University Times

“I was among the first wave of psychologists that Bob Glaser recruited to come here” in 1967, says Charles Perfetti, head of the Learning Research and Development Center (LRDC), referring to the founding director, the late Robert Glaser. “Glaser was trying to build a research center that brought together people from psychology, education and other disciplines who would be interested in the study of learning and the development of ideas for scientifically based instruction. It was about what you can do to make learning more effective in schools. And it is still about that in many ways.”

LRDC has been marking 50 years with its distinguished speaker series and today, May 16, with a symposium, “New Directions in Research on Learning and Education.” The center has more than 175 employees, 40 of whom are faculty or associated faculty. Each faculty member has an appointment in an academic department; most are in the Department of Psychology in the Dietrich School of Arts and Sciences or the School of Education, but others are in law, business and engineering. And many have multiple appointments in other centers of scholarship throughout the University.

The mission of the LRDC has remained steady: “to advance the science of learning by bringing together leading researchers in the cognitive, social and educational sciences.” But the projects, programs and governing scientific viewpoints have taken unexpected turns throughout the years, often leading the region or the world in new research directions.

When Perfetti joined LRDC he was fresh out of the University of Michigan, sporting a psycholinguistics PhD, and LRDC was only four years old. “I had no obvious qualifications for being interested in education,” Perfetti allows, but says Glaser was willing to “roll the dice” and see how his basic research would extend to education.

“It was a small and scattered center at the time,” Perfetti recalls, with Glaser and psychology faculty who needed labs housed in the Mineral Industries Building (now replaced by stairs to Pitt’s upper dorms) and those more focused on the classroom, often with dual appointments in the School of Education, in a building at North Craig and Bayard streets. Perfetti began his LRDC career by investigating how the youngest students understood different aspects of language and why some children had trouble understanding what they had read. By 1985, he had shown that the root of the problem was not students’ difficulties with word comprehension but the degree to which they were able to recognize words quickly and accurately. Eventually, he was examining what parts of the brain were active in word recognition, demonstrating that even adult readers’ brains evoke spoken language while reading print. Most recently, he has compared brain studies of how different languages are read.

Perfetti’s own evolution as a researcher parallels the development of LRDC as a whole.

“A big part of the story is the cognitive revolution in psychology and education,” he says. “In the early days, behaviorism was the dominant perspective. It’s a focus on observing the behavior of students and teachers … and you didn’t make inferences about mental structures.” By the mid-1970s, the cognitive revolution had gained the upper hand, showing that researchers needed to understand learning not just as input resulting in output, but as a series of thought processes. “LRDC moved quickly to bring in thinkers in this field and then became a leader in applying it to education,” he says. However, “the cognitive revolution was incomplete. It stopped short of saying you want to understand the brain. The mind and the brain were not considered the same thing.”

But over the last 20 years cognitive neuroscience, or the study of brain processes mainly with imaging, drastically has shifted the fields of psychology and learning research. “LRDC started having cognitive neuroscience faculty and now we have a significant representation of research of that kind,” Perfetti says. For instance, he has used EEGs to measure brain surface activity during different reading activities. In years since, brain research at LRDC has gone far deeper than that — literally.

And LRDC researchers still are working on observing and improving instructional methods, which early LRDC scientists Lauren Resnick — herself the LRDC co-director and then director 1977-2008 — and Isabel Beck are pursuing today. While still comparing the efficacy of instructional methods and devising fresh ones, staff members also are studying educational reform efforts and broader issues of educational policy. These have been concentrated since 2007 in the Learning Policy Center, directed by Mary Kay Stein, LRDC’s associate director for education research and practice, jointly with the School of Education. This center studies and offers expert advice on learning policies that influence classroom practice and administrative procedures. A new Learning Science and Policy PhD program is also about to graduate its first crop of students.

Since 1995, the Institute for Learning, founded by Resnick, has supported school districts and innovations in teacher professional development, offering new research-based curricula and tools to aid teachers and administrators. What began as a series of quarterly seminars soon focused on training school-district leaders.

“These centers have potentials to have broader impact than developing learning programs,” Perfetti adds.

“You can trace all of these back over the 50 years, where Bob Glaser had the idea that people from different disciplines could come together to conduct research that could be applied to educational problems. Almost anything that is relevant to academics and school-based learning is part of LRDC. Faculty members are in an environment here where they can collaborate and reach out, getting involved with a colleague from very different disciplines.”

THE EARLY YEARS

“When LRDC started, it really wasn’t reputable to be studying real students doing real learning,” recalls Alan Lesgold, who was hired in 1971. “It wasn’t unusual that the only place LRDC scientists could report their results were in our working papers,” since there were no journals like Cognition and Instruction, founded by LRDC faculty, at the time. “The brain, to researchers in the ’60s, looked far too complicated to understand,” because brain imaging wasn’t sophisticated enough to see details of activity and connections. “Back then, when somebody came up with a brain theory, it wasn’t real science. This is a place where, until about the mid-’70s, there were still people using pigeons in basic research.”

Today, Lesgold is a senior scientist at LRDC — and dean of the School of Education — after having served as LRDC associate director 1986-2000. Lesgold credits LRDC’s longevity to its ability to transition between trends and leaders: “Generally speaking, a research institution tends not to survive its founders. One of the stellar things about LRDC is that it was able to have a strong next generation.

“When you’d get on the elevator with Bob (Glaser), he’d say ‘What have you discovered today?’ and he meant it,” Lesgold recalls. “Even today, we live in a world in which we struggle with the fact that people don’t believe in science. Bob had unbounded confidence that we could understand complicated stuff and that it would help us do a better job of making schools work. He was a little bit ahead of the curve in noticing that technology was going to allow us to think bigger than we used to think. The de-sterilization of research in thinking and learning owes a lot to the fact that this was a hospitable environment.”

Glaser arranged for Lesgold’s job interview to occur the day before a national conference LRDC was hosting, which marked the turning point between behavioral and cognitive psychology as LRDC’s major focus, Lesgold believes. However, he adds, “the focus all along has been a problem-solving focus: Given what we know about a student, what can we do to help that student learn better?”

Beyond Pitt, Carnegie Mellon University also was helping to pioneer cognitive psychology and artificial intelligence (AI) under faculty members Herb Simon and Allen Newell, with whom LRDC faculty collaborated. In addition, Lesgold points to the influence of Glaser’s friendship with former Pitt economics professor Alvin Roth, who won the Nobel Prize last year, as influencing LRDC’s focus on large problems rather than simpler and more easily proven ideas. Ralph Tyler, chair of Science Research Associates’ Educational Advisory Board around the time of LRDC’s founding, was another large influence, Lesgold says; Tyler helped the federal government create the laws that funneled research and development funds to centers of education, and then helped Glaser to seek such funds when they first became available.

Lesgold’s own work at LRDC has used AI to design instructional systems for complicated work tasks, such as fixing the machines that manufacture computer chips. The Air Force funded his first-generation work, which aimed to teach avionics — and paid an outside economist to figure out whether the work was viable economically.

It wasn’t, of course — a first-generation system never is, Lesgold says. But, he adds, “LRDC was the kind of environment where, when somebody says that’s not practical to do, we say, ‘Let’s think about it a little harder.’ By the time we built the fifth-generation system, the total investment was recovered the first time one person used it. A lot of work has gone on in LRDC which, if it went on in any other university department, would have your older colleagues telling you, ‘You know, you shouldn’t do that or you will never get promoted.’”

*

Isabel Beck was determined to enter law school in 1967 after working as an elementary school teacher. But having two kids of her own forced her to put off the move and take a research associate post at LRDC instead.

“I had become fascinated with reading and how it worked, as a teacher — the kids who don’t learn and the kids who do without the teacher even turning the page,” recalls Beck, now a senior scientist and emerita faculty member. LRDC helped show her how that fascination could translate into both research and in-classroom work. The idea of law school vanished.

“Sometimes it’s amusing when I look back at how we didn’t understand each other,” she says of the early faculty’s split between psychology lab work and the education faculty’s focus on practical applications. “It was ships passing, early on.”

But she adds, “It didn’t take long for respect and interest to develop in either community. I see it being a good example of how research and practice can use each other’s skills to make progress in understanding phenomena and hopefully using it for the good of the population.”

Beck was very involved in LRDC faculty research inside Falk School — the laboratory school for the School of Education — developing instructional programs to aid reading comprehension, embodied in such textbooks as “Questioning the Author” and “Bringing Words to Life.” The latter, which promotes using student-friendly explanations rather than definitions in teaching vocabulary, has sold more than 400,000 copies.

“We have made an enormous impact on the field,” Beck notes. “Everything we put into schools we develop from scratch with our little group.”

Lauren Resnick has been instrumental in leading that “little group” to prominence, beginning at LRDC as a research associate in 1966 and moving constantly between lab and practical settings for her research into instruction. When LRDC’s own building was finished in the mid-1970s, LRDC faculty decided to bring Falk students to a kind of demonstration classroom, in (and called) Room 314.

“The new building had this funny-looking outreach over the street, so that was a space that it wasn’t clear what to do with,” Resnick remembers. Faculty concluded: “‘Why don’t we open a school in there and we can do research right in our building?’”

Falk students ages 5-8, whose parents applied for entry, spent part of each school day in this room, which had one-way mirrored glass along a wall for observation by LRDC faculty, and microphones suspended above the room’s learning stations. Room 314 tested theories of individualized instruction. Students went to computers each morning to retrieve their unique assignments, then completed each day’s “contract” at their own pace before being rewarded with the chance to try the room’s additional learning activities, which were aimed at developing perceptual and other skills, or the option to read. Director Margaret Wang, then head of LRDC’s design and study of adaptive learning environments unit, reported that participating students spent 76-85 percent of their time on task, compared to 50 percent or less for other groups nationally.

LRDC work in classrooms throughout the United States continued, including National Science Foundation-funded collaboration with CMU and others via the Pittsburgh Science of Learning Center, now called LearnLab. LearnLab conducts “a kind of updated version” of the old Room 314 research, Resnick says, “to see how we can use computational power to aid individualized learning.” Today, Resnick is studying how to promote academically productive talk in classrooms.

FROM INSIDE THE CLASSROOM TO INSIDE THE BRAIN

James Pellegrino was at LRDC in the 1970s, working with Glaser and others “trying to understand what exactly is measured on tests of intelligence,” he says. “We were interested in those kinds of abilities because they tend to predict academic success. It was at the forefront of making those connections between cognitive psychology and instruction.”

Pellegrino, now co-director of the Learning Sciences Research Institute at the University of Illinois-Chicago, was among the scientists running the first conference on cognitive psychology in 1977 in the Netherlands, funded by NATO. He, Glaser and Lesgold then edited the resultant book, “Cognitive Psychology and Instruction.”

“We were helping people connect this and recognize this internationally,” he recalls.

Mary Kay Stein, now LRDC’s associate director, worked in Room 314 in the 1980s, joining LRDC first as a graduate student and continuing as a post-doctoral student and faculty member. “It was very fun training students in the building,” she says. “It brought the place alive.”

Today the Institute for Learning is in Room 314. “Now it’s filled with education practitioners who translate the results of research into tools and protocols and professional development for teachers and leaders out there in the schools.”

Stein, now a senior scientist whose focus remains on the impact of teaching and on how organizational and policy factors influence the improvement of teaching, ventures into classrooms still today, sans the mirrors and mics. “The laboratory environment, even if you have kids in it, not pigeons, is an artificial environment. In a school, you’re getting an authentic slice of what happens in the field. It will give us the kind of results that have a shorter distance to travel to be useful in the classroom.”

One highlight of LRDC’s 1990s work was QUASAR — Quantitative Understanding: Amplifying Student Achievement and Reasoning. It was created by a $10 million Ford Foundation grant to improve middle-school math education in economically disadvantaged cities, focusing on the teaching of reasoning and abstract problem-solving skills over basic computation. It was led by Ed Silver, then an LRDC senior scientist.

“At its height we took up maybe a floor and a half of the building,” recalls Stein, who was involved in the program. QUASAR found that the quality of learning depended on the extent to which high cognitive demand was maintained for students — “how teachers helped kids, without taking over the thinking for them,” Stein says.

*

Before senior scientist Julie Fiez joined the LRDC faculty in 1997, she was studying cognitive neuroscience at Washington University in St. Louis, combining neural operations in psychology with what we know about brain function to study brain regions that support language and reading.

In college, she conducted studies of how well people could produce appropriate verbs when presented with a noun (“gallop” for “horse,” for instance) over repeated trials. This was one of the first projects to use fMRI (functional magnetic resonance imaging) to identify the main brain regions that are activated during such an experiment.

Her study subjects gave faster and more accurate answers as they repeated their task, but that’s just the expected behavioral observation, she says. With fMRIs, she saw “a dramatic reduction in the amount of neural activity required to complete that task” over its length — and that the cerebellum, which was thought to govern mainly motor functions, played a surprisingly large role as compared to the cerebral cortex, which was the expected site of this task.

Did the older cerebellum evolve to add in more neural tissue and connections to support cognitive processing in ways that were poorly understood? The cerebellum had been known as the part of the brain that realigned visual-motor maps, learning to detect and correct errors between people’s intentions and how they actually moved. Now it was somehow helping to detect and correct errors in cognitive processing.

Fiez found a clue by using fMRI on a stroke patient with damage to part of the cerebellum that was very active in undamaged people during the verb-generation experiment. This patient had few motor problems from his stroke, but demonstrated clear deficits in cerebellum activity during the verb-generation task.

At LRDC, the cerebellum is at the forefront of several of her projects, looking at how this part of the brain plays a supervisory role in learning, becoming a kind of teacher itself, in such situations as adapting to hearing one’s native language spoken by a new acquaintance with a foreign accent.

“Being in LRDC has definitely shaped the course of my work,” Fiez says. “It’s added research projects that have a more applied focus, and the focus is to bridge cognitive neuroscience and education.”

Over the past several years, other faculty members have developed another imaging technique to look at patterns of functional connections. Walter Schneider, who joined LRDC in 1985, is at the forefront of high-definition fiber tracking (HDFT).

He began his career here studying the development of automaticity (the ability to do things without thinking about them) and other subjects, examining control processing (how people learn how to execute a function at first, then get so used to it that it feels automatic) behaviorally, computationally and in brain biology. He authored the first Nature paper on fMRIs, helping to start that area of research, then in 2009 began working on mapping anatomical connectivity in the brain using HDFT.

Schneider compares HDFT to X-rays, which finally allowed humans to see living bones as the principal structural component in a noninvasive manner, 1,800 years after human dissections had revealed them. “For about 200 years we’ve known that connections are the major structure of the brain, but up until two years ago we hadn’t developed sufficient technology to map that,” he says.

“I can take and dissect a living person’s connective structure” — without actually making a cut. “And I can create a circuit diagram based on that. I can look at normal or abnormal connection patterns.”

Connectivity disorders affect 10 million people a year. HDFT helps with neurosurgery, locating tumors and pinpointing removal methods that will cause the least damage to surrounding tissue. Schneider and colleagues are looking at brain wiring in people with autism and trauma, epilepsy and Alzheimer’s disease.

“We’re now in the situation where we’ve had 12 popular press reports in the last year,” including a third Discovery Channel special upcoming and recent coverage on “60 Minutes.”

“Too many!” he says.

TODAY

Today at LRDC, work continues to grow from all the seeds cultivated by Bob Glaser and the center’s first faculty 50 years ago.

LRDC senior scientist Kevin Ashley, who also is a faculty member in the School of Law, has used AI to develop a tutoring system that would teach law students argumentation.

With argumentation, “there isn’t one right answer,” just more-or less-reasonable arguments, so any computer system examining law students’ writing had many factors to consider.

“When I first came here, I tried to build a program where all of the intelligence comes from the program,” Ashley says. “Now, after 25 years of banging my head against the wall, I am building programs that provide the best kind of feedback they can provide, but they don’t understand the substance of the argument. So we’re relying on humans to provide feedback on the substance of the argument … and trying to put them together in a useful way, in a structured way.” He is working with LRDC colleagues Christian Schunn and Diane Litman on The Argument Peer.

“This Argument Peer work is going to be one technique for trying to save higher education, which I think is plummeting toward MOOCs” — Massive Open Online Courses that attract tens of thousands of students to some of the free offerings but present too large a class for any professor to interact with members individually. “It’s a technique where, by using the combination of the computer and human peer reviewing, you can create structures for giving students intelligent feedback on their arguments. The Argument Peer approach at least provides the possibility of structuring some intelligent feedback from the students.”

Litman, who joined LRDC in 2001, studies education technology, focusing on natural language processing or how computers can understand human typing and speech, as the iPhone’s Siri dialog system attempts to do.

Working with Ashley and Schunn, she is trying to apply natural language processing to student peer reviews of classmates’ writing assignments. When reviewers make comments that are too vague, she is hoping natural language processing can flag them and point the paper’s author to the source of the comment. It also may help students and professors summarize the gist of many comments, ignoring all the unimportant or redundant information.

For spoken language processing, Litman’s work is trying to teach computers to respond not only to what students say but how they say it — for instance, if they get the right answer but seem hesitant.

Christian Schunn developed his own peer tutoring program shortly after joining LRDC in 2001. SWoRD (Scaffolded Writing and Reviewing in the Discipline) facilitates web-based peer reviews and is used to aid in writing instruction where once no writing assignments were thought possible.

The program began as a practical solution for his own teaching needs in large psych lectures. His tests were multiple choice of necessity. “I felt guilty about that and I found a way to add writing back to the big lecture classes,” Schunn says.

Now SWoRD is used from here to Korea and Estonia in many different disciplines, from nursing, physics and English to “disciplines I’ve never even heard of, like leisure studies,” he says. The 500-plus students in one astronomy lecture at the University of Kansas now have Schunn to thank for the fact that they each have four writing assignments in one semester.

Schunn developed SWoRD by studying what students can learn from getting feedback from peers instead of instructors, and whether they can trust their fellow students’ reactions. As it turns out, averaging at least five peer reviews actually outdoes the instructor for helpfulness, he found, while “the most learning comes from giving feedback,” he says.

Using SWoRD, students also get a grade for how consistently they rate others’ work, and the authors rate the reviewers for how consistent they are as well. Instructors provide a rubric for these evaluations, which are assigned at random to prevent collusion or low ratings given as revenge for any low ratings received.

Now Schunn has National Science Foundation and Department of Education grants to add features, using “the decades of language processing studies LRDC has been a part of.”

He also is working with Mary Kay Stein on new national science standards called Next Generation. Next Generation will challenge high school science teachers, Schunn says, to add engineering practices to their curricula and integrate math and computational tools, such as probability, which can be applied to such science concepts as evolution. As part of the new teaching standards, for instance, high schoolers must devise a breeding plan for lightning bugs that will modify them genetically to turn a different color when they sense unexploded bombs in Laotian villages — still a problem today. And this breeding program must be affordable.

“We want the kids to think some big thoughts,” Schunn says. “We’re also asking teachers to do something really different. It’s not just teaching new things, it’s teaching things in a very different way.

“This is an example of where the work I have previously done … all comes together and really illustrates the value of the LRDC.”

Adds Schunn: “I hated biology in high school. It’s so funny that I’m now in charge of fixing it.”

Timothy Nokes-Malach is one of the newer LRDC research scientists, having started in 2007. He is a cognitive scientist focused on how people learn and transfer their learning to new situations, examining how students use analogies and self-explanations, whether student collaborations are helpful or harmful to learning, and how motivation can affect cognitive processes.

For the last project, Nokes-Malach has identified a variety of student achievement goals: mastery approach, when they want to understand content and master concepts; performance approach, when they want to show that they are the best or can solve a problem, and performance avoidance, when their goal is not to be worse than average or simply not to fail. His work has proven that motivation to achieve one of the approach goals is superior to the avoidance choices — which means teachers should motivate their young charges to pursue approach goals.

THE FUTURE?

What do the next 50 years hold for LRDC? “It is a time of planning for us, for the future,” says Charles Perfetti. “In the past, we have been fortunate, and probably strategically so, to move our field forward instead of being a generation behind. The trick is how to make sure that remains true.

“Twenty years ago we could not have predicted that we’d be studying the brain at the LRDC. The window for change gets shorter and shorter.” So the future, he believes, is simply “hard to predict.”

Says Lauren Resnick: “In the early ’70s, what was going on between people in the LRDC looked very coherent. Now there’s lots of communication but people are working at the edges of their specialties.”

However, she adds, “The mission is very much what it was. Psychology has changed a lot and LRDC has changed with it. I still think it’s academia’s dream job, because there’s no squabbling.”

Isabel Beck believes that brain studies are going to make a huge, practical impact on the field of education, but she worries about this emphasis taking away from the study of education in the field. Mary Kay Stein is similarly confident about the future, but also wants to make sure that LRDC’s researchers continue “trying to look at policy in education that doesn’t black-box the classroom” by examining only a policy and its outcomes without looking at how well policies are implemented on the ground, in classrooms.

The future is “a little unclear” to Alan Lesgold. He labels LRDC “bi-stable,” with strengths in both theoretical and applied science that could shift either way in the next few years.

“LRDC has been very good at allowing both sides to do their stuff without being held back by the other,” Lesgold notes. But each new era sees shifts in how much one side influences the other. “In the end, it’s the same story that was here in the late ’60s and early ’70s. The question about LRDC has always been the same question: Are those folks going to talk to each other and how is that going to shape LRDC’s future?”

—Marty Levine