Learning Science through Collaborative Visualization over the Internet1

Roy D. Pea
Stanford University ([email protected])
Director, Stanford Center for Innovations in Learning

"...If the network idea should prove to do for education which a few have envisioned . . . and if all
minds should prove to be responsive, surely the boon to humankind would be beyond measure." J.C.R.
Licklider & Robert Taylor (1968)

Abstract
Ten years ago, we launched the Learning through Collaborative Visualization, or CoVis
Project. "Collaborative visualization" refers to development of scientific knowledge that is
mediated by scientific visualization tools in a collaborative learning context. Funded by the
National Science Foundation as an advanced networking testbed, our partnership of
Northwestern University, Bellcore, Ameritech, the Exploratorium Science Museum in San
Francisco, and the University of Illinois at Urbana-Champaign's Atmospheric Sciences
Department/National Center for Supercomputing Applications (NCSA) sought to design,
implement and research the promises and problems of a distributed multimedia science
learning environment that used broadband desktop videoconferencing and screen sharing,
scientific visualization tools and distributed datasets, virtual field trips, scientist telementoring,
and a Collaboratory Notebook for enabling project-based learning of science in the high school
using these distributed human and technical resources. Our project vision was to establish
collaborative technology learning environments, or "collaboratories" that would enable project-
enhanced science learning among remote project partners using advanced telecommunication
networks. For example, our collaboration with NCSA scientists provided learners with access
to subject-matter experts, visualization tools and vast databases in the field of atmospheric
sciences. Virtual visits using wireless video over the Internet to Exploratorium exhibits helped
motivate student questions about central scientific phenomena.
At its peak of use, the CoVis network was in use by thousands of teachers throughout the
United States. The project developments were transitioned in 1997 as a set of resources that

1
Nobel Symposium (NS 120), “Virtual Museums and Public Understanding of Science and Culture,” May 26-29,
2002, Stockholm, Sweden. The CoVis Project has been funded by National Science Foundation Grants #9253462
and #9454729, Illinois Board of Higher Education Eisenhower grants, and industrial partners Ameritech and
Bellcore. We are grateful for hardware and/or software contributions by Aldus, Apple Computer, Farallon
Computing, Sony Corporation, Spyglass, and Sun Microsystems. Particular thanks to my CoVis faculty
colleagues Louis Gomez and Danny Edelson, and graduate students Laura D’Amico, Barry Fishman, Douglas
Gordin, Sam Kwon, Steven McGee, Kevin O’Neill, Joseph Polman, and Greg Shrader. Please see
http://www.covis.nwu.edu/ for CoVis contributors and collaborators. Other supports since the early CoVis era
were grants for WorldWatcher (#9720687), the Living Curriculum Project (#9720423) and the extensive
curriculum development and urban teacher professional development activities of the LeTUS Center (#9720383).
continue to be elaborated in the LeTUS Center for Learning Technologies in Urban Schools
based at Northwestern and University of Michigan, which is investigating in partnership with
the Chicago and Detroit public school systems how to support urban school teachers in using
project-based inquiry as their core approach to teaching earth and environmental sciences at
the middle and high school level.
In this chapter, after providing a pedagogical background that lays out the rationale for
bringing together developments in the sciences of learning with advances in high performance
computing and communications in the CoVis Project, I will highlight our design research
methodologies for making the powerful features of professional scientific visualization tools
usable to students and teachers in WorldWatcher.
Introduction
How will learning take place as we stride into the third millennium? I believe we should
emphasize new models for learning that go beyond the antiquated conceptions of education
found in most teacher-centered, formally organized classrooms. Reforms seeking to improve
education would productively focus on understanding and supporting learning in ways that pay
close attention to the skills of social interaction and participation in new activities that
currently constitute lifelong learning in working and living communities. In contrast to
learning-before-doing— the model of most educational settings—I would urge attention to
learning-in-doing, a model where learners are increasingly involved in the authentic practices
of communities through learning conversations and activities involving expert practitioners,
educators and peers. Learning-in-doing requires interactions between groups that traditionally
have been separated by the institutional boundaries of such institutions as school, work,
university, and home. Scientific and business workgroup practices using Internet and high-
bandwidth services recognize that not all partners necessary to an interaction can be co-
located— and education needs to exploit the same insight.
A convergence of socio-technical developments has led to early successes and enthusiasm
about broad-scale potentials for improving learning, education and schooling in these ways.
These developments coalesce in the emergence of “virtual learning environments” or “online
learning communities.” Virtual learning environments may be constituted within or across
classrooms or campuses, within or between businesses (or homes), or involving participants in
heterogeneous settings, such as school-home-community, or school-workplace-university. By
collapsing spatio-temporal barriers through the construction of virtual learning communities,
we may enable greater intimacy and authenticity in the learning process. Such technologies
can be used to facilitate the return of learning models that existed with considerable success
prior to formal schooling, such as apprenticeship, long-term mentoring, and collaborative
groups that learn through work on projects. Expertise that is now geographically dispersed and
isolated will become increasingly interconnected to the benefit of lifelong learning. Virtual
museums, the emphasis of this workshop, offer yet another context for the consideration of
virtual learning environment design and research.
The primary converging trends leading to virtual learning communities have been: (1) the
movement toward more socially-situated conceptions of learning, toward viewing intelligence
as a distributed achievement rather than as a property of individual minds; (2) the advent and
rapid growth of Internet use in support of both informal learning and more formal learning
communities; and (3) rethinking appropriate roles for the teacher in effective learning
environments. Together, these trends may well have synergistic effects far greater than any of

Nobel Symposium (2002) R. D. Pea 2

them could individually for deepening learner understanding and rendering learning more
relevant for enabling the learners’ participation in cultural practices. Participation in cultural
practices is, after all, the kind of “transfer” outcome from one’s involvement in learning
activities that we ultimately seek. Acquisition of knowledge per se is subsidiary to this aim.
Information technologies play multiple roles in these socio-technical developments in serving:
(1) As a meta-representational substrate—in which authors create media-rich documents using
symbol systems such as language, diagrams, video, audio, mathematical notations, and
scientific or data visualizations, and make conceptual linkages between documents or
document components (as in hypermedia) in ways that were awkward or impossible in non-
computational systems like print (34).
(2) As a communication channel for establishing highly interactive conversations among
individuals distributed across space and time (33). These communications in “media spaces”
may transcend in some respects what is possible in face-to-face encounters (sometimes
characterized as ‘beyond being there’), e.g., when the groupware in support of them allows for
joint activities such as computer-assisted design in the production of a common artifact.
(3) As interface to individual, group, and cultural memories—archives of information,
knowledge, and representations of past activities that can be accessed, drawn upon, and
extended as needed when new problems arise, or when reflection leads to new insights.
(4) To establish spaces and places in which individuals come together in real-time
(synchronously) or at different times (asynchronously) to collectively engage in some activity
together (e.g., 1, 20). These social activities are mediated by informational representations of
space (e.g., a graphical city comprised of buildings incorporating rooms in which individuals
throughout the globe may convene through their computers to have meetings, discuss issues,
learn, play, and socialize). The persistence of persons, documents, and rooms for meeting in
cyberspace is being exploited in multi-user virtual environments, or “virtual worlds.”
(5) As cognitive tools for augmenting human performance in complex tasks and for learning2.
Some theorists of mind view the concept of “distributed intelligence” (DI) as crucial for
conceptualizing learning and education (e.g., 7, 32, 36). In this perspective, the main sense of
DI emerges from the image of people-in-action, and in their activity, we see the configuring of
distributed intelligence. Activity is enabled by intelligence, but that intelligence is distributed
across people, environments, and situations, rather than being viewed as a resident possession
of the individual embodied mind. Intelligence is accomplished rather than possessed; the
accomplishments of human activity routinely engage not only an individual’s memories but
also the intelligence embodied in the design of artifacts and in the social resources often
available in situations. As I noted in my chapter on distributed intelligence and education:
“There are both social and material dimensions of this distribution. The social distribution of
intelligence comes from its construction in activities such as the guided participation in joint action
common to parent-child interaction or apprenticeship, or through people’s collaborative efforts to
achieve shared aims. The material distribution of intelligence originates in the situated invention of
uses of aspects of the environment or the exploitation of the affordances of designed artifacts, either of
which may contribute to supporting the achievement of an activity’s purpose” (Pea, 32, p. 50).

2
Early and influential expressions of this perspective were provided by Licklider (24), and by Engelbart (12) in
his seminal work at SRI International. These efforts lead eventually to the “personal computer” (21). See
Rheingold (38) for a fine history of computers as “tools for thought”; see Pea (28, 29) for socio-historical and
developmental considerations.

Nobel Symposium (2002) R. D. Pea 3

 With the explosive growth of the Internet, now connecting hundreds of millions of people
around the globe, the relevance of distributed intelligence to the design of interactive learning
environments has become much clearer than it was in 1993. And its implications have become
all the more problematic, for those who think of designing learning activities and assessing
learning outcomes in education as tool-free, individual cognition. Throughout the world,
virtual learning communities are being established and growing in shape to fit the ecological
niches found necessary to their members’ needs. They may be as small as a few people
scattered over the net in chat rooms devoted to discussing a mutual reading or a prospective
stock investment. Or they may be as large as thousands of individuals coupled together in a
virtual fieldtrip, exploring some part of the globe through teleconferencing video connections
or updated datasets and commentaries that together establish a knowledge-building community
about a learning topic such as global warming (14). Moreover, they are vast as the 50 million
E-Bay buyers, the largest online global marketplace.
In this paper, I will describe the CoVis project that exemplifies the application of distributed
intelligence to problems of designing learning environments. I co-directed the Learning
through Collaborative Visualization (CoVis) Project, with my colleague Professor Louis
Gomez of Northwestern University from 1992-1997, and have been involved in collaboration
and consulting capacity after that era with its newer developments, especially with Professor
Daniel Edelson on WorldWatcher. CoVis provides a broad range of lessons for how
distributed intelligence may be effectively supported through the uses of learner-centered
collaboration technologies and scientific visualization tools.
Virtual learning communities
First, I should say more about the concept of “virtual learning community.” Virtual learning
communities (VLCs) are first distinguished by their “virtuality”—face-to-face teaching
instructions or informal learning activities are actual, not virtual. What makes the learning
communities “virtual” is that they are mediated over computers and using computer networks.
The applications involved in the communications among participants in VLCs may be as wide-
ranging as electronic mail, newsgroups, chat rooms, listservs, remote screen-sharing
collaborative work applications, and multi-user virtual environments.
How about the concept of “learning”? For a socio-technical arrangement to constitute a
“virtual learning community,” its membership should explicitly express “learning” as an
objective of activity participation over the net. Otherwise, any telecommunications-involving
activity would be a VLC, since many of us would argue that most any activity in which one
engages may correlate with learning of some kind, even if only tacitly.
Finally—and most problematically—is the notion of “community.” What makes one?
Certainly each of us belongs to many communities, is peripheral to others, and does not belong
to yet many others. Yet, we cannot be too thrilled that the concept is difficult for sociologists,
anthropologists, and other social scientists concerned with social phenomena. The core sense
of a community, for our purposes, is a collective of participants with significant persistence
whose members align around some common activities. Communities do things, of virtually
infinite variety—they may learn, advance knowledge, collect things, provide help to one
another, read books, do crossword puzzles, design jets, watch sporting events together. In
addition, typically, such communities are comprised of participants who tend to have access to
encounters with one another. Information technologies change the ways in which such

Nobel Symposium (2002) R. D. Pea 4

communities can be formed and sustained in their activities (1, 43). Let us now consider the
design of CoVis as virtual learning community.
The CoVis Project: Learning through Collaborative Visualization
How can the perspective of distributed intelligence be used to guide the design of virtual
learning communities for science? How can we use the different affordances of information
technologies to support the material and social distribution of intelligence that will better
enable the accomplishments of science learning?
In 1992, with NSF funding1 and support from industry partners, my colleague Louis Gomez
and I created the Learning through Collaborative Visualization (CoVis) Project (31). We had a
vision for science educational reform through distributed science learning —taking cognitive,
social, technological, and scientific breakthroughs into account—which we sought to
implement and then empirically study as a community appropriates and evolves its uses. Our
development of the CoVis learning environment and its components was guided by a student-
question-centered, collaboration-focused pedagogy, and our commitment that classroom
science learning should more closely resemble the open-ended, inquiry-based approach of
science practice. As we put it in our January 1992 NSF proposal:
“We believe that science learning environments should look and act more like the collaborative,
connected work environments of scientists. To this end, teachers and students need ways to reduce the
complexity of getting access to resources that are inaccessible locally. These resources include human
expertise in the form of other teachers, scientists and graduate students in business, industry, research
settings, and other learners. They also include tools, instrumentation, hands-on materials and labs,
museum exhibits, and computing and telecommunications infrastructures...
[We] aim: To examine how geographically dispersed teachers, students, and collaborators can integrally
and readily use advanced information technologies to facilitate the types of collaboration and
communication demanded by project-enhanced science learning.
We argue that applications of advanced technologies provide American educators with critical levers for
promoting cognitive apprenticeships in science learning. We aim to build the next generation of
infrastructure for new forms of science learning and teaching, and create a national model for the kinds
of distributed multimedia science learning environments supportable with the future NREN [National
Research and Education Network, later called the NII, or National Information Infrastructure].... We
argue that these new [High Performance Computing and Communications] technologies can provide the
backbone for the transitioning process from didactic science teaching to cognitive apprenticeships in
project-enhanced science learning.... Students need to learn and do science in context of real problems
and with sophisticated tools.”
After its first five years of invention, empirical studies of use patterns, and design iterations,
CoVis came to provide a testbed “collaboratory” (5, 23) for learning science by doing science
over the Internet—that directly worked with over 100 teachers and thousands of students in 54
middle and high schools in 11 states, with thousands of other teachers using CoVis resources.
CoVis schools were connected with high-speed Internet access (commonly T-1) that we used
as a critical leveraging technology to develop and evolve through use a wide array of new
resources for learning. We provided teachers and students with a software suite of learner-
centered tools for doing project-based collaborative learning. Student projects often involved
pursuing inquiries using scientific visualization of datasets in the atmospheric sciences and
geosciences, and use of a full set of Internet communication tools including e-mail, Usenet
news, Gopher, Web browsers, and desktop video-conferencing, that enabled new kinds of
learning relationships. Participating classrooms typically had six or more desktop computers,
and students had individual net accounts. Our ambitious aim was integral use of high
performance computing and communications tools for project-enhanced science learning.

Nobel Symposium (2002) R. D. Pea 5

 Our assessment of learning and teaching needs led us to project-enhanced science (39) as a
fundamental pedagogy for achieving deeper learner understanding and distributed intelligence
among the science learning community. This approach to science education was based upon
several contemporary research strands in the cognitive sciences of learning (2). Chief among
these is the idea that, long before and outside of formal education, people have learned through
participation in “communities of practice” (e.g., 22, 41). In contrast to the common “course
delivery” models of instruction and distance learning, the advantage of such communities is
that learning is situated with respect to community-based goals and activities in which
knowledge is developed and used (4, 30, 40). In the classroom, this benefit may take the form
of “cognitive apprenticeship” (8), with students guided, both by their teachers and by remote
mentors, to think about science in many of the fundamental ways that scientists do. In
designing CoVis learning environments, we acknowledged the large gaps between students and
scientists in motivation, purposes, knowledge, and physical distance. The point of science
education is not to take the "little scientist" that Piaget said was in every child and create an
actual scientist. Heeding these gaps, we sought to design settings in which students could
engage as "legitimate peripheral participants" in “communities of practice” for science.
In the community of science practice, the computational tools used—such as scientific
visualization and modeling environments—as well as the data collected for investigations are
critical to the scientists' abilities to formulate, think about, and work on their problems. In
designing the CoVis testbed, we sought to take advantage of this reliance by providing students
with learning-appropriate ways to access the same real-time weather data and globally gridded
archival earth sciences and environmental data that the scientists study, using tools that we
modeled on those used by scientists (10, 15, 18, 19). Scaffolding access to these tools and the
scientists and other professionals that use them is a central design technique we used to “seed”
a sense of legitimate peripheral participation for students in the CoVis learning testbed. There
are other recent efforts to bridge the science education-scientific practice "gap," including the
GLOBE Project, Global Lab, GREEN, Earthwatch, Kids as Global Scientists, Hands-On
Universe, and BioKids, which have developed related models, of what have been called
"student-scientist" partnerships (6, 40).
A crucial resource that made broad scale CoVis participation in our testbed work out was the
ongoing nature of professional development for teachers, supported in the mid-90’s by our
CoVis Geosciences Web Server (http://www.covis.nwu.edu/), one of the first educational web
projects, with a focus on ironing out the details of effective engagement for learners in project-
oriented pedagogy. This focus made clear how important “fingertip effects” are for teachers’
appropriation of innovations such as CoVis, and as I will describe, we created “model”
learning activities for interschool project collaborations in which students’ learning can be
guided. I will highlight some of the technical features and social affordances of designs that
made more fingertip-accessible such new classroom activities as collaborative knowledge
inquiry, telementoring, and expressive and interpretive uses of scientific visualizations as a
representational vehicle for sense making in science.
CoVis has taken a design-intensive and iterative re-design approach to developing the
learning and teaching software, curriculum activities, and participating groups (schools,
education researchers, content experts and telementors, science educators and science
education institutions) that became part of the “CoVis testbed” (10,13,35). It is challenging to
briefly describe this process of mutual influence among researchers, technology specialists,
and educators in our work, but overall, we shifted from building new tools and resources with

Nobel Symposium (2002) R. D. Pea 6

participatory design advice from educators, to learning more about their situations of
technology use so that we could co-evolve activity structures in which sustained use of
telecommunications in support of project-based science learning would become commonplace.
Because much of the effort in establishing a distributed science learning “collaboratory” has
required technology development and integration, it is not surprising that much emphasis has
been on the “material” aspects of distributed intelligence, as embodied in the affordances of the
properties of learner-centered scientific visualization software (e.g., 10,17,18), and groupware
(9,11), datasets, and other media designed for learner and teacher use. The following table
provides a few illustrations for how design of material aspects of the CoVis distributed
intelligence (DI) systems in which learners and others now participate has fostered, and in
many cases, enabled, the kinds of activities central to our vision of distributed science learning.
Material aspects of DI used in design Social aspects of DI that are enabled
(1) Time-shifting communications (e.g., Facilitates broad access to distributed group inquiry;
email, listservs, newsgroups, web Enables collaboration across global time zones.
site annotations).
(2) Space-collapsing communications. Facilitates broad access to distributed group inquiry;
Enables virtual field trips to remote places.
(3) “Semantically typed” hypermedia More readily-achieved structured scientific inquiry;
links in our CoVis Collaboratory Simplifies tracking learners’ questioning and project
Notebook (e.g., Question, inquiry processes that are central to constructivist
Conjecture, Evidence For, Evidence pedagogy, and navigating the information products of a
Against). group’s knowledge-building project.
(4) Archival memory for Creates ‘living’ database of knowledge building
communication records. community;
Builds persistent and re-usable database of scientists and
other mentors for distributed learning communities.
(5) “Scaffolding” front-ends to scientific By starting with qualitative inquiries, enables broad
visualization tools (e.g. geographic learner access to complex science topics and systems;
context map overlay, semantic units, By making scientists’ tacit knowledge explicit in the tool
and semantic constraints on interface, makes it possible for novice learners to pursue
operators for manipulating their open-ended research questions (e.g. weather
visualizations). prediction; causal patterns affiliated with global
warming).
(6) CoVis telementoring database (e.g., Facilitates remote mentor participation and teacher
forms for mentors to “profile identification of appropriate mentors.
expertise” and for teachers to search
for it; automatic message handling
for project participants, mentors,
teachers)
In our initial work in developing distributed intelligent systems for science learning, we thus
centered on creating tools and resources for learners. Since 1995, a great deal of our attention
and effort shifted to the “social” aspects of distributed intelligence that are required to launch
and sustain these efforts to reform science education toward a learner-centered, project-based
pedagogy. Driven by teacher needs and a massive scale-up from 6 to over 100 teachers in
geographically diverse sites from 1993 to 1995, we focused on creating activity structures that
contextualized the uses of tools and resources by “seeding” the CoVis GeoSciences Web
Server (http://www.covis.nwu.edu/geosciences/) with distributed science learning activities,

Nobel Symposium (2002) R. D. Pea 7

providing models for teachers to appropriate and adapt to their local circumstances (13). We
called our sets of activity structures around a common topic “CIAs” or CoVis Interschool
Activities (http://www.covis.nwu.edu/ geosciences/activities). They included multi-week
resources on the topics of global climate change, water quality, land use management, weather
prediction, and soil science. The World Wide Web was the primary medium through which
these activities were shared with teachers, many of whom helped design them in face-to-face
summer workshops and refine them in the context of their ongoing teaching. CoVis teachers
using these activities formed communities of interest through listservs, to guide and support
one another as they sought to interpret and extend the CoVis vision in new ways.
These predictably scheduled interschool activities help teachers think concretely about
different aspects of project science, educational telecommunications, and collaboration,
because each involve opportunities for telementoring, student collaborative learning, uses of
the many Internet communication tools we have described, and such CoVis web resources as
scientific visualization environments, datasets, collaborative support, and sample assessment
rubrics. These activities provide a curriculum project "shell" or activity structure within which
teachers can create specific projects that can be done across classrooms or within a classroom.
Our central design challenge has been creating activities so as to provide enough structure so
they may be feasibly conducted in a classroom context with limited time and workable
classroom procedures, while also fostering enough freedom in the intellectual environment so
that learners' questions may be developed as novel project directions (44).
Characterizing our aims and achievements in the design and implementation of scientific
visualization environments for middle and high school learning in the CoVis Project and
beyond helps make concrete what this work has been like.
The WorldWatcher Global Visualization Environment for Science Education3
Scientific visualization has dramatically changed scientists' practices in recent years by
“using vision to think,” exploiting human visual pattern perception for complex investigations
in large datasets (e.g., 3, 5, 26, 43). Scientific visualization provides an image rendered
through high-speed computer graphics that is based on a numerical dataset that describes some
quantity in the world (e.g., global temperatures). Our goal in developing scientific
visualization environments for pre-college education was to investigate the socio-technical
conditions under which scientific visualization, integrated into inquiry-based learning
activities, could enable students of diverse abilities (not only the most motivated and talented)
to develop an understanding of complex scientific phenomena such as climate, weather and
global climate change. We say "socio-technical" conditions because key components of the
effective use of learning technologies include software functions that are readily learnable and
usable—but also teacher-guided activities that are part of the social surround of technology
use.

3
For the WorldWatcher Project initiated by Daniel Edelson and Roy Pea at Northwestern University as a
development of the CoVis Project (and also supported by NSF Grant #9453715), we thank Co-PI Louis Gomez,
lead programmers Douglas Gordin, Joey Gray, Brian Clark and Eric Russell, and the contributions of many
others, including Eddy Ameen, Tajuana Bates, Jonathan Berkowitz, Matthew Brandyberry, Matthew Brown,
Kylene Chinsio, Elaine Coleman, Christopher DiGiano, Jason Gandhi, Duane Griffin, Eric Hanson, Ben Johnson,
Steve Juh, Ryan Kowalczyk, Danielle Lessovitz, Ray Liu, Adam Murphy, Kathleen Schwille, Michael Smith,
Michael Taber, Adam Tarnoff, Maya Wallace, Viktoria Wang, and Eshanthika Wijesinha. WorldWatcher
research was supported by NSF grants #9453715 and #972068, by a SMETE Postdoctoral Fellowship
(#9714534), and by Sun Microsystems, ESRI, and Precision Farming Enterprises.

Nobel Symposium (2002) R. D. Pea 8

 Our early investigations in CoVis revealed the need for specific supportive structures in a
scientific visualization software architecture that can provide what learners need (15, 17, 18).
Unfortunately, the tools and techniques scientists use presuppose significant prior knowledge
in the canons of scientific inquiry, data exploration and systematic inquiry, and in strategies for
search, analysis, and transformation of data using the capabilities of scientific visualization
software programs. Figure 1 exemplifies a scientific visualization environment developed for
scientists, not science learners.

Figure 1: A visualization of surface temperature data for the Northern Hemisphere displayed by Transform, a
powerful, general-purpose visualization environment widely used by scientific researchers (circa 1992)
WorldWatcher is a scientific visualization environment designed primarily for supporting
student investigations of two-dimensional, gridded scientific data, and based on the
ClimateWatcher software we first released in April 1996 (10). WorldWatcher built upon
ClimateWatcher's aim to provide an accessible and supportive environment for students to
explore, interpret, and analyze scientific data about the transfer of energy through the earth-
atmosphere system and climate change in ways akin to scientists' inquiry activities. Data is
distributed with WorldWatcher in data libraries that support educational activities centered
around specific datasets, as used in interpretive, analytic and expressive visualization activities
(http://www.worldwatcher.nwu.edu/).
Users of WorldWatcher can now create dynamic color visualizations of many different data
types beyond those initial topics (from NASA and other public domain sources), and can
import their own geo-gridded, as well as point datasets (using a standard spreadsheet format).
WorldWatcher datasets for global climate change (insolation, albedo, absorbed/reflected solar
energy, surface temperature, greenhouse effect/increase, outgoing long-wave radiation, net
energy balance) are complemented by human and physical geography data to allow students to
examine the causes and implications of climate change. Global climate data on precipitation
and relative humidity are provided, as are physical geography data on elevation/bathymetry,
soil type, dominant vegetation and ground cover, and plant energy absorption (FPAR).
Students can also access global information on population magnitude and density, carbon
emissions, and national boundaries.
Student access to the collections of WorldWatcher datasets is supported in two ways: (1) by

Nobel Symposium (2002) R. D. Pea 9

schematic diagrams that graphically depict relationships among variables—such as The Energy
Balance in Figure 2 — to foster students’ conceptual development and which can be
clicked

Figure 2: The Earth Energy Balance diagram user interface to WorldWatcher datasets

through directly to datasets for visualization, or by structured online notebook activities created
by the teacher (see below).
WorldWatcher enables learners to examine and interpret datasets created by the scientific
community and to create their own data using built-in arithmetic operations and climate
models. WorldWatcher offers display features of visualization environments designed for
scientific researchers, such as depicting 2-D global data in the form of false-color maps, but to
provide geographical context for learners, these maps are displayed with latitude-longitude
markings and an optional continent outline overlay (Figure 3). When a user interacts with the
WorldWatcher visualization, a continually updating readout tracks the user's mouse as it moves
over an image and displays current latitude, longitude, country or state/province, and data
values. Users tailor visualizations by modifying color-scheme, mapping of colors to numerical
values, spatial resolution, and magnification. WorldWatcher provides statistical summaries for
user-selected regions4 or whole maps.
Beyond these statistical summaries, WorldWatcher offers unary and binary mathematical
operations for mathematical data analysis. Within an image, users can add, subtract, multiply,
or divide all the values in a region or an entire image by a constant. They can normalize values
in an image, and use the blackbody equation to convert energy values to temperature and
temperature values to energy. WorldWatcher binary operators—enabling users to apply a
function at each location in two images—are addition, subtraction, multiplication, division,
maximum, minimum, and correlation, with the result displayed as a new visualization.

4
Regions can be selected using rectangular and irregular region selection tools, as well as by specifying
geographic areas by name (e.g., South America in Figure 3), or data values by range (e.g., all areas with
temperatures between 60 and 75 degrees F.).

Nobel Symposium (2002) R. D. Pea 10

 Figure 3: A WorldWatcher visualization window with key to what is represented
For example, examining the difference between global surface temperature values in January
and July (all three visualizations included as Figure 4), students can observe the stability of
ocean temperature and equatorial lands, in contrast to non-equatorial land masses, and the
polar regions especially.

Figure 4: Three WorldWatcher visualization windows: Surface temperature for Jan-

94, July-94 and the difference between Jan-94 and July-94

The functions of the WorldWatcher visualization environment allow students to view these
data in the form of color maps at a variety of spatial and temporal resolutions, and to animate
change over time. The activities are supported by a multimedia database of background and
explanatory materials focusing on science topics and scientific visualization techniques.

Nobel Symposium (2002) R. D. Pea 11

WorldWatcher offers a simple on-line notebook feature that supports text, diagrams,
multimedia, and clickable links to open specific datasets. Teachers are using the notebook to
design and disseminate activities, and students are using it to create projects and record their
cumulative progress in their projects, because they can annotate their visualizations
In today’s uses of WorldWatcher advanced by LeTUS, the Center for Learning Technologies
in Urban Schools (http://www.letus.org/), there are both middle school and high school
curricula available, developed through a research center partnership, led by Northwestern
University and involving the University of Michigan and 100 public schools throughout
Chicago and Detroit. Scientific visualization is integrated into inquiry-oriented class work
with hands-on labs, teamwork, and class discussions to develop learner understanding of both
the scientific and social issues affiliated with changes in the environment. For middle school,
The Global Warming Project is an eight-week unit centered on scientific factors contributing
to the controversial global warming debate. The unit closes with student investigations
considering scientific evidence and social concerns as they appraise potential solutions to
issues raised by the global warming debate. Looking At The Environment (LATE) is a year
long, inquiry-based, visually intensive environmental science curriculum for high school
centered on three key issues: (1) The relationship between population growth and resource
availability; (2) Electricity generation and meeting the demand for energy; and (3) Managing
water resources for agricultural use and human consumption.
Conclusion
In the CoVis Project, we had several key objectives:
(1) To demonstrate the feasibility, opportunities, and challenges of establishing a distributed
multimedia learning environment—the “CoVis testbed”—in which high school students could
learn science by doing science over the Internet, using real scientific data and scientific
visualization environments, as well as groupware and desktop videoconferencing tools for
collaborative learning and virtual visits to labs, museums and other schools;
(2) To create a collaborative partnership in the development of the CoVis testbed of
university researchers in the sciences of learning and learning environment design with a
science education museum, a university scientific research group (UIUC/NCSA), multiple
public schools, and industry partners;
(3) To investigate in detail how to make a powerfully general framework for scientific
visualization in learning and teaching that could adapt scientists’ tools, create software tools
and curriculum to exploit that framework, and study how teachers and learners could come to
use it productively in their learning and teaching activities.
When we began these efforts, the vision of establishing a high-end learning collaboratory for
students to learn science by doing science over the Internet, with university-scientist level
computing and communications resources, and engaging scientist telementors and science
education museums in “virtual visits” was considered improbable by many, and exceptionally
costly. Today, we see many of the features of our CoVis vision for science education
integrated in the day-to-day learning activities of urban classrooms, among the most
challenging settings for educational reform using technologies. While these achievements
have been gratifying, we have only begun to explore the transformative roles for advanced
computing and communications in supporting learning and teaching that taps the distributed
intelligence available, so that anyone, anytime, anywhere, and at any pace, can advance their
lifelong learning hopes and dreams.

Nobel Symposium (2002) R. D. Pea 12

References
1. Barab, S., Kling, R., & Gray, J. (in press). (Eds.). Designing virtual communities in the service of learning.
New York: Cambridge University Press.
2. Bransford, J.D., Brown, A.L., and Cocking, R.R. (2000). (Eds.) How people learn: Brain, mind, experience
and school (Expanded edition). Washington, DC: National Academy Press.
3. Brodie, K.W., Carpenter, L.A., Earnshaw, R.A., Gallop J.R., Hubbold, R.J., Mumford, A.M., Osland, C.D., &
Quarendon P. (1992). Scientific visualization. Berlin, Germany: Springer-Verlag.
4. Brown, J. S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational
Researcher, 18(1), 32-42.
5. Cerf, V. G., Cameron, A.G.W., Lederberg, J., Russell, C.T., Schatz, B.R., Shames, P.M.B., Sproull, L.S.,
Weller, R.A., Wulf, W.A. (1993). National Collaboratories: Applying information technology for scientific
research. Washington, DC: National Academy Press.
6. Cohen, K. C. (Ed.), (1997). Internet links for science education. New York, NY: Plenum Press.
7. Cole, M., & Engestrom, Y. (1993). A cultural-historical approach to distributed cognition. In G. Salomon
(Ed.). Distributed cognitions. New York: Cambridge University Press, pp. 1-46.
8. 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 instruction: Essays in honor of
Robert Glaser (pp. 453-494). Hillsdale, NJ: Erlbaum.
9. Edelson, D. C., Pea, R. D., & Gomez, L. (1995). 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 Publications.
10. Edelson, D. C., Gordin, D.N., & Pea, R. D. (1999). Addressing the challenges of inquiry-based learning
through technology and curriculum design. Journal of the Learning Sciences, 8(3&4), 391-450.
11. Edelson, D. C., Pea, R. D., & Gomez, L. M. (1996). The Collaboratory Notebook. Communications of the
Association for Computing Machinery, 39(4), 32-33.
12. Engelbart, D. (1963). A conceptual framework for augmenting man’s intellect. In P. W. Howerton & D. C.
Weeks (Eds.), Vistas in information-handling, Vol. 1. Washington, DC: Spartan Books, 1-29.
13. Gomez, L. M., Fishman., B.J., & Pea, R.D. (1998). The CoVis Project: Building a large scale science
education testbed. Interactive Learning Environments, 6(1-2), 59-92.
14. Gordin, D.N., Edelson, D.C., Gomez, L.M., Lento, M., & Pea, R.D. (January, 1996). Student conference on
global warming: A collaborative network-supported ecologically hierarchic geosciences curriculum,
Proceedings of the Fifth American Meteorological Society Education Symposium.
15. Gordin, D., Edelson, D., & Pea, R. D. (1995, January). The Greenhouse effect visualizer: A tool for the science
classroom. Proceedings of the Fourth American Meteorological Society Education Symposium.
16. Gordin, D., Gomez, L., Pea, R. D., & Fishman, B. (1996, December). Using the World Wide Web to build
learning communities in K-12. Journal of Computer-Mediated Communications, 2(3).
(http://www.usc.edu/dept/annenberg/vol2/issue3/gordin.html)
17. Gordin, D.N. & Pea, R.D. (1995). Prospects for scientific visualization as an educational technology. The
Journal of the Learning Sciences, 4(3), 249-279.
18. Gordin, D.N., Polman, J.L., & Pea, R. D. (1994). The Climate Visualizer: Sense-making through scientific
visualization, Journal of Science Education and Technology, 3, 203-226.
19. Hall, S.E., Sridhar, M.K. Ramamurthy, Wilhelmson, R.B., Plutchak, J., & Wojtowicz, D. (1996). An Online
Hybrid Multimedia Environment for K-12 Education in the Atmospheric Sciences. Proceedings of the Sixth
Symposium on Education. Long Beach, California, American Meteorological Society. (Abstract and Preprint).
20. Harrison, S., & Dourish, P. (1996, November). Re-place-ing space: The roles of place and space in
collaborative systems. Proceedings of the ACM Conference on Computer-Supported Cooperative Work,
CSCW ‘96. Boston, Massachusetts.
21. Kay, A. (1977, September). Microelectronics and the personal computer. Scientific American, 237(3), p. 230.
22. Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge: Cambridge
University Press.

Nobel Symposium (2002) R. D. Pea 13

23. Lederberg, J., & Uncapher, K. (Eds.). (1989). Towards a National Collaboratory: Report of an Invitational
Workshop at the Rockefeller University, March 17-18, 1989. Washington DC: National Science Foundation
Directorate for Computer and Information Science.
24. Licklider, J. C. R. (1960, March). Man-computer symbiosis. IRE Transactions on Human Factors in
Electronics, Vol. HFE-1, 4-11.
25. Licklider, J. C. R. & Taylor, R. (1968, April). The Computer as a Communication Device, In Science and
Technology: For the Technical Men in Management, No 76, pp. 21-31. (Also reprinted, In Memoriam: J.C.R.
Licklider: 1915-1990, Report #61, Systems Research Center, Digital Equipment Corporation, Palo Alto,
California, August 7, 1990, pp. 21-41.)
26. Nielson, G. M., Muller, H., Mueller, H., & Hagen, H. (1997). Scientific visualization: Overviews,
methodologies and techniques. New York: IEEE Computer Society.
27. O'Neill, D.K., Wagner, R. and Gomez, L.M. (1996, November). On-line mentors: Experimenting in Science
Class. Educational Leadership. Alexandria, VA: Association for Supervision and Curriculum Development.
28. Pea, R. D. (1985a). Beyond amplification: Using computers to reorganize human mental functioning.
Educational Psychologist, 20, 167-182.
29. Pea, R. D. (1985b). Integrating human and computer intelligence. In E. L. Klein (Ed.), New directions for child
development: No. 28, Children and computers (pp. 75-96). San Francisco: Jossey-Bass.
30. Pea, R. D. (1992). Augmenting the discourse of learning with computer-based learning environments. In E. de
Corte, M. Linn, & L. Verschaffel (Eds.), Computer-based learning environments and problem-solving (NATO
Series, subseries F: Computer and System Sciences). New York: Springer-Verlag GmbH (pp. 313-343).
31. Pea, R.D. (1993a). Distributed multi-media learning environments: The collaborative visualization project.
Communications of the ACM, 36(5), 60-63.
32. Pea, R. D. (1993b). Practices of distributed intelligence and designs for education. In G. Salomon (Ed.).
Distributed cognitions. New York: Cambridge University Press, pp. 47-87.
33. Pea, R. D. (1994). Seeing what we build together: Distributed multimedia learning environments for
transformative communications. Journal of the Learning Sciences, 3(3), 283-298.
34. Pea, R. D., & Gomez, L. (1992a). Distributed multimedia learning environments: Why and how? Interactive
Learning Environments, 2(2), 73-109.
35. Pea, R.D., Gomez, L. M., Edelson, D.C., Fishman, B. J., Gordin, D. N., & O’Neill, D. K. (1997). Science
education as a driver of cyberspace technology development. In K. C. Cohen (Ed.), Internet links for science
education (pp. 189-220). New York, NY: Plenum Press.
36. Perkins, D. N. (1993). Person-plus: A distributed view of thinking and learning. In G. Salomon (Ed.).
Distributed cognitions. New York: Cambridge University Press, pp. 88-110.
37. Ramamurthy, M., Wilhelmson, R., Hall, S., Plutchak, J., Sridhar, M., Fishman, B., Gordin, D., Pea, R. D., &
Gomez, L. M. (1996, January). CoVis Geosciences Web Server: An Internet-Based Resource for the K-12
community. Proceedings of the Fifth American Meteorological Society Education Symposium.
38. Rheingold, H. (1985). Tools for thought. New York: Simon and Schuster.
39. Ruopp, R., Gal, S., Drayton, B., and Pfister, M. (1993). LabNet: Toward a community of practice. Hillsdale,
NJ: Lawrence Erlbaum.
40. Songer, N.B. (1996). Exploring learning opportunities in coordinated network-enhanced classrooms: A case
study of Kids as Global Scientists. Journal of the Learning Sciences, 5(4), 297-327.
41. Wenger, E. (1999). Communities of practice: Learning, meaning, and identity. New York: Cambridge
University Press.
42. Wolff, R., & Yaeger, L. (1993). Visualization of natural phenomena. New York: Springer-Verlag.
43. Kim, A.J. (2000). Community building on the web. Berkeley, CA: Peachpit Press.
44. Polman, J. (2000). Designing project-based science: Connecting learners through guided inquiry. New York:
Teachers College Press.
45. National Academy of Sciences (1995). National science education standards. Washington, DC: National
Academy Press.

Nobel Symposium (2002) R. D. Pea 14