ACM Logo  An ACM Publication  |  CONTRIBUTE  |  FOLLOW    

Knowledge Transfer from Virtual Environments

By Guy Boulet / August 2009

Print Email
Comments (5) Instapaper

Trainees of high-stakes scenarios, such as firefighting drills, pilot training, and search-and-rescue exercises, benefit wildly from simulation technology and virtual environments. The learners can practice their skills more frequently and in a safer environment, and the total cost of maintaining the complex training platforms are usually much lower. Recent developments in 3D technology make virtual training even more attractive in the other sense of the word, too.

Screen shot from Hazmat Hotzone
But how does anyone know whether the knowledge acquired in the virtual environment is effectively transferred to the real world? How can we ensure that knowledge transfer actually happens?

We first must consider how knowledge is acquired in simulated environments and what basic characteristics will ensure knowledge transfer from virtual training to the real world.

Hazmat Hotzone, a game developed by the
Entertainment Technology Center at Carnegie
Mellon University, allows for otherwise hazardous
training to take place in a safe environment.

The Virtual Environment Defined
Virtual environments are computer-simulated environments representing real or imaginary worlds. Although virtual environments are mostly based on visual representation, they also often include audio.

Standard input devices, such as a keyboard, mouse, and earphones or headset, typically support the primary methods of interaction, but some virtual environments may also use specially designed devices, such as head-mounted displays, wired gloves, or devices representing tools or interfaces used in the real environment.

As I mentioned, one of the key advantages of using virtual simulations is that they are normally less expensive than conventional simulators (particularly if the training is being licensed and not built from scratch). They are also more portable since the technology can often run on laptops.

Another significant advantage of virtual environments is that a single 3D model can be reused for different tasks. For example, a model of a ship can be used for spatial awareness training, console operation training, a firefighting simulation, and so on. All that is required is the capability to create new scenarios.

Similarly, a single personal computer can be used to replicate any number of environments, whereas real-life training would require dozens of separate locations; and even 10 or 15 years ago, simulators, unlike computers, couldn't easily be reused to support different modules.

Pull quote But unless there is an actual transfer of knowledge happening between the virtual and real environments, all these advantages might just be virtual as well.

If the instructional design of the virtual environment is deficient, poorly trained personnel might then become the cause of malpractices, which may result injuries and incur financial costs. It is therefore important to understand how valuable knowledge can be acquired within virtual environments, and how it can be transferred to real-world situations.

Spatial Awareness Knowledge
According to Garrett (2007), there is little difference in the way spatial representations are formed in virtual environments compared to real world environments. But how are these spatial representations formed?

There are two primary types of knowledge [PDF, see pages 1-4] in spatial navigation: route knowledge and survey knowledge (Sebrechts, 2000).

Route knowledge refers to procedural knowledge about the movements required to get from one point to another. The acquisition of route knowledge involves learning the layout of a space by navigating it. Local orientation is updated as the subject turns and the knowledge of the space is created through data collected from successive views.

Survey knowledge, on the other end, implies a structured understanding of the layout of a space and the relationships between the elements it contains. Survey knowledge is normally acquired from outside the space through maps or plans. It is more a global representation of the space than a build-up of images.

An example of the difference between route knowledge and survey knowledge is seen in the way someone might give directions to his house. Using route knowledge, the directions would indicate the street names and instructions for when to turn—but the person navigating won't know much about what lies outside her immediate path. On the other hand, using survey knowledge, a person would give his address and let the navigator find it on a map. The navigator now has a global appreciation of where the house is located within the surrounding context, allowing her to figure out the precise route to take to get there.

Although navigating a space is known to build route knowledge, it has been shown that extensive navigation leads to survey-like knowledge that is equivalent to the knowledge available from map learning (Thorndyke and Hayes-Roth, 1982). This suggests that route learning alone may be sufficient over time to acquire both route and survey knowledge. We can therefore assume that elements of both route and survey knowledge can be acquired from computer models and expect that computer-trained individuals would have a sense of familiarity and feel confident in their ability to find their way around when they move from a virtual to a real environment.

Of course, the more the virtual representation is accurate, the higher the confidence once in the real world.

Task Knowledge
Even though there is evidence that spatial knowledge can be transferred from a virtual environment to the real world, simply navigating a virtual environment may not be sufficient to build the knowledge required to perform real-life tasks. The theory of situated cognition suggests that knowledge is linked to context and, therefore, learning happens through performance across situations, rather than by the accumulation of knowledge.

So for learning to be transferable, the context of the virtual environment must show similarities with the context of the real-world environment it represents. For procedural knowledge to be effectively transferred from a virtual to a real environment, the simulated environment must reflect the real environment it models, and learning events must refer to a realistic context.

However, it is not sufficient for a problem situation to be realistic. The design of the task must incorporate a range of complex facets and options to enable and motivate students to learn from it. For Herrington (2006), the "cognitive realism" of the task [PDF] is of greater importance than the real-life likeness of the learning design. This implies that rich graphics and interfaces are less important than the design of the tasks to be completed by the learner, and therefore it is not sufficient that the virtual representation of a real environment be accurate. The tasks to be performed must also be realistic in their nature and in their end results.

Herrington argues that authentic tasks are an integral component of situated learning environments. To support the construction of meaning, students must have roles similar to those found in the real world, and they need to accomplish authentic activities in contexts similar to those in which these activities will be performed in the real world. In this context, peer interaction can also enhance the construction of meaning and foster the acquisition of knowledge.

6 Requirements for Virtual Training
Macedonia and Rosenbloom (2001), citing also Michael Zyda, identified six characteristics that simulations must have to create realism and allow for the acquisition of knowledge that can be transferred to real situation.

1. Immersion. Immersion is the impression that a user is participating in a realistic activity. Immersion occurs when the learner, through intellectual, emotional, and normative reactions, has to take meaningful actions in order to influence the state of the virtual environment.

2. Networking and databases. The distribution of virtual environments enables a large number of users to interact in the same virtual environment. People can collaborate and perform group tasks over networked virtual environments but in order for this to be realistic, databases must be updated frequently in order for the actions of one user, as well as the effects of these actions on the environment, to be visualized by the others.

3. Story. Constructivism learning theory argues that humans generate knowledge and meaning from their experiences. Success of learning within virtual environments would therefore be linked to interactions, which should be designed to provide the learners with challenging experiences in which they will build new or consolidate existing knowledge.

4. Characters. Animated characters can play various roles in a virtual environment. They can facilitate learning by helping learners accomplish their tasks or by challenging them. To be effective, their behavior must be realistic and responsive to the user's actions. Characters can be automated as part of the scenario, or they can also be instructors or other learners interacting within the virtual environment.

5. Setup. The environment in which the story takes place must be realistic and provide conditions that will foster learning. Not only does the virtual environment need to be properly designed, the physical environment in which learning takes place must also be adequate.

6. Direction. Learners need to be guided and monitored within the virtual environment. They need to be told that what they do is right or wrong, and they need to understand why. This can be accomplished by an instructor observing the learners, or through the use of a virtual coach providing visual or auditory feedback when the learner executes an action or completes a task.

Minimum Realism
For knowledge to be acquired in a virtual environment and transferred to the real world, a minimum of realism is required. This realism should be more contextual than physical, meaning that knowledge has to be acquired in a context as close as possible of the one in which the knowledge has to be applied.

An acceptable level of realism can be achieved by providing learners with a simulated environment that resembles the work environment in which they can play their own role and where they can interact and perform realistic and meaningful tasks in realistic conditions and under proper supervision. These are, in our opinion, the conditions required for virtual environment to be efficiently used to prepare personnel to perform tasks in their real work environment.


  • Fri, 18 Jan 2008
    Post by Damon

    Did all the Britney Spears references increase the number of hits for this article? Could you maybe give us a rough number, like percentage above average?

  • Thu, 17 Jan 2008
    Post by Lauren

    Lisa - I agree that quality in ideas and writing are important. However, I wouldn''t undermine web analytics as a way to help you determine whether your articles are connecting with your audience. Analytics doesn''t just track how many people read your article, or open the web page, but also tells you how long someone stays on that page (a sure indicator whether your article is interesting to the reader). You can also see which pages / or articles on your site are read the most, or even what time of day or week gets the most traffic. Certainly web analytics only tells part of the story, but used well, those numbers are really helpful.

  • Tue, 15 Jan 2008
    Post by Mari Cruz Garcia

    Answering your question, Mark... Yes, by mentioning "Britney Spears", you will get more hits that by tagging "Aldous Huxley", or "Kafta", for example. These are the easily-manipulated and manitulative societies in which we are living

  • Thu, 10 Jan 2008
    Post by Karl

    Lisa, I agree 100%, it is like a professor who is tough on his or her students, you might get poorer evaluations than the instructor who is "easy" and gives out pizza at the end of the semester but the students learn more because the professor or teacher is tougher. Number of hits, to me, is not a direct evaluation of quality of writing. Karl

  • Mon, 07 Jan 2008
    Post by Mark Notess

    So Lisa, are you suggesting that mentioning Britney in your article will get you more hits than I got for mentioning Beowulf in mine? What''s our culture coming to? :)