Researchers examining situated cognition [9,18,19] have argued that a major reason for student disengagement in schools, and why students may perform below expectations, is partly because the curriculum is disconnected from their lives. In other words, many classroom skills acquired based on a rote approach, fail to transfer to real life. According to situated cognition research, the disconnection lies in the separation of curriculum content from the situations or context in which that content would normally be used. The researchers argue that meaning is lost in this decontextualization of content—so ultimately engagement, learning and student achievement are minimized [17,18,19].

Finding new methods and materials to foster engagement, as well as upgrading teacher skills to use these new materials and methods, has become an important goal for educators. With the growth of Internet applications and tools, the increase in broadband availability, an increasing number of K-12 teachers are expected to incorporate technology into classroom instruction. Researchers contend that as teachers should for new methods, they should be considering looking games and virtual environments because these activity forms are highly engaging and present students with chances for deeper learning and problem-solving opportunities not typically found in school classrooms [20,21,22].

In a study of elementary students, aged 9 to 12 [22], students showed a clear preference for playing video games in their leisure time. The researchers observed proficient behaviors in self-regulation, qualitative thinking, and decision making, all desired characteristics of successful students. From what was learned about the students’ involvement with video games, the researchers argue that video games represent a learning form that engages students and deserves attention from educators. But computer learning approaches in elementary and middle school have typically used games designed for factual content and explicit test practice [9]. Multiple-choice and fill-in-the-blank software for standardized test practice in math, reading, and other subjects has been available to schools for years, while outside of schools, noneducational commercial video games continue to rise in popularity and sales. Given that youth spend significant time with video games, researchers argue that commercial gaming companies are, in a sense, educating our young people. And, although commercial games with educational potential do exist, there remain too few examples that would satisfy teachers and parents, and support engaging academic learning [23].

With technology advancements and the rising popularity of Internet-based multiuser games, skills-based educational games are no longer the only option for schools. Examples of educational virtual learning environments can be found at Arizona State University (Quest Atlantis/Atlantis Remixed) Harvard University (River City and EcoMUVE), North Carolina State University (Crystal Island) and the California Institute of Technology (Whyville). Each of these incorporates avatars in a virtual world, communication/collaboration among players, using tools, simulations, and academic content including science, math, economics, and literacy. One reason these worlds are effective is because students assume the role of the experts working in an authentic virtual circumstance; that is, the students are working in similar conditions as real-world experts [24,8]. Also central to these virtual worlds is the concept of play, where students are free to experiment with learning and try new identities or roles [7,19]. Play itself is an important component of learning [25], and observations of children at play, have shown that in assuming new roles, children exhibited levels of thinking and performance beyond their age levels. Gee [21] and Klopfer [9] argue similarly that games provide students with opportunities in risk-free scenarios to explore and experiment as both novice and expert, engaging in activities and using competencies that are key to being a successful student.

Dede’s [26] virtual world work on the transfer and usability of knowledge, points out that intentionally designing virtual experiences situated in specific tasks, activates the human senses, and creates a feeling of presence, of actually being in the virtual setting, while interacting with its characters and facing its challenges. Dede extends the case for digital environments to include the idea of multiple perspectives: learning is enhanced when a student is able to change his point of view or frame of reference at will. This can be accomplished by “seeing” an object or location from the inside, such as a laboratory in a virtual world, or viewing that same object or location from a distant point in the virtual world, available for exploration and choice, rather than the being isolated in prearranged lessons. The student has control of what lies ahead, can change direction, and can adjust progress—opportunities not found in a regular classroom situation. The virtual environment selected for this study exemplifies Dede’s contention.

This study is concerned with student learning. Content knowledge has been the major focus of testing in schools, but the education experts, who authored the National Education Technology Plan [27], concede that content is not enough. Learning is a complex phenomenon; it includes more than just exposing a student to facts and expecting those facts to be retained. Deeper learning requires connections among the learner, the context of the learning, and the content of the learning [18,19,21].

This study uses a virtual world scenario of a science challenge not typically found in a 4th grade curriculum—sustainability science and the study of genetics. As part of a narrative, students learn about the idea of sustainability, of serving today’s needs without wasting resources needed by people in the future. Along with sustainability learning, students are introduced to the concept of genetics. While most 4th graders might have a sense of genetics from observing traits in their own families, such as eye and hair color, the academic language and vocabulary of genetics are typically unknowns to this age group, since the content is taught in later grade levels. Students in this study were invited to take on the identity of a scientist, and to complete tasks that involved learning how to use scientific tools and concepts in order to breed virtual dragonflies of a specific size and color, and to understand the probability of such occurrences given the genetic characteristics of the mother and the father dragonfly. In the virtual environment narrative, a particular dragonfly called the drako is a pollinator for the Cassip flower which in turn is the source of a rare medicine used to treat serious illnesses. Both the flower and drako are in jeopardy as the environment is being altered by as machines clearing land for new construction. The student player is needed to help change the world, to find alternative solutions to ensure that the drako survives which in turn will ensure a critical medicinal supply needed for the future. Students start the mission by reading the Task List (Figure 2):

The study occurred over a five day period in a classroom equipped with eleven Internet-connected computers. Quest Atlantis was implemented during an extended morning learning centers time with each student working for 2 or more hours online. Students began the quest by teleporting to a virtual location called the Sustainable Science Conservatory on Healthy World, and there they are asked to locate certain characters at specific directional locations. The students were experienced QA/ARX players, so teleporting and navigating in virtual worlds was already a competency they possessed. As the mission progressed, the students visited the Conservatory, and talked with in-world programmed characters (Figure 3, Figure 4 and Figure 5) Ithnus, Uther, and Ekon as they learned about genetics and breeding dragonflies. The in-world characters helped tell the story, and offered choices and assigned tasks to students.

Students received no additional prompts from teachers while navigating or proceeding through the virtual world. Students navigated the environment collecting virtual dragonflies, using mating tanks, working with genetics machines, and solving trait matrices called Punnet Squares in order to produce specific genotypes of dragonflies. Figure 6 shows the Punnet Square tool students used for designing specific dragonfly offspring.

In the virtual science experience, Dr. Uther (a programmed character) teaches students to match dominant and recessive characteristics among red and blue dragonflies to produce at least one small blue dragonfly. In the process, students were involved with learning new vocabulary, gene splicing, breeding, and finding cures for diseases. As tasks were completed, students had to submit written observations of their activities explaining what they had learned. As students interacted with the in-world characters, they made choices with each encounter, which charted their own pathways within the narrative, defining individual learning trajectories, a design feature intended to make player actions result in consequences on the environment as well as on their own learning. Because their actions have a consequence, students understood that they had a direct effect on what happened next in the virtual experience. While living the role of a scientist, they were reading for meaning, comparing ideas, solving problems, collaborating with other students in-world, deciphering puzzles, operating scientific machines, making choices based on time and resources, and writing and reflecting on the content of the mission—and all while they were playing a game.

In one test, students were introduced to six hamsters with varying descriptions of fur length, presence of a tail, and color of fur. Matching genotypes were provided for each of these. For example, test directions showed students that BB or Bb represented a long tail, while bb was a short tail. RR or Rr was presence of a tail, while rr indicated no tail. Directions instructed students to use the genotypes in identifying specific hamsters that would be associated with changing genotypes. A graphical organizer, typically used in genetics courses, called a Punnet Square was incorporated to show how genotypes can be paired to result in specific traits. Finally, students were introduced by test examples to the fact that some phenotypes such as color and size were dominant and some recessive, thus further affecting possible offspring combinations. After 5 days of work in the virtual world, where students talked to virtual scientists, learned about breeding processes, captured virtual dragonflies, and used scientific breeding tools to create certain sizes and colors of dragonflies, a posttest was given.

A common and widely used method for evaluating differences in means between two groups is the paired samples t-test [30]. For the test data in this study, the t-test was used to check for a difference in students’ pre- and posttest scores. Before and after working in the Drakos mission, the fourth grade students were tested on their knowledge of genetics. Table 1, Table 2 and Table 3 show the data from the pre- and post- t-test scores. Provided are results from paired samples correlations, paired sample statistics, and paired samples test comparisons. Shown are improvements with the pretest to posttest mean increasing from 6.633 to 10.133. Pretest to posttest standard deviation changed from 2.0219 to 4.3072. The paired-samples t-test indicated that scores were significantly higher for the posttest subscale (M = 10.1, SD = 4.30) than for the pretest subscale (M = 6.63, SD = 2.02), t(14) = −3.42, with r = 0.46 and d = 1.04. The gain is small, but statistically it is significant. Students showed improved results on the posttest following the 5 days of using QA/ARX to learn the genetics content. See Supplementary A for the pre- and posttests used.

These pre- and posttest scores illustrate that student learning in a virtual worlds environment can be accountable, as in traditional educational measuring methods. The engagement survey showed the students’ reaction to a challenging, nontraditional learning situation, the Drakos mission. The pre- and posttests show the connection between a traditional format for obtaining knowledge (testing) with a nontraditional method of exposing students to intended content (genetics). The outcome is clear by the data—students not only reported being engaged in the virtual worlds environment while learning about genetics, but the knowledge they learned in the virtual environment transferred to a traditional written test, and significant gains were shown.

Engagement is critical to learning. It is a part of the process that keeps a student connected and involved in an experience such that he or she will persevere with challenging situations [17,31,32]. Understanding engagement is important to this study based on research findings stated earlier tying performance and general satisfaction to engaging educational experiences [15,16,22,33].

Table 4 shows the student responses of the engagement survey.

Table 4 shows a mean value of 5.1 with a small standard deviation of 1.10, indicating tightly grouped values around the mean. The most frequently occurring value, the mode, was 6. The median value for this survey was 5. When values were grouped by agreement and disagreement, the two resulting sets clearly show that 90% of students indicated they were engaged versus 10% indicating they were not engaged.

To compare with the data from the pre- and posttests and the engagement survey, qualitative data regarding observed 21st century competencies were obtained. Survey participants were thirty elementary teachers from Australia, England, and the United States, all of whom were experienced using the QA/ARX virtual learning environment in elementary school settings. The skills targeted in the survey questions were selected as desired learning behaviors from definitions of 21st century competencies based on the work of other educators and researchers [26,34,35,36]. Because this survey was newly developed, it required validation. Experts in educational technology, multiuser virtual environments, and game design were consulted on the validity of the survey tool. After their review, expert feedback on content, as well as adjustments to the survey, was implemented in the survey.

The survey, as shown in Table 5, targeted the following range of 21st century skills: technology use, communication, global awareness, collaboration, critical thinking/inquiry, and problem solving. See Supplementary B for the survey content.

This study showed that students learned genetics content through their experiences in the Drakos virtual mission, and the evidence of that learning was shown in a traditional testing format. Students gained knowledge of dragonfly phenotypes through immersive virtual world experiences, then transferred their understanding to a posttest requiring a basic understanding of genetics. With the focus of today’s schools overwhelmingly on standardized testing, educational gaming may be perceived as inappropriate for meeting accountability requirements. This study has shown otherwise. Evidence from the literature on virtual world learning benefits [39,40,23,21,41] confirms that participants consistently gain knowledge at high levels, and further, use their acquired knowledge and experiences to continue their personal learning paths (i.e., they only move forward in the game if they learn and succeed with each mission task). Gee [21] wrote extensively on the intrinsic engagement of virtual learning coupled with the acquisition of in-world knowledge and competencies that are not only readily useful in the real world, but are quickly becoming requirements by companies seeking competent innovative, collaborative workers.

As shown earlier, engagement is critical to learning. It is a part of the process that keeps students connected and involved in an experience such that he or she will persevere in challenging situations. Survey data in this study show that students experienced high levels of engagement while working in the virtual world as students had opportunities for choice, socialization, exploration, and individual curriculum pathways (differentiation). Actions by students in the virtual environment have consequences and meaning that enhance engagement with learning activities and associated content [19,39]. Peer teaching and collaboration happen naturally as students solve problems and navigate the terrain of the virtual environment, teleporting to different worlds, encountering role models, negotiating meaning, and all while learning academic content embedded in the mission narrative.

Examples of learning situations in the virtual world along with observational data from teachers show students demonstrating 21st century competency behaviors while working in QA/ARX. The importance of these competencies is supported in the literature from the perspective of practice fields [38], learning by doing [42], situated cognition [18], collaborative learning in communities [19,34] and deliberate practice [37]. In the act of using these competencies, students are better able to understand and use associated academic content. Learners find themselves challenged to complete missions, which directly and subtly take the learners on a practice path to embark on systematic use of technology, critical thinking/inquiry, problem solving, collaboration, communication, and global awareness.

As educators are considering new, engaging ways to expose students to science (and STEM) experiences, this study seeks to contribute possible ideas, resources, and directions. This study works in tandem with recent Next Generation Science Standards (NGSS) to support deeper learning in science. Additionally, recommendations at the national level suggest a broader approach to learning beyond traditional classroom methods. The U.S. Department of Education’s National Educational Technology Plan [27] points out that academic content can be taught and learned in a variety of ways using educational technology and should include a focus on competent behaviors: “Twenty-first-century competencies and expertise such as critical thinking, complex problem solving, collaboration, and multimedia communication should be woven into all content areas” (p. 13). Included in the recommendations for engaging students are virtual worlds, games, and other interactive, exploratory technologies with embedded academic content. The information from this study may help educators make selections for learning approaches in science that include virtual worlds among other active learning scenarios.