by Henry Kelly

Henry Kelly is president of the Federation of American Scientists, headquartered in Washington, D.C. Kelly has worked on technology policy for the Congress and in the White House and is the Executive Director of the Learning Federation, a group of corporate and university research leaders dedicated to building a major national research program aimed at discovering ways to use advances in cognitive science and information technology to improve the quality, accessibility and affordability of learning.

After years of rhetoric about how information technology would revolutionize education, the nation’s education and training institutions have not delivered anything approaching the promised change. Most of the innovations that have occurred have been driven by students using word processors instead of typewriters and using the web as a supplement to library research. Students with web-literate teachers are able to get advice and submit papers by e-mail. But progress has stalled at this point. Even efforts to put university curriculum online seldom amount to much more than posting syllabi and lecture notes.

New financing mechanisms have done a good job of providing computer connectivity to schools and market forces will make hardware costs affordable in the years to come. But this hardware will be little more than ornaments in the back of 19th century classrooms unless our schools and universities manage an unprecedented change in their approach to innovation in their own core enterprise.

What Can We Aim For?
We’re so accustomed to the limitations imposed by today’s classroom instruction that it is hard to imagine anything else, but cognitive scientists tell us that there actually is a better way for people to learn. Their research argues that the best way to acquire and maintain expertise is to provide a framework that helps learners to organize information into a logical structure and to reinforce this logic through a series of inherently motivating experiences that fixes the knowledge in a learner’s mind. The goal is to keep a student motivated by maintaining a state of disequilibrium, where there is some dissonance between their concepts and the reality they are confronting. The dissonance should not be so great that the learner becomes frustrated or despairs.

Timely, informative assessment of a person’s expertise is also essential to this process. The measurements should be as valuable to the learner as to the instructor, helping the learner become aware of the success of his or her own learning strategies and approach to expertise.
It has been easy to ignore these recommendations because many seem implausibly expensive and impossible to implement. But what would happen if technology suddenly made them affordable?

What Do We
Need To Know?
The recommendations of modern cognitive science strongly resemble the learning environment that was universal before the worthy desire to provide universal education forced us to use classrooms: learning apprenticeships; learning through games that imitated adult behavior; learning through mistakes and trial and making mistakes corrected by experts; learning supported by tutors giving advice in the context of a task the learner was motivated to accomplish. What’s striking is how little we really know about these lost approaches to teaching. The most accomplished tutors in today’s society have little but their own experience to go on; few have been tutored themselves. Laboratories and field trips are highly artificial experiences that seldom allow real exploration or provide anything approaching the experience of a real practitioner.

How to think outside
of the classroom
The most fundamental research challenge is finding out how to design the most effective learning strategy for each subject and each student given that all problems of design and cost could be overcome. We know that computer games can be extraordinarily compelling, and that people will spend weeks mastering obscure skills that help them move to ever-higher levels of play. Some of these techniques can be used to motivate learning. Some subjects lend themselves to exploration and tutorials and some are better learned by curling up with a good book. Some learners thrive in groups and some are exasperated by them. Socratic dialogues driven by experience and lots of questions are undoubtedly valuable in some situations for some students. The painful reality is that we really don’t know much about when or how to use these new approaches.

How to build
and use simulations
Many of the new approaches to learning depend on finding ways to make exploration and discovery practical, and for most purposes that means building plausible simulations and virtual environments. It is now possible to run astoundingly realistic simulations on computers that cost only a few hundred dollars.

The animations that are integral parts of most modern action films provide a good example of the kinds of simulations that are feasible and will function on inexpensive computers in the near future. In the movie The Hollow Man, a figure moves around as layers of skin, muscle and internal organs peel away. The heart beats, muscles and ligaments stretch in completely plausible ways because the animation was built by modeling every major bone, muscle and organ. Jurassic Park created complete Jurassic ecosystems; Gladiator included a model of most of 2nd century Rome.

Highly sophisticated simulation visualization tools are used in research and engineering. In a growing number of cases, these simulations are the primary way information and insight are encapsulated. But classroom teachers, on the other hand, have none of these advantages. A major reason for the difference is that each class is, in effect, a cottage industry; each burgher is expected to build his or her own tools. Moving a complex simulation to an instructional environment, much less building one from scratch, is an extraordinarily complex and expensive task. It’s also a rather pointless task unless accompanied by innovations in pedagogy and a host of other support systems. No individual instructor, not even one equipped with a National Science Foundation grant, is going to be able to accomplish this.

How to answer questions and encourage discussions
There’s overwhelming evidence that dialogue and discussion between learners and experts is a powerful way to build knowledge. Classrooms make it difficult or impossible to hold a sustained dialogue with anyone. During a typical hour lecture to a class of thirty, there are only three questions. An experienced tutor generates thirty questions or more per hour for an individual student. When you’ve crashed your simulated airplane, you probably have a lot of questions and are likely to pay close attention to the answer.

Designing an artificial system that can provide accurate, timely answers to multi-dimensional questions (such as, What does this do?) is an enormously complex problem. The problem is compounded if the system must adjust the answer to the context of the question (What was the learner trying to do?), the state of the individual learner (What is the person’s level of understanding, his state of mind?), and reflect strategic goals of the instructional design (Should a factual answer be given to a question or should dialogue be encouraged to help the learner discover the answer?). Part of the design goal, of course, is to build a system smart enough to know when to summon human assistance and what information to provide the human before he or she intervenes. A significant amount of research in this area has been conducted by groups designing new search engines, help systems, web-based retail sites and support systems for intelligence analysts. Unfortunately, virtually none of these advances have migrated to education.

How to personalize learning
Good tutors know a lot about their students. This knowledge includes both understanding of the learner’s background and the learner’s grasp of a wide variety of concepts. It also means carefully watching the learner’s reactions to a situation—is the learner frustrated, bored, angry or tired? It’s possible for information systems to capture much of this data and store it as a detailed record of the learner’s interests, learning style and other factors. This record, much of which should be as private as medical records, can be invaluable for future teaching—both human and automated. The information can also provide information critical for adjusting the pace or style of artificial tutors.

What Are The Showstoppers?
Obviously there is an enormous amount that we don’t know about how to build and use new learning systems. It’s probably the most difficult research challenge the nation has ever faced, but it’s also the most important. How then to account for the complete absence of any serious program of research and invention in government or in industry? We invest nearly a trillion dollars a year in education and training and $5.5 billion on computers and other information hardware for education, but we spend, at most, $100 million addressing the critical problems of learning technology research. We probably spend more on improving pet food.

A number of factors explain this bleak situation. The core of the problem is that education and training institutions at every level have no experience in research on their own core business of teaching, and no competence in managing change. There’s no market for innovation in institutions that are unwilling to engage in the kind of deep examination of the real needs of their clients or willing to face the kinds of wrenching transformations in jobs and management needed to achieve real gains.

The fear factor on the part of education managers is understandable. The changes are likely to be dramatic. A generation ago, a family physician was in a similar situation, relying just on personal knowledge and what could be carried in a little black bag. But it would be unthinkable today to imagine a physician not backed by a vast body of research and a phalanx of specialists and sophisticated equipment.

Even though teachers themselves are strong advocates of change, the education establishment has been lukewarm in its support for research and innovation—in part because of its deep suspicion that innovation is the first step to a corporate system that will replace the human dimension of teaching and leave students languishing in darkened rooms staring at screens managed by corporate behemoths. In fact, the absence of a coherent national program of research and evaluation is more likely to lead to unattractive outcomes and maintain the mass production model of instruction.

What Is The Next Move?
(And Whose Move Is It?)
One critical stumbling block in the path of educational innovation is the absence of a coherent program of research—or even an outline of priorities, for that matter—that comes close to the scope or scale of research focused on health care, energy and environmental technologies, or other areas of great public interest and concern. With administrative management from the Federation of American Scientists, a group of the nation’s leading researchers in the learning sciences, along with software designers and private sector and educational personnel, have formed the Learning Federation Project. Its first goal is to develop such a plan. It will be essential that government and business move aggressively to support the needed research. It’s also essential that the nation’s education and training institutions find a way to make the dramatic changes needed to take full advantage of the new technologies and learning strategies that will emerge. And all of this requires leadership and significant new resources.

One promising new direction is offered by The Digital Opportunity Investment Trust*, a proposed initiative that would sponsor research in learning science and technology and support large-scale digitization and software development projects aimed at making digital materials available to broad public audiences. It would finance this work with revenue from auctions and fees for licenses to the publicly owned electromagnetic spectrum (the frequencies that transmit radio and television signals, for example). A concept like this is critical for capturing public imagination and lifting the debate over whether we can achieve dramatic gains in learning out of the research labs of a few enthusiastic scholars. Of course there are risks, but there’s a very real possibility that the investment would translate into dramatic gains in the quality, the affordability and the accessibility of learning worldwide during the next two decades. It would be hard to find a risk that’s more worth taking.

* More information about the Digital Opportunity Investment Trust can be found in the Fall 2001 Carnegie Reporter, (also available at www.carnegie.org), in A Digital Gift to the Nation: Fulfilling the Promise of the Digital and Internet Age, by Lawrence K. Grossman and Newton N. Minow, ©2001 The Century Foundation Press, and at www.digitalpromise.org.

Copyright information | Masthead | Carnegie Corporation of New York web site