Abstract
This paper presents
our findings after the first year and a half of a multi-year deployment of an
ImmersaDesk® to a local elementary school, investigating its effectiveness
in enhancing science education. These findings deal with how VR can aid in the
coordination of multiple representations, and how to integrate the technology
into the existing school culture.
1. Introduction
In August 1999 we moved an ImmersaDesk into Abraham
Lincoln Elementary School in Oak Park, Illinois as part of a multi-year study
into the use of VR in
elementary school science education. So far over 250 students have used the ImmersaDesk
at the school to interact with a series of
conceptual learning environments designed to study the effectiveness of
VR-based learning strategies for coordinating multiple external and internal
(mental) models of scientific phenomena. These studies focus both on supporting
individual learning, and providing it effectively within the constraints
imposed by the classrooms of contemporary schools.
The objective of our
effort is to maximize the effective use of this technology in a single working
public elementary school by providing a rich resource and support base, and a
broad range of participation opportunities. A central theme running through our
integration efforts is the alignment of all of our activities with the
school’s learning goals, curriculum, and operating practices so the
teachers feel confident that participation is not adding irrelevant activities
to an already crowded curriculum. While our long-term goal is to impact these
criteria, we can not impose a decontextualized solution, but rather we must
discover together an evolutionary path to the future.
Our long-term goal is to help prepare the nation's schools for the
advanced visualization technologies such as VR that are becoming operative in
research laboratories. The
ability of children to successfully adapt to today’s clumsy controls and
low-resolution displays, and their accelerated learning capacity provide at
least prima facie justification for exploring these new media. Recent empirical
studies [3, 8, 14, 17] provide initial evidence of learning gains using
VR-based environments with students ranging from second grade through high
school. Whether this technology
will be adopted depends on whether we can find more solid evidence of
effectiveness for the individual learner, and whether we can successfully align
its use with school goals, curricula, practices, and culture
While the history of meaningful
technology adoptions in schools has been discouraging [4], we have the
advantage of the lessons learned from the computing integration efforts over
the past two decades. The central theme of that experience has been the
recognition of the importance of addressing the needs of the individual learner
while respecting the constraints imposed by the educational context within
which learning occurs. Therefore we are seeking answers to the following
research questions:
1. How does VR interact with the cognitive processes involved in
learning, and can we exploit this knowledge for the development of effective
educational interventions?
2. How should VR be incorporated in the classroom, into current teaching
practices, and into the life of the school?
Section 2 discusses
our use of this technology in constructing and coordinating mental models.
Section 3 describes the VR resource in the school, and section 4 describes the
three tracks of research that we are pursuing using that resource. Section 5
gives some of our current findings and section 6 looks at our current
directions in our second year at the school.
2. Multiple Representations
We are focusing on the role of
coordinating multiple representations as the central process in science
learning. The adage that ‘a picture is worth a thousand words’ has
intuitive appeal, but cognitive scientists do not yet understand under what
circumstances multiple representations have positive consequences for learning.
Learning fundamental ideas in
science requires that the learner construct mental models of the system
to be understood – integrated dynamic representations that support
argumentation, explanation and prediction. Visualizations facilitate this
construction. However construction is not enough. Learning fundamental ideas in
science requires that the learner coordinate multiple mental models.
There are several kinds of mappings that can be used to accomplish this.
2.1 Constructing a Mental Model
Knowing a fundamental idea in science is not merely being able to repeat
its standard verbal formulation, or to solve a canonical set of textbook
problems. Verbal formulations and problem-solving skills can be acquired
without understanding. Instead, cognitive and educational researchers have
converged on the hypothesis that understanding an idea requires that the
learner construct an internal representation, often referred to as a mental
model, of the type of system for which that idea is relevant. A mental model is
an integrated, dynamic representation that allows the learner to simulate the
relevant system in the mind’s eye. Mental models support argumentation,
especially counterfactual reasoning, explanation and prediction.
Educational materials abound with
visualizations. Visual representations support mental models in numerous ways
and the more powerful the visualization technology, the more complete the support:
·
Integration.
Visual representations provide simultaneous and parallel access to multiple
parts or components of the visualized system, while a text necessarily presents
them in some sequence, leaving the learner with the task of integration.
·
Dynamics. Dynamic
visualizations demonstrate how components of a system interact and change over
time.
·
Reification.
Visualizations can convert abstractions into perceivable objects and engage
perception in support of conceptual learning.
·
Activity.
Interactive visualizations allow a learner to manipulate a system, and draw
upon the near-universal principle that knowledge is constructed in the course
of activity.
·
Immersion.
High-end visualizations combine all of these features to let the learner feel
as if he or she is directly experiencing the visualized system, thus drawing
upon children’s natural capacity for experiential learning.
However, recognizing the power of advanced
visualizations to support the construction of mental models does not specify
which visualizations are most useful for particular topics. Additional
principles are needed.
2.2 Coordination of Multiple Models
If the coordination of multiple representations is central to science learning, then pedagogical visualizations should be designed to support these different types of mappings. In VR a common mapping is one between exocentered and endocentered representations [5]. This is a relation between two different perspectives, in the literal sense of views from different positions in space. However, there are many different types of mapping relations, and different mappings are relevant for different learning targets. For example:
· Time. the successive stages in mountain formation, or time slices
of an eroding shoreline.
· Transformation. a process running at different speeds, or an
object undergoing changes in shape.
· Scale. the mapping between the normal and the molecular view of a
substance or between a tissue and its components cells.
· Part-whole. the mapping from the components of an electrical
circuit to the behavior of the circuit as whole, or from orbit, rotation and
the tilt of the Earth’s axis to the march of the seasons.
· Form-function. the relation between the shape of a DNA molecule
and the process of duplicating the genetic code.
· Abstraction. the relation between a photo of a baseball homerun
and the corresponding force diagram for projectile motion.
· Interpretation. the relation between hypothesis and evidence, or
between alternative hypotheses.
Each
type of mapping is involved in learning some scientific concepts; each presents
the learner with a challenge. However, noticing that particular subject matter
topics include certain mappings does not exhaust the importance of coordinating
multiple representations and models. Cognitive research has also provided
empirical support for at least two types of mappings that have greater
generality: analogy and schema articulation.
According to the
analogical transfer hypothesis, the learner can construct a new mental model by
mapping a novel system onto an already familiar one. For example, by mapping
conduits onto wires, water sources onto batteries, and so on, a learner can
construct an initial (partial) mental model of how electricity works. Unlike
the mappings considered above, analogical mapping is not tied to any particular
subject matter. The main limitation is the availability of a useful analogue.
According to the
schema articulation hypothesis, a mental model can be constructed by filling in
an abstract schema in a novel way [12, 18]. For example, the Darwinian schema
of change in a population via variation and selection can be articulated to
explain why bacterial diseases become resistant to antibiotics and how the
immune system works. Like analogy, schema articulation is not intrinsically
tied to particular subject matters. The main limitation is the availability of
a suitable schema.
The issue of multiple
representations is not straightforward. Several researchers have suggested that
the presentation of multiple representations is beneficial in and of itself,
without any additional support. Multiple representations have been hypothesized
to help because they provide redundancy at different levels of abstraction [6],
present complementary aspects of the subject matter [2], influence the
allocation of attention [16], induce distinct cognitive processes [10], or provide
a better fit to individual learning styles. Salzman, et al. [17] recently
provided support for this idea in a conceptual learning task involving electric
fields. In their study, high school students who interacted with both an
exocentered view and an endocentered view of an electrical field exhibited a
significantly higher level of mastery than students who interacted with only
one of these visualizations.
However there are strong reasons to
believe that the issue is more complicated than “two representations are
better than one.” Constructing a mapping between multiple mental models is
itself a cognitive task at which the learner can succeed or fail. This is
illustrated by Ainsworth, et. al. [2], who found that two visual
representations in combination produced a significantly lower learning
outcome than either representation by itself. Presumably, the students in this
study could not compute the relevant mapping, or else allocated so much working
memory to creating the mapping that it interfered with other aspects of their
learning [19].
Consistent with this outcome, research on analogical
reasoning has demonstrated that neither children nor adults are strong mappers.
Children in particular have a tendency to relate representations to each other
on the basis of surface features, rather than the deeper features that would be
more helpful. Teachers, tutors, and transitions
between the various representations are needed to coordinate the representations.
3. The VR Resource in The School
In 1997, we conducted VR-based
learning studies with over 80 elementary school children within the Electronic
Visualization Laboratory at the University of Illinois at Chicago. It became
obvious that a much better solution would be to conduct these studies within an
elementary school itself. In December 1988 we moved an ImmersaDesk into a
classroom at Abraham Lincoln Elementary School in Oak Park, Illinois for a
month-long research project with 76 2nd graders looking at these issues of
multiple representations in terms of the shape of the Earth [8]. In August 1999
we returned to Lincoln on a more permanent basis.
Abraham Lincoln
Elementary School is a K-6 elementary school in Oak Park, Illinois, a racially
and economically diverse inner-ring suburb bordering Chicago's West Side.
Lincoln is attractive as a research site for its size, diversity, and state of
technology adoption. It is a large school (620+ students), nearly always
allocating four (20-30-student) classrooms at each of the K-6 grade levels.
Besides a racially and economically diverse student body (64% white, 29%
african american, 4% hispanic, 3% asian) and faculty, Lincoln offers diversity
of subject mastery, as reflected by IGAP (Illinois Goal Assessment Program) and
Stanford-9 achievement tests administered at the school. While performing
moderately above average as a school, Lincoln has significant representation in
all performance quartiles. The school is also roughly average with respect to
technology infusion, with about one computer for every five children,
distributed both in classrooms and in the school’s Media Center, and an
orientation more toward computer literacy and technology education than
conceptual learning.
We have the advantage
of a strong pre-existing relationship with Lincoln. One of the members of our
team was an active Lincoln parent for twelve years and a member of the Board of
Education. We bolstered that relationship through a series of meetings both at
our laboratories, familiarizing teachers with our VR technology, and at the
school, familiarizing researchers with the realities of a contemporary elementary
school.
The current equipment at Lincoln
consists of an ImmersaDesk driven by a 4-processor SGI deskside Onyx IR, and a
19” stereo monitor driven by a dual processor SGI Octane. While most of
the work takes place at the ImmersaDesk, the Octane allows us to explore collaborative
virtual environments, specifically those with heterogeneous views of a shared
space. We run both screens at 1024 x 768 in 96Hz stereo. Both computers are
wired to send audio to the ImmersaDesk’s speakers. When we are conducting
more formal learning studies we bring in recording equipment such as cameras
and microphones.
The equipment is located in a room
next to the school’s Media Center. This location was determined in
consultation with Lincoln administration and staff, and is designed to minimize
the impact of ‘pull-outs’ from regular classroom activity. Since
classes regularly visit the school’s Media Center, its possible to
‘pull over’ students to work individually or in small groups while
the rest of their class is doing other activities in the Media Center. Since the Media Center
has its own instructor, a teacher can accompany his or her students in the
activity at the ImmersaDesk while the media center teacher works with the rest
of the class. The Media Center also contains a cluster of 26 networked iMacs,
allowing us to investigate collaborative activities between children who are
immersed and children at desktop computers.
4. Three Tracks of Research
We are currently
performing three tracks of research at Lincoln. All three focus on multiple
representations, but in different ways. The Cognitive Studies are
in-depth studies of learning that focus on the coordination of multiple
perspectives in the formation of mental models. The QuickWorlds focus on
augmenting the existing classroom visualizations (textbooks, filmstrips, 3D
models) to give the students an additional, more interactive, visualization of
the material. The Virtual Ambients help students understand the
scientific method by following data through the process from collection through
abstraction. These three tracks of research also support different types of VR
experiences. The Cognitive Studies focus on individual learners, the
QuickWorlds focus on small groups, and the Virtual Ambients focus on whole classrooms.
4.1 Cognitive Studies
The Cognitive Studies track [8, 13] involves conducting
in-depth learning studies with an entire grade level of individual learners.
These types of studies take several weeks to complete the pre-testing, VR experience,
and post-testing. It requires a high level of commitment both from the teachers
and the researchers as it involves pulling out each student from class three to
four times. During the time the children are pulled out of class, they are
interacting with the VR technology and the researchers, but not their usual
classroom teacher. This creates good conditions for a formal study but is not
very realistic in terms of common school usage.
Our first study in 1998 at Lincoln, with the second
grade, showed a very significant improvement in the children who went through
the VR experience of walking on a small-diameter asteroid and the subsequent
dialogue relating the experience to the Earth. Our second study at Lincoln,
with the first grade, compared students who experienced the VR asteroid and
took part in the dialogue to those who only had the dialogue. Both groups of
students showed significant improvement, and we are still analyzing the
differences in the various knowledge components.
Working with the 1st graders brought up developmental issues
which were not obvious with the 2nd graders. Quite a few of the 1st graders had
difficulty giving directions to their partner who they saw as an avatar. They
had trouble taking their partner’s viewpoint. Because of this we are augmenting our Round Earth work with
another world. ‘Piaget World’
draws upon the classical demonstration by Jean Piaget of what he called ‘egocentrism’, but
instead of demonstrating the inability to imagine a different point of view
than one’s own, we will use a VR version of Piaget’s set up to help
young learners overcome it [15].
The Piaget World visualization consists of an
exocentered view of an alpine landscape with various objects (cottage, river,
etc), replicating Piaget’s pasteboard model. Depending on the
viewer’s point of view, different objects are occluded by features of the
landscape. Four heads provide multiple possible viewpoints, and the child has
the ability to ‘become’ a particular head, changing their visual
perspective, to verify what that eyeball can see. We are currently pilot
testing this world.
Figure 1: A selection of
QuickWorlds that were developed in our first year and a half at Lincoln: a wood
ant and bee with internal organs, a beating human heart with lungs showing
blood flow, Mt. St. Helens before and after the eruption, an iceberg, the solar
system in scale to show the vast distances involved and out of scale to show
the details of the planets, the distribution of earthquakes over the Earth, and
the interior of the Earth
4.2 QuickWorlds
The QuickWorlds track is intended to provide a
fast-turnaround mechanism for teachers who would like to make virtual dynamic models available to their students as part of the regular learning program. VR
provides a mechanism for making additional models accessible to students
without the costs of physical fabrication. VR also makes models accessible that
would not otherwise ‘fit’ inside a regular classroom.
This track involves a teacher bringing several of his or
her students to the ImmersaDesk for 15 to 20 minutes to view an interactive 3D
model as part of an existing curricular unit. Depending on the number of
children per group, five to eight sessions may be needed for all of the
children in a class to participate. The teacher controls the virtual experience
and the researchers remain in the background as much as possible.
Figure
2: Photograph of four 6th graders exploring one of the virtual ambient
environments to collect data. The children rotate through the roles of
navigator, spotters, and note taker. Afterwards, they will then return to their
classroom and integrate their data with the data of other groups
The teachers request these models and tell us what
features we should emphasize. We then build the virtual models and make them
available at the school. Several of the QuickWorlds that we have made on request
from the teachers are shown in Figure 1. We wanted this track to require a
minimal commitment from the teacher. It focuses on the ImmersaDesk as just
another presentation medium for the teachers to use, and we are interested in
observing how they use it.
For example, the physical education teacher requested a
model of the human heart for her 4th graders. She currently teaches about the
heart using ‘heart land’ where the children roll around on the gym
floor moving dodgeballs from place to place. She wanted to augment this with a
more realistic animated model showing the blood flow through the various
chambers. This allowed her to help the children relate their participatory
model of the heart on the gym floor to the actual shape and function of the
heart. She used the VR experience to ask questions of the children and prompt a
dialogue.
Initially the teacher would talk to the students while
we would manipulate the model, but the teacher quickly took over and began
manipulating the model herself, eventually giving control over to the students.
All of the QuickWorlds have a dynamic cutting plane allowing the teacher or
students to dynamically slice through the model and see the internal structure
– this became a very popular feature in helping to focus attention on key
parts of the model. Head tracking the teacher was helpful because it made it
easier for her to set up appropriate views so the students could see exactly
what she was talking about. The large screen was useful in allowing the groups
of children to all see what was happening while they were talking about it.
4.3
Virtual Ambients
The
Virtual Ambient Environments track [11] uses VR as a small part of a larger
full-classroom unit where the students survey a large virtual space and then
integrate their data as a class. The nature of inquiry (methodology, or the
philosophy of science) is seldom taught as a subject matter in its own right.
Benchmarks [1] claim that it should be, and the district’s science
curriculum identifies it as a central learning goal. Data collection has long
been believed to be a valuable component of activity-based learning strategies,
but for practical reasons is often replaced by pre-constructed databases. VR appears
attractive for this purpose because it can provide access to simulated
environments which might otherwise be impossible to visit in person, while
still providing experiences analogous to those undertaken by a scientist in
real experimental work.
We also believe that the act of collecting and transcribing data will promote conceptual understanding by making the connection between the environment being studied and the ‘scientific’ representations of data drawn from that environment explicit. Evidence from other domains gives credence to this theory; Holst [7] for example, found that degrading the user interface to algebra-learning software – requiring the students to do manual data within a fixed time period - resulted in both fewer interactions and a significantly higher level of content mastery. Such evidence highlights the need to balance between scaffolding on the one hand, and engaging in low-level activities associated with scientific investigation on the other.
In our first year at Lincoln two classrooms of 6th graders and one classroom on 2nd graders visited the same virtual space. The current space is a large flat field, the size of several football fields, laid out in the form of a tic-tac-toe board. Eight different kinds of plants are scattered throughout this space in clusters.
In the first phase of the experience a ‘scouting team’ of 3 or 4 children spend about 30 minutes wandering around the space on their own. They take notes on what they see in the virtual world and then present their findings to the entire class. The entire class spends an hour taking about how to systematically survey the space, then the class breaks into groups of three to four students, and each of those six groups spends 20 minutes in VR. Each of these groups surveys their piece of the virtual world taking notes on what they find. See Figure 2. For the second graders their task was to find how many different plants there are in the space and to draw each of them. For the sixth graders their task was to find correlations between which types of plants grew together. After the VR experience, two more hours of class time were used to integrate the data, and in the case of the sixth graders to convert their collected data into symbolic representations and discover the ‘rules’ of the space.
So
far in our second year at Lincoln we have had one classroom of 4th graders
survey the space. This time instead of dividing up the class to collect data
from different areas of the field, the different groups of children visited the
field at different times to investigate the relative growth rates of the
plants; the growth rate of the plants over six months was simulated over two
days.
These
experiences combine work with the classroom teacher and the researchers, and
try to more directly investigate how to use VR as part of a classroom activity.
As with the QuickWorlds, the large screen was valuable in allowing the groups
of children to see the space and talk about its features at the same time. The
children typically divided the labour into rotating jobs such as the navigator,
and the note-taker, and one or two spotters on the look out for plants in the
space.
5.
Lessons Learned
At the end of our
first year and a half at Lincoln, over 250 children have had a learning
experience incorporating the ImmersaDesk. Almost all of the 2nd and 4th graders, half of
the former 6th graders, and one quarter of the 2nd, 3rd and current 6th graders
have participated. We have learned lessons in terms of the use of the
technology, the logistics involved in making this technology available, and how
to evaluate the learning that goes on.
5.1
Technology
Many of the children are familiar
with using videogame controllers. As such they had an easy time using the
buttons on the Wanda™ we are using as our main interaction device, but
had a harder time with the isometric joystick. They also tended to treat the
Wanda as a two-handed controller, like a videogame controller, rather than a
one-handed tracked pointing device. Typically we have used the joystick on the
Wanda for movement and the buttons for actions within the environment. Since
the Wanda has only three buttons, its not easy to map movement onto the
buttons. Because of this we have been investigating more conventional video
game controllers with four buttons in a diamond for movement as well as
additional buttons for additional functions. Ideally we’d like to
transition to specific physical input devices targeted to the particular applications.
By remaining in the school over
several years we are able to investigate repeated usage of the technology.
Almost all of the 4th graders taking part in the virtual ambient work this year
participated in our Round Earth work two years ago. These children were
familiar with the technology from their earlier experience and easily
transitioned to the new application.
Figure
3: Photograph of the prototype plasma panel and SGI Linux machine running one
of the Virtual Ambients that we will place in a Lincoln classroom.
Even with the
children’s experience with videogames, most of them found the ImmersaDesk
and its virtual worlds to be quite impressive: ‘Sweet!’ being the
most common comment. Several students also favourably compared the experience
to their Sony Playstation®, but of course they wanted to know if we had
more games. While this does motivate the children, it is exactly this novelty
effect that we are trying to get beyond through this long-term deployment to
the school.
We have tried to keep the frame
rates high (at least 15 frames per second), avoided high-vection imagery, and
focussed on large open spaces in the worlds. Even so, we have had one serious
case of simulator sickness out of the 250 children that have experienced the
ImmersaDesk in the school. This child was in a group of four children exploring
one of the Virtual Ambients and was not being tracked when she became ill. We
found out later that this child was susceptible to carsickness and had trouble
at OmniMax films. Interestingly enough this incident did not dissuade any of
her classmates from taking their turn. We also had one other student who was
interacting alone with the ImmersaDesk as part of the cognitive studies. She
had no trouble using the non-tracked glasses at the stereo monitor, but decided
to stop as soon as she put on the tracked glasses at the ImmersaDesk. In both
cases it seemed like the large amount of sudden movement covering their visual
field (either from the non-tracked child seeing the world move dramatically
when the tracked child moved his head quickly, or from putting the tracked
glasses on) overwhelmed them.
Because of these incidents we have
adopted the following strategy for head-tracking:
·
If we have one child in
front of the ImmersaDesk, then that child is head-tracked.
·
If we have a teacher
leading a group of children at the ImmersaDesk then the teacher is head-tracked
to give the children the most appropriate view. The teacher knows not to make
sudden head movements.
·
If we have several
children in front of the ImmersaDesk then we turn head-tracking off since this
avoids the intrusion of regularly exchanging the head-tracked glasses, and
keeps the image stable.
We have noticed several times,
children who are not head-tracked acting like they are tracked, trying to look
around virtual objects; there is clearly an interest in being tracked. Our goal
is balancing that interest with safety.
More details on what we have
learned about working with the technology in the school can be found in [9].
5.2 Logistics
The children not only enjoyed the VR experience, but also the pre-testing and the
post-testing. We believe this is partly because they are out of class, but also
because it gave them a chance to interact with an adult one on one.
While we
initially hoped to pull students over from the Media Center, we have mostly
relied on pull-outs from class since the students’ time in the Media
Center is generally limited, and very focused. While we can rely on pull-outs
for a research program like this one, that path will not be a sustainable model
in the school.
Because of this there are a couple
different routes that we are going to explore in terms of how to integrate the
technology into the school. The first path involves a scenario where high end,
rare, technology is deployed in a single space in the school. This is basically
the setup that we have now. Thus the ‘VR room’ would be similar to
the music room or the art room in the school where equipment is staffed by an
expert who can support all of the students in the school on a scheduled pattern.
An alternative
strategy is to rely on more consumer level technology and move the device into
the classrooms. The virtual worlds that we have been using at Lincoln are based
on SGI Performer, and Performer runs on both SGI machines and on Linux
machines. We have ported all of these worlds to Linux and have run them on a
Linux based ImmersaDesk. The stereo support on graphics cards under Linux is
still weak, however the worlds do run at roughly Octane level speeds in monoscopic
mode.
In the spring of 2001
we will deploy a 50-inch plasma panel, running some of these worlds, driven by
a Linux machine, to the 4th grade open space at Lincoln where the four 4th
grade classrooms share a common large room. The panel will not be head-tracked,
for the same reasons we have not been head-tracking the ImmersaDesk when
multiple students are interacting, and will not show stereo visuals. The panel
will include hand-tracking, and a wanda for navigation and interaction, as a
tracked hand is useful for pointing at objects in the virtual ambient field,
and in manipulating the cutting plane in the QuickWorlds. We may also add a
touch-screen to the panel. This setup reduces the cost of the hardware by a
factor of 10 from roughly $250,000 to roughly $25,000. See figure 3.
We are trying to leverage the children’s willingness to spend
hours with a couple friends in front of their TV with its game console, but
providing more meaningful content. This will give the teachers more of a chance
to use the world opportunistically; the screen will always be left on giving a
‘fifth wall’ to the classroom. For example this fifth wall could be
constantly showing one of the virtual ambient environments allowing the
children to easily visit it and explore regularly, just as they might regularly
visit a field next to the school.
Our goal at this point in the
research is to move beyond being an interesting novelty and towards being an
integral part of the curriculum. Teachers are busy people, and we need to prove
to them that the time they spend working with us is really valuable to their
students. But its not simply a matter of increasing the number of curricular
units that we can augment with VR, but also offering a capability that is
useful in the long-term.
5.3 Evaluation
Probably the most important lesson we have learned is that we have not found the ‘silver bullet’ for assessing the effectiveness of these interventions. Assessment of learning has defied consensus in the education community, so this should not be a surprise. Simple comparisons between this new way and the ‘standard’ way of teaching the material are not very meaningful since there are so many factors that could contribute to the success or failure of the new intervention
In the Cognitive
Studies we are employing pre-tests and post-tests with questions are asked
verbally, through pencil and paper drawings, and through 3D models looking for
statistically significant improvements in the different treatments, and looking
more in depth at individual learners. In the QuickWorlds the evaluations are
more anecdotal at this stage, based on the impressions of the classroom
teacher. For the Virtual Ambients the evaluation will be longer term, looking
for improvements in the students methodology, not just the end product.
6. Current Directions in Year Two
The first year and a half of this deployment have shown us that there are a great many issues related to the coordination of multiple perspectives that we will pursue more in the coming years. The lessons learned from the previous year suggest that the hardware is robust enough to deploy in a school, but being in the school is not enough. We need to be in the classroom. In either case we need to continue working on our tools for learning assessment.
We are currently
running a pilot study with the Piaget World that will lead into the third large
cognitive study with the Round Earth next spring. We are building more
QuickWorlds, though this term we may focus their use on smaller groups of
children who need extra help with a topic. We are continuing with the Virtual
Ambient work involving 4th graders and the 6th graders looking at issues of
sampling. In the spring the 6th graders will return to the same space three
times, each time with an increasingly complicated space to sample. We will
continue to have the Deskside Onyx and the Octane in the Media Center with the
ImmersaDesk, and will deploy the plasma panel driven by a Linux machine to the
4th grade Open Space.
We are also beginning
an NSF funded collaboration with the University of Michigan, Northwestern University,
and Georgia Tech to combine their PC-based science inquiry tools with our VR
work to create more comprehensive immersive inquiry environments.
Acknowledgements
This project involves the hard work of a large number of people including Ya Ju Lin, Mark Orr, Joseph Alexander, Josh Hemmerich, Yong-Joo Cho, Dave Pape and Greg Dawe. We would especially like to thank Carol Dudzik and the teachers, staff, and students of Abraham Lincoln Elementary School.
This research was
made possible through major funding from the National Science Foundation,
specifically EIA 9720351, DUE 9979537, BCS 9907839,
and EAI 0085946.
The virtual reality
research, collaborations, and outreach programs at the Electronic Visualization
Laboratory (EVL) at the University of Illinois at Chicago are made possible by
major funding from the National Science Foundation (NSF), awards EIA-9802090,
EIA-9871058, ANI-9980480, and ANI-9730202, as well as the NSF Partnerships for
Advanced Computational Infrastructure cooperative agreement ACI-9619019 to the
National Computational Science Alliance. EVL also receives major funding from
the US Department of Energy (DOE) Science Grid program, awards 99ER25388 and
99ER25405, and the DOE ASCI VIEWS program, award B347714.
ImmersaDesk is a
registered trademark and Wanda is a trademark of the Board of Trustees of the
University of Illinois.
References
[1] American Association for the Advancement of Science. (1994). Benchmarks for Science Literacy. New York: Oxford University Press.
[2] Ainsworth, S., Bibby, P., Wood, D., Information
technology and multiple representations: New opportunities, new problems. Journal
of Information Technology for Teacher Education, 6, 1997, pp. 93-104.
[3]
Bowman,
D., Hodges, L., Allison, D., Wineman, J., The educational value
of an information-rich virtual environment. Presence, 8(3), 1999, pp. 317-331.
[4] Cuban, L. Teachers and Machines: The Classroom use of technology since 1920. New York: Teachers College Press, 1986.
[5]
Gentner, D.,
Toupin, C., Similarity and the
development of rules. Cognition, 65, 1998, pp. 177-211.
[6]
Hennessy,
S., Twigger, D., Driver, R., et. al. Design of
computer-augmented curriculum for mechanics. International Journal of Science
Education, 17(1), 1995, pp. 75-92.
[7] Holst, S. Directing learner attention with manipulation style. ACM CHI 96 Doctoral Consortium, Vancouver, CA, 1996.
[8] Johnson,
A., Moher, T., Ohlsson, S., Gillingham, M., The round earth
project: collaborative VR for conceptual learning. IEEE Computer Graphics and
Applications, Nov/Dec, 1999. pp. 60-69.
[9] Johnson,
A. Deploying VR in an elementary school - Pipe Dreams and Practical Realities,
In the Proceedings of IPT 2000, June 19-20, 2000, Ames, Iowa, CDROM
[10]
Larkin, J., Simon,
H., Why
a diagram is (sometimes) worth ten thousand words. Cognitive Science, 11, 1987,
pp. 65-99.
[11] Moher,
T., Johnson, A., Cho, Y., Lin, Y., Observation-based inquiry in a virtual
Ambient Environment, In the proceedings of the Fourth International Conference
of the Learning Sciences, June 14-17, 2000, Ann Arbor, MI, pp. 238-245.
[12]
Ohlsson, S. Abstract schemas. Educational Psychologist, 28(1), 1993, pp. 51-66.
[13] Ohlsson,
S., Moher, T., Johnson, A., How to teach children that the Earth is round In
the Proceedings of COGSCI 2000, Aug. 13-15, 2000, Philadelphia, PA, pp.
364-368.
[14] Osberg, K.,
Winn, W., Rose, H., Hollander, A., Hoffman, H., Char, P.,
The effect of having grade seven students construct virtual environments on
their comprehension of science. AERA Annual Meeting, Chicago, IL, May 1997.
[15]
Piaget, J.,
Inhelder, B., The child's conception
of space. London, UK: Bourledge & Kegan Paul, 1956.
[16] Presson, C., DeLange, N., Hazelrigg, M. Orientation specificity in spatial memory: what makes a path different from a map of the path? Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 1989, pp. 887-897.
[17] Salzman,
M., Dede, C., Loftin, R. B., Chen, J., A model for
understanding how virtual reality aids complex conceptual learning. Presence,
8(3), 1999, pp. 293-316.
[18]
Schank, R. Learning via multimedia computers. Communications of the ACM, 36(5),
1999, pp. 54-56.
[19]
Sweller, J. Cognitive load during problem solving: effects on learning.
Cognitive Science, 12, 1998, pp. 257-285.