• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Dissecting the Process of Reasoning

Posted on: Thursday, 30 December 2004, 03:00 CST

When Sir Isaac Newton published his theory of gravitation in Principia Mathematica in 1687, he prefaced the work with a disclaimer: "Therefore let the reader beware of thinking that by words of this kind I am anywhere defining a species or mode of action of a physical cause or reason." Newton insisted he was making mathematical observations about centers attracting and forces acting on objects, not explaining causal relations. Is it possible to propose or even conceive a concept such as gravitation without explaining the relationships between observable data?

"Much research on causal or scientific reasoning ignores the role of theory, or explanation," says Barbara Koslowski, associate professor of human development. "Furthermore, researchers often note that relying on theory has disadvantages. It can lead to a confirmation bias-a tendency to avoid searching for information that can undermine one's theory and to avoid taking account of such information when one is presented with it."

Many researchers argue that scientific reasoning constitutes a method that must be learned if one is to avoid confirmation bias. Koslowski disagrees. "Scientific reasoning is just a special case of good reasoning in general," she says. "It's not that physicists, for example, reason differently from mere mortals; it's that they know a lot about physics that most mortals don't know. The basic principles for reasoning remain the same."

According to Koslowski, those principles include theorizing, or generating explanations, as well as generating alternative explanations, for phenomena. She explains that it is theories, or explanations, that determine other aspects of good scientific reasoning, such as looking for data that distinguish between competing explanations rather than data that are compatible with both, and deciding when to reject rather than modify an explanation in the face of disconfirming data. In addition, distinguishing evidence and disconfirming data help us evaluate our theories and decide whether they should be rejected or modified. She has found that, like scientists, most people apply these principles fairly routinely.

Koslowski began studying the process of reasoning in sixth- and ninth-grade students; she found significant developmental differences between those age groups. She also compared sixth and ninth graders with college students. She found that young adolescents will treat as reasonable information that college students find implausible. Because the younger age groups had less background information than the college students did, they were less able to judge that the theories being presented were implausible. "If you don't know how deep holes in the earth usually are, it's difficult to realize that you should be impressed by the Grand Canyon," she explains. "For example, we had one sixth grader who explained a correlation between color of car and gas mileage by suggesting, with great enthusiasm and conviction, that blue cars probably attract bugs, and the bugs hit the front window and that causes wind resistance, which makes blue cars go slow. The college students who are asked to evaluate such explanations laugh, albeit indulgently."

Since the intent of her research was not to test subjects' knowledge, it made sense for Koslowski to choose an age group that was developmentally mature enough to evaluate data. She began working with college students in a laboratory situation where she can control the information she gives participants. She takes her content from actual, debated topics. For example, she may introduce a controversy in archaeology about whether an early group of nomadic women were a kind of Amazon group, fighting alongside men on horseback and killing, as well as getting killed, in battle. Or why people living in the mountains are typically shorter in stature than people who live at sea level.

One aspect of the way in which theory-including alternative theory-affects how science is conducted is that it determines whether scientists treat information as evidence that is relevant to explaining a phenomenon. One of the things Koslowski has been finding is that information becomes evidence when a theory is available that can fit it into a broader explanatory framework.

She gives an example. One of the phenomena that some scientists are trying to explain, based on fossil evidence, is why there were mass extinctions of large mammals roughly 15,000 years ago, when humans migrated to North America. One possibility is that the migrating humans hunted the mammals to extinction. However, people who work in this field also know, for example, that this was the first time that humans had lived in that part of North America, that their hunting tools were primitive at best, that they brought with them various birds and dogs when they migrated, and that the animal fossils were from animals of all ages.

If we now consider an alternative explanation, some of this information might be seen as evidence relevant to explaining the extinction. Specifically, if we consider a possible explanation- that the migrating humans brought with them diseases to which the animals were not immune-we can see that it's relevant that there had been no humans in the area earlier, and that it's relevant that the humans brought birds and dogs with them, because these small animals might have been the vectors by which the diseases were transmitted to the large animals. We also can see that information about animals of all ages is relevant, because hunters usually target the very young and the very old, but a new disease can spread through an entire herd. We might not realize how these pieces of information are evidential unless we view them through the lens of a particular explanation-in this case, an alternative one.

Since the ability to generate alternative explanations or hypotheses is one of the hallmarks of scientific reasoning, Koslowski wanted to observe under which conditions her college-age subjects were likely or not to propose alternative explanations. She wondered whether they would exhibit better reasoning skills if they were taught a set of formal rules. To test this idea Koslowski used four groups. One group of students was advised to "always consider alternative hypotheses" and then was given a story problem. A second group was taught only facts about the story in question. A third group was given the formal rule as well as information about the story. And a fourth group was given causally irrelevant data.

What she found is that people who were trained to use only the formal rule and were not given much content did not do very well in proposing alternative hypotheses. This makes sense to Koslowski.

"If you know a lot of data and someone gives you an explanation, you're likely to propose an alternative hypothesis." she explains, "Whereas you are not likely, when presented with data, to say to yourself mechanically and formally, 'Hmmm, I must consider alternative hypotheses, even though I can't seem to think of any right now.' Content, it seems, triggers the formal rule, rather than vice versa."

Koslowski also has found that people are more or less likely to search for different kinds of information depending on the type of information they are given at the outset. "If you give people a single explanation, they will ask for data that could prove the explanation is plausible, but they also will ask questions about alternative explanations that could be true," she says. "In contrast, if you give people a target explanation plus an alternative explanation, it shuts down the alternative-generating capacity. They will focus only on the two explanations they were given. We're finding that the information people look for on their own might be heavily determined by the information that they are initially presented."

Acceptance of the fact that theories are embedded in and judged with respect to a broad network of related information-including information about theories-might help explain why it is that some beliefs are so difficult for people to reject. Koslowski suggests that to undermine a belief, you need to undermine the network of beliefs in which the target belief is embedded.

Her graduate students, Amy Masnick and Melanie Swiderek, tested this notion by conducting extensive interviews with college students about their beliefs regarding capital punishment and affirmative action, two topics typically considered to be examples of emotion- laden cognition and thus resistant to change.

Students were asked not only about their beliefs about capital punishment but also about the network of related beliefs-for example, beliefs about capital punishment as a deterrent, whether it is cost effective, whether it is applied in a racially unbiased way- that supported their general beliefs. No matter whether they were pro or con on the issue of capital punishment (and thus, respectively, pro or con on the beliefs that it was a deterrent, cost effective, and applied in a racially unbiased way), they then were given information that undermined their beliefs. For example, participants who supported capital punishment and believed that it was a deterrent were shown statistics citing that, although there is an initial decrease in the rate of capital crimes after the adoption of capital punishment, after three months the rate reverts to its level before capital punishment. In con\trst, interview participants who opposed capital punishment and believed it was not a deterrent were presented only with data from the first three months after capital punishment is adopted-namely, that adoption of capital punishment is indeed followed by a decrease in the rate of capital crimes.

Each supporting belief was methodically disconfirmed. The researchers used the same approach to undermine deterrent beliefs about affirmative action.

Koslowski found that the single best predictor of whether participants were likely to change their positions was the extent to which they had changed their supporting beliefs. The way to change someone's viewpoint is not to focus on the single overarching opinion, she explains, but rather to focus on changing, or disconfirming, the network of related beliefs within which the opinion is embedded.

"In terms of attitudes like sexism, racism, anti-Semitism, and homophobia, if you believe that people adopt a confirmationbiased attitude that never goes away, that is a dismal outlook," she says. "But if you realize that by chipping away at the network of related information you actually might induce people to change their minds, that is encouraging."

Although Koslowski's work on disconfirming beliefs has the most obvious practical applications, her work on the role of theory in reasoning also is relevant to how we teach people to conduct science and how we teach people to evaluate scientific research.

For more information, contact

Barbara Koslowski

Cornell University

Department of Human Development

Martha Van Rensselaer Hall

Ithaca, New York 14853-4401

607-255-0837

bmk2@cornell.edu

"...Individual nonscientists pay attention to the type of problematic information they encounter and to the network of related information."

"If you know a lot of data and someone gives you an explanation, you're likely to propose an alternative hypothesis."

"In actuality it is very difficult to do any kind of reasoning without some background beliefs or theories."

Copyright Cornell University Oct 2004

Source: Human Ecology; Ithaca

More News in this Category