Researchers Go From A to B to Discovery
Washington Post Staff Writer
Monday, January 26, 1998; Page A3
The mythology of science is tall with tales of lone scientists whose dramatic discoveries came from flashes of insight or strokes of good luck. Friedrich August Kekule, the 19th century German chemist, claimed to have deduced the circular structure of the chemical benzene after dreaming of a snake biting its own tail. Alexander Fleming discovered penicillin when he had the good fortune to sneeze into a petri dish containing a natural culture of the germ-killing mold.
But are these stories representative of how advances are typically made?
Not according to a new field of study that focuses on the psychology of scientific discovery. Far from the movie image of the white-coated iconoclast shouting "Eureka!" in the middle of the night, most scientific breakthroughs are collaborative, gregarious affairs involving many participants with varying backgrounds.
Moreover, discoveries rely very little on blind luck or grand strokes of genius and much more on solid logic, a talent for apt comparison and a mind so steeped in a discipline that it can recognize an unexpected clue for what it's worth.
This more nuanced, if less cinematic, view of science has emerged in part from a remarkable series of studies led by a psychologist who spent two full years directly observing scientists at work in their laboratories. That effort, by Kevin Dunbar of McGill University in Montreal, suggests that scientists themselves often are surprisingly deluded about how they and their colleagues work. Many scientists are even mistaken about how they arrived at their own discoveries.
"There are all these myths and stories," Dunbar said. "But no one has really watched [scientists] to see what they really do. No one has asked, 'What are the mental processes that scientists use when they think and reason every day, and how do they make discoveries?' "
Cognitive psychologists specialize in understanding the thought processes that get a person from A to B. They look at problem-solving strategies, the ways that people compare new information against old and how people deal with unexpected or seemingly contradictory observations.
Ultimately, cognitive studies could help scientists become more efficient in their work and perhaps change the way science is taught in schools. Already, such studies have begun to debunk supposed differences in the ways that men and women do science.
Early efforts to understand the psychology of discovery consisted mostly of studies of scientists' notebooks, the best of which contain enough detail to allow insights into the authors' thought processes.
Bowling Green State University researcher Ryan Tweney, for example, used the elaborately documented notebooks of 19th century physicist Michael Faraday to study "confirmation bias," the potentially misleading tendency to play down findings that contradict one's hypothesis and to read too much into supportive findings. To his surprise, Tweney found that Faraday and other successful scientists benefited from some degree of bias. At least in the early stages of research, he found, it pays to stick to a hypothesis even as evidence to the contrary accumulates. Those who change course too easily do as badly as those who cling too long to their wrong views.
Other psychologists interested in the dynamics of discovery have developed computer programs that mimic the process, and still others study people given make-believe problems to solve. One common finding is that discoveries often have roots in unexpected findings.
"If you look at the Nobel laureates, in case after case after case the critical event was a surprise," said Herb Simon, a professor of computer science and psychology at Carnegie Mellon. Of course, a surprise is not enough by itself, Simon added. "You need to recognize a surprise for what it is. These people are very knowledgeable in their fields. They know a surprise when they see one."
Dunbar, a researcher at McGill's cognitive neuroscience center, took a different research approach. For two years, he directly observed working scientists in eight biological research laboratories at two major universities in North America. He tracked the planning and execution of experiments, the interpretation of results and the preparation of articles for publication in journals. He also attended private lab meetings, where scientists talked candidly about their preliminary findings and figured out what to do next.
First, Dunbar found that scientists' common refrain, "I was lucky," is at best false modesty and at worst a failure to recognize how logically they came to their own discoveries.
"This is a myth, the myth of chance in science," Dunbar said. "You ask a scientist, 'How's your lab structured?' and he says, 'Oh, it's anarchy, it's chaos, anything goes.' But you go in there for a couple of months and you see it is a beautifully designed lab. He may say, 'Oh, that just happened.' But the truth is, they've got things so narrowed down in the first place, and there is so much basic knowledge they have, that in the end they may see their final decision to perform a certain experiment as luck or chance, but that's hardly ever true."
That doesn't mean the process of discovery is predictable. From his own observations of hundreds of experiments, Dunbar found that 50 percent to 60 percent of all results are significantly different than expected that is, they do not support the hypothesis that the scientists had when they designed the experiment. But even unexpected results can be informative; Dunbar found that 50 percent to 70 percent of scientists' conclusions about how things work were born of unexpected results.
Dunbar's research also shows that to make sense of such findings, scientists rely largely on analogy the process of applying knowledge in one area to solve problems in another. But he found that the most successful scientists use analogies differently than many had presumed.
The most famous examples of analogies in science are the "distant" analogies that span disparate fields of study. For example, Ernest Rutherford's knowledge of how planets revolve around the sun is said to have led him to propose that electrons orbit the nucleus of an atom.
But Dunbar found that distant analogies like Rutherford's account for only 2 percent of all analogies used by scientists and are relatively uninformative. More common and more useful are "close" analogies, such as when a scientist uses knowledge about the workings of a gene in one virus to gain insight into how a related gene in another virus works.
Surprisingly, when Dunbar asked researchers how they had come to their conclusions, most had quickly forgotten their reliance on analogies, even though Dunbar had documented their use. That helps explain why so many researchers later credit good luck or a "flash of insight."
Dunbar's work also points to the advantages of "distributed reasoning," in which several scientists put their heads together to solve a problem. In contrast to the classic image of the reclusive scientist suddenly "getting it," the most important discoveries Dunbar witnessed arose when several participants built on each others' analogies and interpretations.
This advantage, however, appeared only when members of the group had varying areas of expertise; when members of the lab all had similar backgrounds, progress was no faster than by individual scientists working alone.
Finally, Dunbar's work sheds new light on the long-standing question of whether men and women do science differently. According to surveys of scientists, the prevailing assumption among both male and female researchers is that male scientists are more competitive, and women avoid the more cutthroat areas of research.
Dunbar's observations suggest, however, that female scientists are as likely to challenge a finding or a colleague as are men. "They are not meek people," Dunbar said. "And if you look at the ways they think and reason, and how they make deductions, we found no significant differences whatsoever except in one small place." That difference is small but statistically reliable, Dunbar said: When confronted with a clue that they are off track, men are more likely to ignore the warning while women are more likely to form a new hypothesis or try a new approach.
Which way is best? Dunbar suspects the male method may have advantages early in a project, when it pays to be headstrong, while the female approach may prevail later on, when the bullish advantages of sticking to your guns are outweighed by the need for refinements.
But there's no better way to explore the question than by direct observation, Dunbar said. So he will spend several months in Italy next year, studying labs run entirely by women.
© Copyright 1998 The Washington Post Company