be “parts per cubic feet,” a concentration that they noted compares favorably to the readings found in Cape Cod (a case study they had learned about via their interview). They also read that they need to “create a model of how TCE moves through the groundwater,” and subsequently compared their readings to the data they were presented from the case study. Next, Jenny added that they could use this data to pinpoint the source of the spill, but new information about phytoremediation (a process of planting trees to remove toxins from the soil) changed the topic: Jenny: The 30 should tell us something about the source of the spill being closer to the … Bill: Yeah, I think so. Steve: [reading] What should we do about remediation TCE? Planting trees, phytoremediation. Bill: [writing, reading? aloud what he writes] Plant trees to suck TCE out. So we’re nervous about the effects of TCE on the environment. We don’t know that TCE is, like, infecting trees. We have higher readings, which is contradictory to … Steve: Higher reading where there’s less trees. No more trees. Jenny: Because it’s sucking the TCE out. Steve: Hold out there’s more information. It’s expensive and you could get water treatment part … something about backyard. [As Steve finishes reading, the team begins walking.] Jenny: Pumps are the best because trees don’t do anything. Here, introducing the concept of phytoremediation did two things: First, it made the team realize that the existence of vegetation could be affecting their results. Second, it introduced the issue of “what to do about TCE.” Unlike other teams, this team realized that phytoremediation is a partial remedy at best. However, the team also realized that they knew very little about TCE as a chemical, its health effects, or what might have caused this spill. Bill began by suggesting that they drill more samples, but the team realized that this would not help them learn about TCE. They went back and forth between querying one another on what they knew, what they needed to know, and where they might find the information: Bill: We could just start digging holes to get more information. Jenny: Since we’re going to the classroom, let’s ask Eric [one of the in-game characters whom they can interview].

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

391

Steve: What is he going to know about TCE? Jenny: Who knows … Bill: We never learned what TCE is at all, did we? We have no ideas what its effects are on the environment. Jenny: Trees suck it up. Steve: For all we know, TCE is just another form of water particle—we don’t know that there’s anything bad about it. Steve raised a critical point here; the team was not exactly certain why TCE is a dangerous chemical. They knew that there have been other spills, and they knew that trees absorb some amounts of it, but they were not sure in what form or concentrations it is actually dangerous (if at all). In the next exchange, Bill connected these concerns to his existing knowledge of the Charles River: Bill: We know that it’s in the Charles, which is already disgusting. It’s possible that TCE is such a ridiculously small effect compared to the big mess of the Charles, and I have friends by the way who study the Charles River and are not impressed. So, that’s a possibility. We also know that the water isn’t used for drinking … Jenny: We used to go canoeing on the Charles River. And we always had to watch out. People fell out of their canoe their eyes were stinging and stuff. This exchange illustrates a common phenomenon in augmented reality games. Facing gaps in their knowledge about chemicals, health effects, or the history of their local space, players would frequently begin taking what they already knew (or thought they knew) about the environment (in this case, the fact that the Charles River is polluted and not used for drinking) and applying it to the problem at hand. Given the importance of activating prior knowledge in learning for deep understanding, this tendency to build connections between the game space and their existing, lived knowledge of the space was encouraging. In the debriefing, this team made the case that there were significant concentrations of TCE in the groundwater and it had been there for at least a few years, as evidenced by the size of the plume (the three-dimensional disbursement of TCE through the groundwater). They believed that it would soon be in the Charles River, but they were not sure of the precise health effects. They believed that it was a cause for concern and that some sort of pumping would be required to remove the toxin. They were one of the few teams to advocate cleaning the groundwater rather than “planting trees and monitoring the situation.”

392

SQUIRE AND KLOPFER

High School Case Studies Most college students had framed the problem as one of collecting samples to obtain the one correct solution of where the spill occurred, as opposed to an investigation into a socially situated problem. However the problem-solving approach tended to differ among high school participants. As such, we focused the subsequent investigation of high school usage on weighing the potential value of interview data in context with the quantitative data. Also, given the broader audience in these cases, we paid additional attention to the quantitative reasoning applied by the high school students to understanding the patterns in the data. In these case studies, we focused more specifically on team talk to examine the processes by which the problem was framed. Across the teams we examined, four main motifs emerged in the talk: (a) negotiation of the environment in the investigative process; (b) within- and intergroup interpretation of the problem as gathering information to complete a puzzle; (c) discussion and problem solving that integrated the physical world, paper-based resources, and personal digital assistant (PDA)-mediated resources; and (4) emergence of intergroup power dynamics. This section reports results primarily from two teams, which were chosen to represent the ends of the spectrum of responses. One team (Team 1) struggled with making sense of the quantitative data patterns as well as integrating the quantitative and qualitative information. The other team (Team 2), although unable to fully address the problem, demonstrated significant success in finding patterns in the data and identifying where additional research was needed. In this section we use a brief discourse analysis to examine emergent learning practices. In the following passage, Team 1 discussed the best method for reaching an interview with an expert who was in the horse farm. Several physical structures entered into their thinking: 1. Stacey: There’s a fence there. I can’t get over it. 2. Gina: Then I don’t know what we’re going to do. We’re stumped. Let’s call the guy [facilitator on the walkie-talkie] so we can find out what we’re doing. 3. Stacey: What does it look like? 4. Gina: We’re close. That’s the thing. 5. Stacey: Ok, fine. Can we go over this [barbed wire] fence? 6. Gina: I don’t know. 7. Stacey: Maybe we can get on the other side by walking somewhere else. 8. Louis: Maybe we can walk the fence. No, there are trees. Environmental constraints and affordances immediately had an impact on students’ problem-solving process. The constraints of the environment, namely fences (1,5,8), barbed wire (5), and trees (8) guided their problem-solving path. All

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

393

of Gina’s statements were declarative, assessing their progress and directing activity, whereas other students raised ideas as suggestions, couching them with qualifiers (i.e., “maybe”). The problem was about designing strategies in relation to local environmental affordances. Roughly 10 min into the activity, students in Team 1 had negotiated the particulars of the environment, with Gina having taken a lead in defining team work. They had conducted their first virtual interview and now met another team (Team 3), who asked them how many interviews they had gathered. A shared understanding emerged whereby the point of the activity was framed as collecting “boxes” (the screen icons that correspond to virtual interviews), akin to a scavenger hunt: 9. Girl (Team 3): 10. Louis: 11. Stacey: 12. Girl (Team 3): 13. Boy (Team 3): 14. Gina: 15. Boy (Team 3): 16. Gina: 17. Boy (Team 3): 18. Gina:

How many [interviews] did you get so far? None, nothing. We’ve only gotten one box. How many have you got? One so far. We were going for another one. Three. Oh. You meant the boxes? Did you dig? Yeah. Can you dig anywhere? Yeah. I think so—I did. Cool. We got an interview. That’s all we did. We don’t have much time. We have to go.

The girl from Team 3 initiated the conversation by asking “how many they got so far,” framing the problem as one of collecting the most interviews as efficiently as possible and establishing the activity as one of collecting “boxes.” Gina turned the topic to digging, but Team 3 offered little information on what they had dug. Gina did not pursue the conversation and declared that the team was running out of time and needed to go. Shortly afterwards Team 1 set out in pursuit of an additional site at which to dig. Along the way they discussed the readings that they had received thus far: Gina: So we’re digging a well at 144 … [reading the coordinates] Stacey: And we’re near the chickens. [writing down notes] Gina: Sample sent to field lab. What does that mean? Stacey: Is there a location? Gina: Oh no, what did we do? [Pocket PC sound effect indicating a sample is available] Gina: Reading at whatever is 27. What does that mean? Reading at 140 94 is 27. Whatever that means. Stacey: So I’m just going to say we dug a well at this spot and it was 27. Gina: Yeah.

394

SQUIRE AND KLOPFER

Stacey: Well, actually it was that spot. [pointing to where they dug] Gina: Yeah. And it got sent to the field lab. Here we see that they were collecting additional data, but the incoming data were interpreted merely as a stream of numbers. The students did not relate the readings to previous readings or to the spatial arrangement of the readings. Here they also did not know what the numbers mean, but they did not identify that they needed additional resources to ascribe this meaning. We compare this with Team 2 upon receiving their first data: Abbey: The reading is 10. Maya: Ok. Abbey: We got a well reading of 10 so now we should find someone who can tell us what that means ‘cause we don’t know. Team 2 immediately identified that they didn’t know what the readings mean and should find someone that could help them with this interpretation. They went in search of a virtual character who could possibly give them this information. A while later they got the interview they were looking for, and one of the researchers asked them about what they got out of that interview: Abbey: Information about wells and sending water samples to labs. But we need to get information about reading, like what the reading means. 10 … I have no idea what that means. They were able to identify that this interview gave them helpful information, but it didn’t give them the information that they needed to provide some absolute meaning to the reading of 10. A while later they obtained an additional data sample from a virtual well: Abbey: Reading 56! That’s a lot higher than the 10. Ok … Maya: 15 [announcing a third new reading] Abbey: 15. So it seems to be closer, higher when we’re near the water. And we’re at a higher elevation here too. Do you want to head back? Here Team 2 had obtained and interpreted information based on three readings (10, 15, and 56). They understood the geographic relationship of these points, such that the one that was closest to the water source was the highest (having the highest concentration of TCE). Additionally, they looked at the physical landscape, showing that the readings were highest at a high point on the landscape, which may have had implications for where the water flowed.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

395

As the teams worked their way toward the end of the project, they tried to interpret their findings and decide what they were going to say when they presented their recommendations and evidence. On the way back, Team 2 met up with another team (Team 4), and they discussed what they had found: Abbey: So what do you guys think, you know? About buying the land? Nick (Team 4): I didn’t really find any overwhelming evidence that there is TCE or any other toxic chemical. Abbey: Did you say you took just one reading? Nick: Yeah. Abbey: What did you get? Nick: 173? Abbey: It has a question mark. Wait, what does that … Where did you? [interprets the information in contrast to her own computer, and perhaps determines that although Team 4 has drilled one well, they have not taken a sample from that well, thus they have no real data] Brett: I’d buy the property, because there’s enough property … Abbey: But if these animals are getting sick ...? I mean, liver problems? Something’s up though. That librarian we talked to seemed disgruntled, didn’t she? Here we see Team 2 had collected more quantitative data and made progress in interpreting that information. Additionally they had obtained interviews from the characters and integrated that information as well (referring to the disgruntled librarian). For Team 4 there was no connection between the disjointed (to them) information. Similarly, we pick up Team 1 as they headed back to the lab with the information that they had collected. They continued to struggle with making sense of the data: Gina: I am so happy that we have at least one box. Louis: Yeah. Gina: And we have that it is the TCE chemical. That is what they think it is, so we have something to say. I am quite happy about that. Although they are happy about having collected their information, they had not been able to connect any of the pieces, either during the investigation or as they attempted to offer summative explanations. The teams gathered at the end of the experience to give their recommendations and share evidence to the entire group. The teacher selected some teams to make

396

SQUIRE AND KLOPFER

oral presentations. One of the members of Team 4 (whom Team 2 had run into before coming back) presented their recommendation: Nick: Ok, I think that the state should buy the land because there’s studies that shown that if there’s TCE here it can be cleaned up effectively. TCE and CT (Carbon Tetrachloride) and whole bunch of other chemicals can be cleaned up effectively. So I see no overwhelming reason not to buy it because the problem is solvable. He made this recommendation without knowing anything about the data indicating what was actually there, but surmising that whatever it was could be cleaned up. When asked if they had found any significant amount of any chemicals, he commented, “No, we didn’t … Large amounts of TCE … I guess it’s only harmful if it’s large amounts or large exposures to it …” Their teacher then asked them if they could define what “large” was. Nick responded, “No, no. We know that large is just big.” When probed further for their evidence of where they learned that the problem could be solved, they cited a single interview: Nick: The librarians down at the library said that at Cape Cod … I guess there was a similar problem at the Massachusetts Military Reservation … and they cleaned it up. In Illinois also … Their case hinged on the recommendations of one interview that they had found. Subsequently the reliability of this interview was questioned by one of the members of Team 2: Maya: The only thing about the librarian is that she kept saying “I think ...” so it’s kinda like we weren’t sure if her information was exactly accurate. That was just something that I noticed. Here Maya, from Team 2, showed that she was reading deeply into the information that she had found, even questioning the language with which the librarian presented the data. Their more thorough analysis and interpretation became evident in their recommendations: Abbey: Ok, well, we didn’t really come to definite conclusion. We found readings. One of them was right over by the water right before you go into the tunnel and we got a reading of 50 [rounded from 56]. And our other ones were 10 and 15. So those ones are away from the water, but we couldn’t find anyone who could give us information about what these numbers mean, so we didn’t come to any conclusion, because we’re not

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

397

exactly sure how to interpret those, so we’re not sure if the land should be bought.

Their evidence showed patterns in the data and also identified the gaps in their information, specifically citing items where they needed additional information. Unlike other teams, Team 2 was very aware of the limitations of their knowledge and structured their recommendations accordingly. Team 1 did not participate extensively in the presentations but did describe some of their thought processes during the investigation: Stacey: It was kind of confusing at first. Gina: It kinda seems like you’re supposed to go on a certain path. ‘Cause it kinda seems like we took a reading and went to the next site, and it gave us information about taking readings. But we’d already done that. It’s just, we had to stop… Through this dialogue the team was indicating that part of their failure came from seeing the process as totally linear. They described their experience as one in which they followed a path through each of the different “points” in the game, as opposed to being a dynamic process that evolved over time as they collected more information and interpreted that information. How each team perceived the role of the physical environment varied greatly, contributing to their success or failure in the investigation. Those that saw the environment as a barrier, or that simply couldn’t incorporate the real surroundings, struggled, whereas those who could “read” their physical surroundings incorporated them with the virtual information that they collected to create a better response. Here we follow Team 1 as they used real maps, the actual environment, and the Environmental Detectives-based maps interchangeably. They had just collected an interview and were now about to get another one. The students were concerned that they did not have enough information to solve the problem adequately. We pick up the discussion as they decided what to do next: 19. Stacey: Let’s go to that one [pointing to the learning center]. We just traipsed through a field. 20. Louis: I like how he [the character in the video] was standing up there [pointing toward the house] and reading it. 21. Gina: Yeah, I know. 22. Louis: He got to stand at the house, and we had to stand in the water [in the field]. 23. Stacey: I know. I am so wet. 24. Louis: My socks are so wet.

398

SQUIRE AND KLOPFER

25. Camera: We should head back soon. 26. Gina: Yeah, it is 12:50. 27. Louis: How far away is the thing [the location of the debrief]? 28. Gina: Where do we have to go again? 29. Stacey: Alan Morgan Center? That is … 30. Louis: [looking around] Not around here. 31. Stacey: Right here [points at paper map]. 32. Gina: And we’re right here [points at pocket pc]. 33. Stacey: That’s not bad. 34. Louis: But we have to go through the tunnel. 35. Stacey: How are we supposed to make recommendations? 36. Gina: I don’t know. 37. Louis: Just read off of the information that we got. 38. Gina: I thought we could dilly-dally but we actually did work. 39. Louis: For once. Stacey initiated the conversation by suggesting that they go to the learning center, as the team was tired of “traipsing through a field,” which “got their socks wet.” Louis noted that their path back to the nature center would take them through the tunnel (34), a feature of the environment that earlier had been the cause of considerable discussion, as a group of birds had flown out and scared the team. Stacey noted their lack of information (they had located several interviews but had dug few, if any, wells) and asked the team how they were supposed to make recommendations (35). As in the other exchanges, Stacey queried the team for strategies and Gina gave the response (36). Louis (37) suggested that they just “read off their information.” Gina summed up the team’s dilemma: They had thought that the exercise would be relatively thoughtless—that they could “just dilly-dally”—but instead they “actually did work,” (38) which Louis agreed with (39). Students used maps (19, 31), the real environment (20, 30), and PDA resources as tools (32) for communicating. Later this team encountered another visitor to the site (clearly not from their class). The visitor asked what they were doing: Stacey: We’re trying to find if there are any toxins here. Do you know of any toxins? Visitor: Toxins. I don’t know of any toxins. Gina: It is in the game. I think it is all in the game. The members of this team were negotiating the reality of the situation. The one team member asked the visitor if he knew anything about toxins, as if the information were real and may be accessible outside of the game. It was the other team member who suggested it was probably just in the game.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

399

In another discussion, one of the members of Team 2 was pondering the dynamics of the game: Abbey: It would be cool if there were real people. You’ve heard of Sturbridge Village [a local living history museum]? They make candles and stuff. It would be cool if there were real people you could ask your own questions, you know? This statement seems to suggest that she understood the simulated nature of the environment, that the game enacted this situation just as actors do in a mock historical site and that the students had some interest in expanding the experience beyond the virtual. After the game and debrief were completed, the students were asked to reflect on their experience. One of the boys from Team 4 responded: Nick: We didn’t get to read everything, because we were just going [snaps three times—boom boom boom] … running and getting chased by a guy with a knife … well, it was metaphorical knife. Maybe we could have all of the people in one room and talk to them all like around different places in the room. Their teacher asked if they thought that would have been better than the outdoor experience: Nick: It would be more efficient, but maybe the point of it to go out and walk around and see everything too. I don’t know what the objective is, but if the objective is to get all the info real quick, then the best thing to do it here [in one room]. This team expressed that they didn’t know what the purpose of the outdoor portion was and that if they were just expected to learn the information, then it would have been more efficient to give it to them. This failure to put the different pieces together—the physical environment, along with the virtual information—seemed to have contributed to this team’s failure to make sense of the situation. One of the girls from Team 2 responded to the same question about whether it would be better to put everyone in one room. Maya suggested some things she may have learned from doing the activity outside in the real space: “The way the water traveled? If we were up on the hill and the water would go down… So we thought if it was the water contaminating down…” As the team debriefed, some students expressed value in working in the real-world environment, although these understandings were relatively shallow, showing the limitations of this particular enactment for producing learning.

400

SQUIRE AND KLOPFER

CROSS-CASE DISCUSSION These cases suggest both the opportunities and challenges to using handheld technologies to situate learners in environmental engineering practices. Augmented reality simulations can create a compelling context for environmental investigations. Taking learners into the field to conduct a virtual simulation enabled learners to gain a situated experience of environmental science, although the value of this was not always clear to students. This section further explores the significance of the activity’s occurrence in a real-world location, exploring the role of the environment in students’ activity, challenges in conducting virtual investigations, and the role of reflections on “failure” in learning within augmented reality simulations. Student Practices in Environmental Detectives A primary goal guiding the design of this project was to recreate core environmental engineering practices (balancing multiple data sources and the evolving, competing needs of an investigation) within a context where students could test out new ideas and identities without fear of failure. The university and high school students encountered different sets of difficulties in trying to mount their investigations. Yet these different deficiencies led to similar failures in mirroring environmental engineering practice and ultimately determining a solution to the problem. The university students were driven almost exclusively by the collection of water quality data from the wells. Most college students collected samples at the starting location or traveled to where the initial reading was found. When students did conduct interviews, it was because interviews were (not coincidentally) located near desirable sampling sites. In fact, each team collected between 6 and 10 water samples before they ever determined what the units meant or what level was considered toxic. This problem (not knowing toxic levels) was often discussed but dismissed in favor of collecting more samples, perhaps hoping that a pattern would emerge that would put the readings in perspective. In short, wherever there was a problem, the answer was to drill more samples. The holes in students’ understanding were made more evident when they presented their assessment and remediation plans. For example, several teams reported that the TCE was unlikely to reach nearby surface water because it was “far away,” even though they did not know how fast the TCE was moving or how long it had been in the ground (which might indicate that it had already spread to the river). Other teams made assumptions about the use of groundwater for drinking water, though they had no evidence to support these assertions. When collecting water quality samples, the majority of university teams who actually collected data used a “warmer/colder” strategy for locating the source. They would take two samples and move in the direction of the sample with the higher concentration. This method proved to be largely successful, though it was

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

401

susceptible to getting stuck on local minima (due to local variability, random variation in the underlying model, or a smaller secondary spill built into the game) and was very data intensive. Two other strategies that were employed were triangulation and concentric circles. Triangulation (perhaps suggested by the three simultaneous wells limitation in the game) involved drilling three wells in some relatively small area and then moving in the direction of the highest concentration. The concentric circles strategy was designed to start at the original site of contamination and then move out from there, sampling along different radii. Neither of these strategies was more successful in the context of this game, though they might have involved fewer wells and been less susceptible to local variation. These findings suggest that games afford good opportunities for complex problem solving, but that they also need to scaffold players’ thinking and action. The importance of supporting academic game play with other media (books, texts, video) is something well known within the literature of games in social studies education (Squire, 2006; Wentworth & Lewis, 1973). Commercial video games, such as Ninja Gaiden or Viewtiful Joe, are structured so that they can be learned by their players, with levels functioning as essentially embedded tutorials (Gee, 2003; Squire, 2005). Other times game tools, resources, and characters function as embedded scaffolding for players, suggesting ways of thinking to the player. Academic games such as Environmental Detectives might do well to use these same techniques in their designs. Although Environmental Detectives incorporates many game mechanics and features, building them in a way to scaffold thinking may be a productive route for future designs. Regardless, additional scaffolding, whether in game or included via teachers or peers, is needed to further develop scientific thinking. Constructing the Problem Students often recognized shortcomings of their information, citing the lack of data on flow rates or toxic levels, but then proceeded to make recommendations based upon these incorrect or incomplete assumptions. Regardless of this information, the proposed solutions were fairly consistent: Because this is largely a drinking water problem, and because humans do not drink the groundwater, the solution is to plant trees (which have been found to have a measurable, though minimal, effect on reducing groundwater levels of TCE) and subsequently monitor the situation. We have classified this solution as the “political solution”—on the surface it seems like it should satisfy the parties involved (it does not alarm the population, detract from the aesthetics of campus, or call attention to any environmental wrongdoings), but it would be largely ineffective against any real problem. In reality this problem has no one solution that could satisfy everyone and address the real environmental and legal concerns: The pollution is likely to eventually flow into the river, which might upset environmentalists, and although it might not have

402

SQUIRE AND KLOPFER

real environmental consequences, any amount flowing off the campus property has legal implications. Students seemed unable or unwilling to make the hard tradeoffs and address this solution, likely because they were not used to these authentic problems with real trade-offs. The high school students, described in the second part of the study, also struggled to understand the nature of the environmental investigations, but in a different way. Indeed, the high school students generally struggled to balance the need to gather background information with drilling and sampling (as environmental engineers might predict). These students typically defined the activity as a scavenger hunt in which the moment-to-moment goal was to collect interviews as quickly as possible. This meaning was negotiated through both intragroup and intergroup communications (i.e., Dialogue Lines 10–18). Through intergroup exchanges, students negotiated and agreed upon a focus of the activity of one as collecting information, as one might collect pieces of a puzzle. How and why the activity got framed as a scavenger hunt (which is unique and contradicts earlier case studies) was the result of several factors, including the nature of the field trip and students’ past experiences (as evidenced by Gina and Louis’s comments that this “actually was work,” Dialogue Lines 38–39). Across cases, students failed to discern what information was needed for an effective solution. Both groups believed that there was one right answer to the problem and that conducting an investigation was merely a matter of tracking down the source of the toxin. For the engineering students, this problem was about identifying the physical source. For the environmental science students, it was a scavenger hunt to collect as much information about the spill and toxin as possible. Across both cases, students playing Environmental Detectives initially tried simple strategies based on naïve conceptions of environmental science investigations. The task recruited very different strategies among different populations of students, and the game play served as a way for teachers to discuss students’ beliefs about environmental science investigations. This suggests that in designing new educational theories, platforms, and interventions, developers should undergo rapid iterations with a variety of students to understand how problems are constructed and inhabited by players. This suggests that the creation of a problem not only should include particular design features (is novel, contains opportunities for creative expression, helps complete a problem) but also interact with students’ prior knowledge, drives, motivations, interests, and goals in pedagogically desirable ways (Blumenfeld et al., 1991). We find value in doing multiple iterations across different groups to understand how that problem is taken up in different contexts, and to theorize authenticity, cognitive complexity, intrigue, and hence pedagogical value not as properties of the problem, but as emergent phenomena occurring at the intersection of user, designed object, and context (see also Barab, Squire, & Dueber, 2000). Interestingly, much of the foundational work in motivation that problem- and pro-

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

403

ject-based learning advocates developed their theories upon was originally derived from games, suggesting fruitful opportunities for reintroducing gaming conventions into such learning environments (cf. Malone & Lepper, 1987; Squire, 2002). Role of the Physical Environment Across both high school and university students, we found that the teams had relatively little difficulty negotiating the hybrid real and virtual components of augmented reality and within minutes were diving into this mixed-reality environment. Students mapped virtual data onto the real-world context or pointed to locations in the real world and described the concentrations at those locations using data and information off of the handhelds. Using maps and computers, they continuously worked across the spatially distributed problem-solving context. More importantly, students often used knowledge of the surroundings to solve the problem. The college students, who were more familiar with the environment than the high school students (who were on a field trip), investigated sites of known printing presses, metal shops, and other places with large machinery, which had been identified as being associated with TCE early on in the investigation. College students used hypotheses of the activities in each building to guide their thinking, yet they were less personally connected to their surroundings. Situating students’ activity in the physical environment where physical space is part of the learning experience may be the strongest pedagogical value of Environmental Detectives. Across groups, students drew upon their existing knowledge of the terrain, chemicals, or environmental problems associated with the area. The ease with which students synthesized information from the physical and virtual environments suggests that a pedagogical benefit of augmented realities may be in how they encourage learners to draw upon existing knowledge and apply new information to understanding the world around them. The high school cases show how the environment can function as a constrainer of action, as in the first high school case, where students had to traverse rough terrain. In this way, environmental constraints affected their problem-solving paths to an even greater extent. From the first challenge of climbing a fence to the final challenge of negotiating a tunnel, students’ problem solving was concrete, and specific environmental constraints (fences and trees), affordances (such as the tunnel), and local demands (time considerations) were a part of students’ thinking. Students rarely, however, used the physical environment to talk about toxin spreads, as they framed the activity as collecting and synthesizing information rather than gathering data, constructing a narrative, and designing a solution. Consistent with instructional goals, students’ environmental investigations were deeply embedded in the particulars of this physical location. Fences, trees, fields, tunnels, and marshes played a role in students’ thinking and problem solving. In most instances, students used these features as navigation devices, seam-

404

SQUIRE AND KLOPFER

lessly thinking across partners and other classmates, paper-based resources (e.g., paper maps), and Pocket PC-mediated data. Students used the physical environment in deciding which interviews to get (few teams, except the most physically proactive, got a critical interview that was located at the top of a steep hill) but rarely used the physical environment to talk about toxin spreads, as they framed the activity as one of information collection and synthesis. Those teams who framed the problem as a dynamic investigation tied to the landscape were more successful in coming up with well-founded solutions. Instructional Supports for Learning Augmented reality simulations may have communication advantages (i.e., gestures, facial expressions) over their purely virtual counterparts. These groups debated in real time using their voices, gestures, and physical locations as tools. Although similar representations exist in virtual worlds, they require negotiated standards that must be adopted and accepted over time. Emoticons in chat and hand signals by avatars are two examples of these emergent standards. Students in augmented realities do not need to learn these standards, as evidenced by these cases, because they employ the modes of communication with which they are the most familiar. More importantly, team members frequently “voted with their feet” in determining the next location to go. Although this did not always result in democratic decision making (the person holding the computer seemed to have a larger vote), it did make immediately apparent what people’s opinions were and provoked critical dialogue. These affordances show the promise of this technology in structuring learning activities. We learned from these cases, however, that the students struggled with a number of fundamental concepts that will require additional scaffolding in subsequent designs. Students across both cases had difficulty negotiating and making connections between soft qualitative information gained in interviews and hard quantitative data gathered through physical samples, suggesting that the game captured a real and relatively hard problem for these students. The differences in appropriation can be attributed to several factors, but minimally these cases remind us of the power of local cultures in shaping how tools are used. The game was only one object in the activity system, and encompassing cultural models of schooling, specific academic practices, as well as students’ goals, shaped the activity. As the student who complained about the difficulty would suggest, this activity was more complex than ones normally demanded from them at school, and one can easily understand how this activity could be misconstrued in this context as a relatively simple fact-collecting exercise. This suggests that the design needs additional learning scaffolds to promote this insight. Students’ difficulties with the investigations suggest that there may be great learning benefits derived from integrating more extended investigations in envi-

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

405

ronmental engineering curriculum. If learning to redefine the research problem given new information is a central dilemma in environmental engineering investigations, then perhaps allowing students to make these mistakes—to make choices and experience their consequences within a sandbox-like virtual world—is a good thing. One way to think about the pedagogical value of competitive games (as some students seemed to see this activity) in education is to consider their role in inducing failure states (and subsequent reflection), and in providing a socially acceptable context for trying different strategies, experimenting with ideas, and then revising those ideas. Play theorists (e.g., Salen & Zimmerman, 2003) emphasize the importance of creating safe spaces where people can experiment with new ideas and new identities. These cases suggest that Environmental Detectives creates a context in which different teams of students can explore ideas and confront ideas by enacting strategies in game strategies. One particularly promising pattern we observed was different participants arguing for different investigative strategies. In most teams, dominant personalities (as in the high school case) or a combination of social factors (as in the college students’ cases) drove students to prefer one approach over another and prematurely close off strategy discussions. In future iterations of Environmental Detectives, we hope to explore ways of creating game dynamics and groupings so that these approaches are seen in a more even light. Furthermore, we have begun scaffolding students’ problem solving by lengthening the game and providing mid-activity reviews that scaffold students’ articulation of what they have learned thus far and what they still need to find out. The spatially distributed nature of the game makes in-game coaching difficult, although communication technologies (e.g., walkietalkies) that allow teachers to better monitor and scaffold students’ work could be integrated into the game. Summary Augmented reality simulations hold promise for science educators hoping to help students understand science as a social practice, as opposed to an isolated set of facts or procedures. In these enactments of Environmental Detectives, we saw students negotiating complex problem spaces that demand the integration of multiple information data sources. Positioning students in virtual investigations made apparent their beliefs about science, particularly that conducting an investigation was a matter of sampling until the “correct” cause of a spill was located or interviewing experts until the “right” information was gained. As students participated in the activity, they began to gradually unravel the complexity behind conducting an investigation, with some teams coming to deeper understandings than others. The relatively deeply situated experience of conducting a virtual investigation, scaffolded by the design of the simulation, gave students a narrative to draw from as they studied science. Despite the lack of sophisticated teacher mentoring and facilitation in

406

SQUIRE AND KLOPFER

the enactment we studied, Environmental Detectives helped students understand the socially situated nature of scientific practice.

IMPLICATIONS Over the past few years, games and simulations have been criticized for their contrived nature and contrasted with the social “authenticity” of engaging in communities of practice, either through participation in extended communities of practice or through establishing classroom-bound communities of practice engaging in authentic inquiry (e.g., Barab & Duffy, 2000). Quoting Lave (1993), Barab and Duffy pitted learning environments predicated upon a practice field metaphor against those predicated on Lave and Wenger’s (1991) communities of practice, arguing that in practice fields the problems, although authentic in the complexity they bring to the learner, are not authentic in the sense that they are an integral part of the ongoing activity of the society. With the practice field, education is viewed as preparation for some later sets of activities, not as “meaningful activity in its own right” (pp. 48–49). Our results suggest that augmented reality games such as Environmental Detectives have unique potential for learners to experience intellectually productive problems central to science in a psychologically safe space where they can try new ideas (and identities) and learn through failure. Environmental Detectives draws from traditional practice field models of education but deviates from most of these forms in that students are placed under time pressures and forced to make decisions that have consequences on students’ success, and to do so within an environmentally meaningful, authentic, but safe environment where failure is acceptable. What sampling strategies students use, what information students decide to pursue, and when students decide to jump from subgoal to subgoal can have critical societal ramifications. This decision structure is designed not only to be engaging, but to model authentic scientific and engineering practices, including planning research strategies, evaluating the value of data sources, and constructing arguments in debates with team members. Perhaps one way to think about the role of strategy games in learning environments is as precursors to conducting full-scale investigations. The teachers we worked with saw Environmental Detectives as a useful tool for helping students understand some of the trade-offs in doing larger research projects. Perhaps games can provide one way for overcoming some of the challenges to more open-ended forms of inquiry-based learning, such as a lack of student engagement or the experience of cognitive overload by students at the challenges of conducting openended inquiry. Games such as Environmental Detectives might provide scaffolding for conducting larger investigations, serving as simplified but authentic conditions for larger, more complex tasks. The fact that Environmental Detectives explicitly

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

407

bridges real and simulated worlds suggests that it may also help to bridge these practice fields with subsequent actual fields. We believe that by bringing the physical world into the game space, augmented reality gaming applications have unique educational affordances when compared to their purely virtual counterparts. In purely digital simulations, students are asked to make connections between wholly constructed digital virtual environments and the physical landscape. Augmented reality applications allow the physical environment to enter both the problem space and students’ thinking, and these cases suggest how environmental affordances can affect a problem-solving path within an augmented reality environment. In future studies, we will examine how the physical environment enters students’ thinking across a variety of environmental engineering tasks and compare students’ thinking in virtual and augmented reality environments. The design of Environmental Detectives shares many core features with the learning-by-design approach (see Kolodner et al., 2003). It seeks to engage learners in cycles of investigations and explorations and reciprocally design and redesign (within a collaborative classroom culture with particular ritualized practices). The approach detailed here scaffolds the player by introducing challenges and embedding investigative practices (such as comparing evidence to predictions) within game play. We believe that situating a game such as this within a more extensive project-based or learning-by-design-based framework, using these more structured investigations as a source of preparation for more open-ended design tasks, might help avoid some of the pitfalls of open-ended project-based learning described by Kirschner, Sweller, and Clark (2006). The key here is not to pit these various approaches against one another, but to understand how they can be used in concert within a broader curricular framework. They may also have some unique affordances when compared to project-based learning and learning-by-design frameworks that also prepare students for sustained inquiry (see Barab & Luehmann, 2003; Blumenfeld et al. 1991, 2000; Bransford, Franks, Vye & Sherwood, 1989; CTGV, 1993; Kolodner et al., 2003; Krajcik et al., 1998; Soloway et al., 1995). Games have the potential to give much more directed scaffolding to players, in that they can subsume all activity within a particular identity, can introduce new tools to players to enable them to progress into more and more complex thinking, and can present just-in-time multimodal information for feedback (Gee, 2003). Environmental Detectives makes some use of these features, but other games played over more sustained time might make further use of them. Combining and comparing such pedagogical approaches, and researching how each prepares one for sustained inquiry, could be a productive line of future inquiry. We believe that distinguishing between practice fields and communities of practice is useful, particularly with respect to what Lave (1988) called the commoditization of knowledge—the idea that students create contrived products for schools, as opposed to participating in socially meaningful practices. However, to disregard practice fields as inauthentic because they use fictional, imaginary

408

SQUIRE AND KLOPFER

FIGURE 2 Players beginning a round of Environmental Detectives spot their current location on a handheld computer and await readings from a recently placed sample.

worlds in the process of learning is assuming a simplistic notion of authenticity. To equate fantasy, play, and simulation with inauthenticity is misguided. Simulations, which are in effect fictitious worlds, exist at the heart of many scientific endeavors and are used to help scientists explore systems that are otherwise difficult, if not impossible, to explore (Feurzeig & Roberts, 1999). Although this process of learning through imaginary worlds is now aided by computer modeling, learning through imaginary worlds or play is a cross-cultural phenomenon with historical roots as least old as Plato and worth revisiting given the capacity of digital computers to simulate worlds (Caillois, 1979; Jenkins & Squire, 2002). Such game-based environments have the potential to recruit new identities in students, asking them to try on the perspective of environmental investigators (see figure 2). Structuring the game around core disciplinary dilemmas allows students to try on these new ideas and identities, which will inevitably include failure and reflection, within the safety of a classroom environment. Future designs of Environmental Detectives and other augmented reality simulations may do well to learn from the design of strategy games. Whereas one of the goals of Environmental Detectives was to learn about where students were having difficulty planning and conducting their investigations, another was to help them learn about how to better plan and implement an investigation and how to navigate across the necessary data. The struggle that was observed on the part of both the university and high school students to navigate data and properly plan their studies suggests that some additional in-game scaffolding was necessary to help the students improve. Looking to the way that games build skills in novice players, we

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

409

might divide students’ task into some simpler starter tasks or episodes so they could improve their requisite skills and “level up” as they mastered those skills. Also looking to game design, we may make the narrative itself more interactive, responding to the actions of the players. This would allow the players to have more control over their outcomes and provide a tighter feedback loop to help steer them along the right path. Finally, through these cases we gain insight into the important influence of the location and context on student engagement in this augmented reality. The university students were engaged in a situation that was taking place on their campus— the place where they lived and worked. The seniors had been on campus for years and had gotten to know it quite well. The freshmen may have only been on campus for a few months, but that was still quite a bit longer than the high school students, who had only previously spent a couple of field trip days at the nature center. The high school students didn’t know their site as well as the college students knew theirs and, perhaps more importantly, they may not have cared about the site as much. The consequences, even if they were real, were somewhat removed from them. One way to solve this problem, of course, is to situate these games in places that the participants know well and care about. The challenge is balancing that against the unique attributes of a location that one might want to incorporate into a game. If the students don’t deeply care about the location through its inherent real-world meaning, one can try to create that connection through the virtual part of the augmented reality. The game can be designed to connect in ways that the students do care about—deeper information about more well-developed characters, relevant feedback systems, connections to their own communities, or narratives that more tightly link to people, places, and events that are inherently meaningful to them. This requires a partnership between the game developer and game player. In the case of classes such as these, it means getting students and teachers communicating with developers or, better yet, having students and teachers become the developers. That is the direction that this project is headed in, namely integrating teachers (and students) into the game design process to better link location and curriculum to the augmented reality game play.

REFERENCES Barab, S. A., Cherkes-Julkowski, M., Swenson, R., Garrett, S., Shaw, R. E. & Young, M. (1999). Principles of self-organization: Learning as participation in autocatakinetic systems. Journal of the Learning Sciences, 8 , 349–390. Barab, S. A., & Duffy, T. (2000). From practice fields to communities of practice. In D. Jonassen & S. Land (Eds.), Theoretical foundations of learning environments (pp. 25–55). Hillsdale, NJ: Erlbaum. Barab S. A., & Kirshner, D. (2001). Methodologies for capturing learner practices occurring as part of dynamic learning environments. Journal of the Learning Sciences, 10, 5–16.

410

SQUIRE AND KLOPFER

Barab, S. A., & Luehmann, A. (2003). Building sustainable science curriculum: Acknowledging and accommodating local adaptation. Science Education, 87, 454–467. Barab, S. A., Squire, K., & Dueber, B. (2000). Supporting authenticity through participatory learning. Educational Technology Research and Development, 48(2), 37–62. Barron, B. J. S., Schwartz, D. L., Vye, N. J., Moore, A., Petrosino, A., Zech, L., et al. (1998). Doing with understanding: Lessons from research on problem- and project-based learning. Journal of the Learning Sciences, 7, 271–311. Bhandari, A., & Erickson, L. E. (2005). Case studies can fill a critical need in environmental engineering education. Journal of Environmental Engineering, 131, 1121. Blumenfeld, P., Fishman, B., Krajcik, J., Marx, R., & Soloway, E. (2000). Creating usable innovations in systemic reform: Scaling up technology-embedded project-based science in urban schools. Educational Psychologist, 35, 149–164. Blumenfeld, P. C., Soloway, E., Marx, R. W., Krajcik, J. S., Guzdial, M., & Palinscar, A. (1991). Motivating project-based learning: Sustaining the doing, supporting the learning. Educational Psychologist, 26, 369–398. Bransford, J. D., Brown, A. L., & Cocking, R. R. (Eds.). (1999). How people learn: Brain, mind, experience, and school. Washington, DC: National Academy Press. Bransford, J. D., Franks, J. J., Vye, N. J., & Sherwood, R. D. (1989). New approaches to instruction: Because wisdom can’t be told. In S. Vosniadou & A. Ornony (Eds), Similarity and analogical reasoning (pp. 470–497). New York: Cambridge University Press. Brown, J. S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18(1), 32–42. Caillois, R. (1979). Man, play, and games. New York: Schocken Books. Casti, J. (1998). Would-be worlds: How simulation is changing the frontiers of science. New York: Wiley. Cobb, P. (2000). Conducting teaching experiments in collaboration with teachers. In A. E. Kelly & R. A. Lesh (Eds.), Handbook of research design in mathematics and science education (pp. 307–333). Mahwah, NJ: Erlbaum. Cognition and Technology Group at Vanderbilt. (1990). Anchored instruction and its relationship to situated cognition. Educational Researcher, 19(6), 2–10. Cognition and Technology Group at Vanderbilt. (1993). Anchored instruction and situated cognition revisited. Educational Technology, 33(3), 52–70. Cole, M. (1995). The supra-individual envelope of development: Activity and practice; situation and context. In J. Goodnow, P. Miller, & F. Kessel (Eds.), Cultural practices as contexts for development (pp. 105–117). San Francisco: Jossey-Bass. Colella, V. (2000). Participatory simulations: Building collaborative understanding through immersive dynamic modeling Journal of the Learning Sciences, 9, 471–500. Colella, V., Klopfer, E., & Resnick, M. (2001). Adventures in modelling: Exploring complex, dynamic systems with StarLogo. New York: Teachers College Press. Collins, A. (1992). Toward a design science of education. In E. Scanlon & T. O’Shea (Eds.), New directions in educational technology (pp. 15–22). New York: Springer-Verlag. Collins, A., Brown, J. S. & Newman, S. E. (1989). Cognitive apprenticeship: Teaching the crafts of reading, writing, and mathematics. In L. B. Resnick (Ed.), Knowing, Learning, and Instructions: Essays in Honor of Robert Glaser (pp. 453–494). Hillsdale, NJ: Lawrence Erlbaum Associates. Dewey, J. (1938). Context and thought. In R. Bernstein (Ed.), On experience, nature and freedom (pp. 88–110). New York: Library of Liberal Arts. Dorland, D., & Baria, D. N. (1995). Hazardous waste processing in the chemical engineering curriculum. Chemical Engineering Education, 29(3), 178–181.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

411

Dorweiler, V., & Yakhou, M. (1998). Environmental education for the nonenvironmental engineering student: An imperative for the next generation of engineers. Journal of Environmental Education, 29(4), 52–58. Edelson, D. C., Pea, R. D., & Gomez, L. (1996). Constructivism in the collaboratory. In B. G. Wilson (Ed.), Constructivist learning environments: Case studies in instructional design (pp. 151–164). Englewood Cliffs, NJ: Educational Technology. Engestrom Y., & Cole, M. (1997). Situated cognition in search of an agenda. In D. Kirshner & J. A. Whitson (Eds.), Situated cognition: Social, semiotic, and psychological perspectives (pp. 301–310). Mahwah, NJ: Erlbaum. Feurzeig, W., & Roberts, N. (1999). Modeling and simulation in precollege science and mathematics. New York: Springer. Friedman, J. S., & diSessa, A. A. (1999). What students should know about technology: The case of scientific visualization. Journal of Science Education and Technology, 8(3), 175–195. Fruchter, R. (in press). Degrees of engagement in interactive workspaces. International Journal of AI & Society. Games-to-Teach Team. (2003). Design principles of next-generation digital gaming for education. Educational Technology, 43(5), 17–33. Gee, J. P. (1992). The social mind: Language, ideology, and social practice. New York: Bergin & Garvey. Gee, J. P. (1999). Discourse analysis: An introduction to theory and method. New York: Routledge. Gee, J. P. (2003). What video games have to teach us about learning. New York: Palgrave. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory. Hawthorne, NY: Aldine de Gruyter. Glenberg, A. M. (1997). What memory is for. Behavioral and Brain Sciences, 20(1), 1–55. Goodwin, C. (1994). Professional vision. American Anthropologist, 96, 606–633. Greeno, J. G. (1998). The situativity of knowing, learning, and research. American Psychologist, 53, 5–26. Hoadley, C. (2002). Creating context: Design-based research in creating and understanding CSCL. In G. Stahl (Ed.), Computer support for collaborative learning 2002 (pp. 453–462). Mahwah, NJ: Erlbaum. Holland, W., Jenkins, H. & Squire, K. (2003). Theory by design. In Perron, B. and Wolf, M. (Eds.). Video Game Theory Reader (pp. 25–46). London: Routledge. Jenkins, H., & Squire, K. D. (2002). The art of contested spaces. In L. King (Ed.), Game on! (pp. 64–75) London: Barbican Press. Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75–86. Kirshner, D., & Whitson, J. (Eds.). (1997). Situated cognition: Social, semiotic, and psychological perspectives. Mahwah, NJ: Erlbaum. Klopfer, E., & Squire, K. (in press). Developing a platform for augmented reality simulations. Educational Technology Research and Development. Kolodner, J. L., Crismond, D., Fasse, B. B., Gray, J., Holbrook, J., & Puntembakar, S. (2003). Putting a student-centered Learning by DesignTM curriculum into practice: Lessons learned. Journal of the Learning Sciences, 12, 495–547. Krajcik, J., Blumenfeld, P., Marx, R., Bass, K., Fredricks, J., & Soloway, E. (1998). Inquiry in project-based science classrooms. Journal of the Learning Sciences, 7, 313–351. Latour, B. (1987). Science in action: How to follow scientists and engineers through society. Cambridge, MA: Harvard University Press. Lave, J. (1988). Cognition in practice. Cambridge, England: Cambridge University Press.

412

SQUIRE AND KLOPFER

Lave, J. (1993). Situating learning in communities of practice. In L. B. Resnick, J. M. Levine, & S. D. Teasley (Eds.), Perspectives on socially shared cognition (pp. 17–36). Washington, DC: American Psychological Association. Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. New York: Cambridge University Press. Malone, T. W., & Lepper, M. R. (1987). Making learning fun: A taxonomy of intrinsic motivations for learning. In R. E. Snow & M. J. Farr (Eds.), Aptitude, learning and instruction: III. Conative and affective process analyses (pp. 223–253). Hillsdale, NJ: Erlbaum. Nelson, L. M. (1999). Collaborative problem solving. In C. M. Reigeluth (Ed.), Instructional-design theories and models: A new paradigm of instructional theory (pp. 241–267). Mahwah, NJ: Erlbaum. Nepf, H. (2002). Groundwater pollution curriculum package. Retrieved September 7, 2005, from http://web.mit.edu/nepf/www/guide.html Patterson, J. W. (1980). Environmental engineering education: Academia and an evolving profession. Environmental Science and Technology, 14, 524–532. Peirce, C. S. (1992). Questions concerning faculties claimed for man. In N. Houser & C. Kloesel (Eds.). The essential Peirce (Vol. 1, pp. 11–27). Bloomington: Indiana University Press. (Reprinted from Journal of Speculative Philosophy, 2, 1868) Roth, W. M. (1996). Knowledge diffusion in a grade 4-5 classroom during a unit on civil engineering: An analysis of a classroom community in terms of its changing resources and practices. Cognition and Instruction, 14, 179–220. Salen, K., & Zimmerman, E. (2003). Rules of play. Cambridge, MA: MIT Press. Salomon, G. (Ed.). (1993). Distributed cognitions: Psychological and educational considerations. New York: Cambridge University Press. Schank, R. C. (1994). Goal-based scenarios: A radical look at education. Journal of the Learning Sciences, 3, 429–453. Silverman, D. (1993). Interpreting qualitative data: Methods for analysing talk, text, and interaction. Newbury Park, CA: Sage. Soloway, E., Grant, W., Tinker, R., Roschelle, J., Mills, M., Resnick, M., et al. (1999). Science in the palm of their hands. Communications of the ACM, 42(8), 21–26. Spitulnik, J., Studer, S., Finkel, E., Gustafson, E., Laczko, J., & Soloway, E. (1995). The RiverMUD design rationale: Scaffolding for scientific inquiry through modeling, discourse, and decision making in community-based issues (Tech. Rep.). University of Michigan: Highly Interactive Computing Research Group. Squire, K. D. (2002). Rethinking the role of games in education. Game Studies, 2(1). Retrieved May 11, 2005, from http://www.gamestudies.org/0102/squire/ Squire, K. D. (2005). Educating the fighter. , 13(2), 75–88. Squire, K. D. (2006). From context to context: Video games as designed experiences. Educational Reseracher, 35, 19–29. Squire, K. D., Makinster, J., Barnett, M., Barab, A. L., & Barab, S. A. (2003). Designed curriculum and local culture: Acknowledging the primacy of classroom culture. Science Education, 87, 1–22. Stake, R. E. (1995). The art of case study research. Thousand Oaks, CA: Sage. Staudt, C. (2001). Curriculum design principles for using probeware in a project-based learning setting: Learning science in context. In R. Tinker & J. Krajcik (Eds.), Portable technologies: Science learning in context. Innovations in science education and technology (pp. 87–102). New York: Kluwer Academic. Stratford, S. J., Krajcik, J., & Soloway, E. (1998). Secondary students’ dynamic modeling processes: Analyzing, reasoning about, synthesizing, and testing models of stream ecosystems. Journal of Science Education and Technology, 7, 215–234. Wentworth, D. R., & Lewis, D. R. (1973). A review of research on instructional games and simulations in social studies education. Social Education, 37, 432–440.

AUGMENTED REALITY SIMULATION ON HANDHELD COMPUTERS

413

Wertsch, J. (1998). Mind as action. New York: Oxford University Press. Wolfram, S. (2002). A new kind of science. Champaign, IL: Wolfram Media. Zolin, R., Fruchter, R., & Levitt, R. (2003). Realism and control: Problem-based learning environments as a data source for work-related research. International Journal of Engineering Education, 19, 788–798.