Corrected: Clarification: A photograph of a Lego robot that appeared on Page 35 in the print edition of this story should have noted that the photo was published courtesy of professor Nira Granott of the University of Texas at Dallas, and was used in research she conducted at the Massachusetts Institute of Technology Media Lab.
Forget about big ideas and large-scale studies. When it comes to studying children’s learning and development, a limited but growing group of researchers says thinking small is the way to go.
By “small,” these researchers are not just talking about the size of the humans under their microscopes. What they have in mind is studying learning in much shorter time increments than traditionally has been the case in the field.
New studies suggest that learning occurs in a process of fits and starts, steps forward and stumbles backward, rather than the steady upward progression that many people imagine.
The idea behind the new school of thought, which is called “microdevelopment,” is to give scientists a better picture of how learning occurs in real time, rather than relying on snapshots of what children learn at particular points. And, if developmental and cognitive scientists can puzzle out how children learn, they might be better able to figure out what teachers can do to nudge the process along.
Researchers predict that the new lens they are putting on the study of learning has the potential to revolutionize it in much the same way that newer brain-imaging technology has transformed neuroscience. Already, studies suggest that learning occurs in a process of fits and starts, steps forward and stumbles backward, rather than the steady upward progression that many people imagine.
“There is a view that this is a real paradigm shift in developmental learning,” says Nira Granott, a co-editor of Microdevelopment: Transition Processes in Development and Learning, a book on the subject scheduled to be published later this year by Cambridge University Press. “I really think it’s the way the future is going to go.”
How does the new thinking differs from traditional practice? Psychologists typically assess children’s learning by giving a pretest to a group of children, putting an educational intervention in place, and then returning six weeks or six months later to administer a post-test. But a microdevelopmental scientist might be more interested in videotaping children as they deal with a specific task and tracking their comments, gestures, and answers as they work.
The traditional time scale in developmental science is even greater. Watching children grow, for example, Jean Piaget, the famous Swiss psychologist, identified four distinct stages that characterize children’s learning abilities from birth to about age 15. The microdevelopmentally oriented researcher, in contrast, wants to know what happens between those developmental stages.
Microdevelopment—sometimes also referred to as microgenesis—is not completely new to the study of learning. Even some of Piaget’s collaborators called for taking a closer look at the processes and patterns of learning.
Researchers’ interest in the approach picked up in the early 1990s, however, as videotape equipment became cheaper and easier to use, and more sophisticated computer-software programs came on the market. The technology made it easier for researchers to take on the mammoth task of recording hundreds of hours of videotape, coding the responses or words and gestures of the learners they were studying, and systematically analyzing all of the information.
‘There is a view that this is a real paradigm shift in developmental learning. I really think it's the way the future is going to go.’
“You can do sophisticated analyses now without having to become a mathematician or a computer scientist,” says Kurt W. Fischer, a professor of human development and psychology at Harvard University’s graduate school of education here.
Fischer, steeped in the work of scientists such as the psychologist B.F. Skinner, the linguist Noam Chomsky, and Piaget, was first attracted to more process-oriented, shorter-time-frame approaches as a Harvard graduate student in the early 1970s.In the 1990s, Fischer also noticed that microgenetic approaches were drawing more and more prominent scientists from other fields— cognitive scientists, for example, such as Robert S. Siegler and Robbie Case. And his own graduate students were showing more and more interest in such research.
“I think it’s well on its way to becoming mainstream,” he says. “It’s at least respectable now.”
A Learning Roller Coaster
If microgenesis is in fact not yet completely mainstream, some of the reluctance to embrace it may come from critics who complain that microdevelopmentally-oriented researchers use too few subjects in their studies. Proponents contend, though, that the critics may have misunderstood their approach.
Even though they tend to look at fewer subjects, researchers taking this newer, shorter view of learning note that they take many more measurements—perhaps just as many as traditional researchers doing larger-scale studies. Moreover, they say, much of the work is still at the basic stage, when figuring out what is going on is more important than proving “what works” through large-scale studies.
Among the more consistent findings coming out of this new body of work is the idea that learning patterns look nothing like a steady, upward stair climb. Nearly a dozen studies suggest instead that, when examined at the micro level, learning looks a little more like a roller coaster ride. Learners of all ages grope for a solution to a problem, get a glimmer of an idea for a new approach, test it out, revert to an older and less efficient strategy, discard it, and then gradually incorporate the newer, more efficient strategy into their bag of skills.
“If you took a pencil to draw it, you would have to go up and down, up and down,” says Granott, who is also an assistant professor of psychology at the University of Texas at Dallas.
She says the patterns so far seem surprisingly consistent, showing up in studies of school-age children and adults, average learners and Ivy League graduate students, and even among children labeled as developmentally delayed.
‘It looks like these are regressions, but they're happening to support progress. Teachers need to know that this is important, because we are all used to thinking of regressions as something bad.’
“It looks like these are regressions, but they’re happening to support progress,” Granott says. “Teachers need to know that this is important, because we are all used to thinking of regressions as something bad.”
Three decades ago, an education researcher turning in data showing those kinds of up-and-down learning patterns probably would have faced criticism from more seasoned colleagues, says Fischer.
“They probably would have said, `You didn’t collect the data right,’” he says. “What we see now is that kind of fluctuation up and down is what learning behavior is all about.”
For her own studies of learning processes, Granott gave eight graduate students Lego robots, called “wuggles,” that reacted to light, sound, and touch by changing their movement patterns. The students were told nothing about how their wuggles worked. Their task was to figure that out. Along the way, they could experiment with the robots any way they chose. When the wuggles broke down, technicians were on hand to put them in working order. Granott videotaped the pairs’ interactions and analyzed the responses.
“I wanted to take bright people and put them in this really interesting, psychedelic environment, where there were lots of sights and sounds and robots moving, and not tell them anything,” says Granott, who at the time was conducting her study at the Massachusetts Institute of Technology Media Lab. “A place where they could really learn.”
Besides finding the same up-and-down learning patterns that characterize so much of the microgenetic research, Granott noticed something else. Typically, she found, students would blurt out vague comments indicating they were close to discovering a cause and effect—even before they had a real grasp on a solution.
The example she writes about is Kevin and Marvin, one of the pairs under study. When they first encountered the robot, the two spent some time playing with it, placing their hands around it in different ways, and observing its movements. Finally, as Marvin put his hand around the robot, Kevin said, “Looks like we got a reaction there.”
“By putting that word ‘reaction’ on the table, it shows they’ve already started to explore the reaction,” Granott says.
She, along with Fischer, and their colleague Jim Parziale, the other editor of Microdevelopment, have come to call these flashes of insight or brief performance peaks “bridges.”
The researchers liken them to the grappling hooks that rock climbers use to pull them upward, or to empty containers that learners put out and then fill in later with more solid information. They differ from hypotheses because they are something less than an educated guess.
Fischer has even documented the same sort of cognitive occurrences in the writings of Charles Darwin, the great evolution theorist. In a study with Zheng Yan, Fischer examined notebooks the naturalist religiously kept over a period of years as a young man.
“He came up with the principle of natural selection several times and, initially, he didn’t recognize that he had it. He didn’t know it was there,” Fischer says. “You can see things happening in days sometimes.”
Typically, students would blurt out vague comments indicating they were close to discovering a cause and effect—even before they had a real grasp on a solution.
Other researchers, using other labels and slightly different theoretical underpinnings, have noticed the phenomenon as well. At the University of Chicago, for example, Susan Goldin-Meadow and colleagues have observed that people’s words and gestures frequently start to become mismatched before they arrive at a new understanding. Looking at two glasses of water, for example, a person might say the one on the right is full while pointing to the one on the left.
Using the Research
The practical goal in all of this research, of course, is to find ways to help students better build bridges and create lasting understanding. One way to do that in teaching mathematics, suggests Siegler, who uses microdevelopmental approaches in his cognitive research at Carnegie-Mellon University in Pittsburgh, might be to ask students to explain why some strategies for answers work and others don’t.
Siegler recently tested that approach with 87 3rd and 4th graders. First, he asked them to solve a series of simple arithmetic problems that required multiple problem-solving strategies. No one got all 10 problems right.
After completing that test, one group of students was given a similar problem and told whether their answers were correct. A second group, facing the same problem, was also given feedback on their responses. In addition, however, the researchers told those students that a child at another school had given the right answer—whatever it might have been—and then asked them to explain why they thought the child had done so.
In a third group, the 4th graders were asked to explain the reasoning for both the hypothetical child with the right answer and for another hypothetical child who had responded incorrectly.
Tested again after those discussions, the children in the third group outperformed their peers in the other two experimental groups. They increased the number of correct answers from zero before the experimental sessions took place to 70 percent afterward. The children who were asked to explain only right answers correctly solved half the problems at post-test.
“I think undercutting less advanced or older strategies is just as crucial to helping children learn as inculcating or teaching new strategies,” says Siegler. “Now we assume that teaching them new strategies will work. We very often teach them new strategies but then, when they are solving problems later, they go back to using the same, old flawed strategies.
“Children are conservative learners,” he continues. “As long as what they are doing is working reasonably well, they will continue to use it quite a bit as they get more and more evidence that says a new strategy is superior.”
Here at Harvard, Marc S. Schwartz, the assistant director of the “Mind, Brain & Education” program that Fischer runs, is building a middle school science curriculum based in part on “learning-skills theory.’'
‘Sometimes in science, one of the most important advances is having new ways of measuring things.’
Developed by Fischer in the early 1980s, learning-skills theory is a microdevelopmentally oriented take on Piaget’s theory of human development—in part, because it looks at what happens between developmental stages, and in part because it provides a frame for learning that happens on a very small scale. Fischer came to his ideas through hours of videotaped studies looking at children’s behavior and responses to new learning tasks.
One of the tenets of the learning model is that, faced with a new problem, learners move through their entire developmental history--sometimes in a matter of minutes or hours— as they attempt to solve it. In other words, students would first manipulate physical objects, then come to understand representations of new ideas, and then move to more abstract models of concepts. If the terms sound familiar to educators, it’s because they’re partly drawn from Piaget’s developmental stages.
In addition, the theory holds that people can possess a skill, yet be unable to demonstrate it without the proper supports—perhaps a good teacher or a carefully planned learning experience—to help them draw it out.
Keeping all that in mind, Schwartz has put together science lessons that require students to move through all the stages. Although the approach looks like “discovery learning,” another teaching approach that involves hands-on learning, the lessons here are much more carefully plotted out.
On a chilly afternoon in March, Schwartz was testing out one of those lessons with his graduate education students. The subject of the unit, which would span several weeks, was conservation of matter.
A week earlier, Schwartz told these students a story about a girl in ancient Egypt who noticed that, when she spilled vinegar on the ground, the rock bubbled. To test the idea for themselves, the students were given pieces of trona, the rock from which baking soda is milled, and vinegar. Some students noted, for example, that the mixture produced a gas that caused a lighted candle to go out, or that it produced temperature changes.
This week, the group is combining varying quantities of baking soda and vinegar to see how high the resulting gas can push a pingpong ball up a 90-centimeter glass tube.
The students will later graph the results, decide what proportion of vinegar and baking soda produces the most gas, and eventually toy with—and try to perfect—a physical model to describe the phenomena they witnessed. The prototype model students are using—made of pieces of cardboard and paper clips— is intended to give the students a figurative ladder so that they can move up to a higher level of understanding.
Dan M. Record, one of the students in Schwartz’s class, says the approach is a far cry from his own learning experiences in the subject.
“I don’t think I totally understand everything in physics, but I have a degree in it,” says Record, who is planning to become a high school science teacher. He adds, however, that the teaching approach he is learning also poses a dilemma for classroom teachers, expecially those at the high school level, who might feel compelled to feed students facts that they can regurgitate later on tests.
‘By teaching students more for understanding, you put them at an advantage. But, if you don't focus on the facts that they're going to be tested on later on, you put them at a disadvantage. That's what I'm struggling with right now.’
“By teaching students more for understanding, you put them at an advantage,” Record says. “But, if you don’t focus on the facts that they’re going to be tested on later on, you put them at a disadvantage. That’s what I’m struggling with right now.”
Typically, what happens in classrooms, says Schwartz, who is also a lecturer on education at Harvard, is that teachers start with abstractions, such as E=MC², and then fill in the blanks. “That’s not built on what students understand about their world. It’s built on what professionals or experts understand about the world.”
In 1999, Schwartz tested a similarly designed unit on electromagnetism with six 7th and 8th grade classes in Rhode Island and Massachusetts. The classes were equally divided into three groups. Some were taught using the “learning-skills theory” approach drawn from Fischer’s microdevelopmental studies. Others were taught with traditional methods, and the remainder experienced a “discovery” teaching approach.
Of all three groups, the students in the classes using teaching methods developed around microgenetics made the greatest learning gains. A former high school science teacher, Schwartz turned to Fischer’s theories after discovering that, even though he had been perfecting his teaching skills, his students weren’t learning any more science.
“I started keeping tests and putting them in a file drawer,” he says. “When I discovered my students didn’t get any better, that was very, very disturbing to me. I decided I had to change my perspective from what is going on in me to what is going on in them.”
Another practical application of microdevelopment, Fischer believes, may be in student assessment. With the advent of more sophisticated computer technology, he says, teachers should be able to keep closer, more accurate tabs on individual students’ learning. Traditional standardized tests, in comparison, offer only a snapshot of students’ capabilities. Yet-to-be-developed software programs, for example, might be able to show teachers daily which students are “getting it” and which ones need more targeted help.
“Teachers assess all the time. They need to assess,” Fischer says. “If we could provide a system that would provide that kind of immediate feedback, I would think that would be very helpful to teachers.”
Mathematical-modeling techniques may even enable researchers to predict how learning occurs in a classroom in the same way that meteorologists use mathematical modeling to predict how weather systems will behave.
“Sometimes in science,” says Fischer, “one of the most important advances is having new ways of measuring things.”
The Research section is underwritten by a grant from the Spencer Foundation.
A version of this article appeared in the April 24, 2002 edition of Education Week as Research: Under the Microscope