In 2002, New York University political scientist Russell Hardin wrote a brilliant essay called "The Crippled Epistemology of Extremism." Hardin contended that many extremists, including terrorists, are not stupid, insane or badly educated.
The real problem is that their information comes from a sharply limited set of sources, all of which are supportive of their extremist beliefs. Many extremists listen only to one another. They live in self-reinforcing information cocoons. Their "crippled epistemology" can lead to utterly baseless, but firmly held, convictions (and sometimes even violence).
Most Democrats and most Republicans are not extremists. But Hardin's argument offers lessons about 21st-century political campaigns in the United States — and about some of the most serious difficulties in contemporary governance.
How do you know what you know? You undoubtedly have firsthand knowledge about many things. But how do you know whether George Washington or James Madison really lived, whether matter consists of atoms, whether Bob Dylan wrote "Like a Rolling Stone," or whether Mars and Venus exist?
Most of what we know is what we learn from other people. By itself, that is inevitable and nothing to lament.
But here is the problem. When we listen mostly to people who already agree with us, our pre-existing convictions get fortified, and we start to think that those who disagree with us are evil, dumb or duped. Is it any wonder that our politics are highly polarized, so much so that it sometimes seems as if Democrats and Republicans don't merely disagree but live in unfathomably different universes?
A few years ago, I participated in some experiments designed to shed light on how people's political beliefs are formed. We assembled a number of people into all-liberal groups and all-conservative groups. We asked the groups to discuss three issues: climate change, affirmative action and civil unions for same-sex couples.
We requested group members to state their opinions at three stages. The first occurred before they started to talk, when we recorded their views privately and anonymously. In the second stage, we asked them to discuss the issues with one another and then to reach a kind of group "verdict." In the final stage, we asked people to record their views, after discussion, privately and anonymously.
On all three issues, both liberal and conservative groups became more unified and more extreme after talking to one another, both in their public verdicts and private statements. Discussions with one another made conservatives more skeptical of climate change and more hostile to affirmative action and same-sex unions — while liberals showed exactly the opposite pattern. After liberals spoke only with liberals, and conservatives only with conservatives, the divisions between the two groups grew dramatically.
Why do groups polarize in this way? One reason involves people's concern for their reputations. If you find yourself in a group of people who hate affirmative action, you might be reluctant to say that you like affirmative action, and your agreement with the group in a public setting might affect what you say privately.
The more interesting reason involves the exchange of information. In conservative groups, for example, people tend to offer a number of arguments against affirmative action, and very few in favor of it. Group members learn from what they hear. Having heard the set of arguments in their group, people become more confident, more unified and more extreme.
Can anything be done to address this problem? The most obvious answer is to break out of information cocoons. That is a central goal of the American constitutional system, which was devised to ensure that diverse people would speak with one another.
The Anti-Federalists, opponents of the Constitution, urged that self-government required homogeneity and that diversity could create paralysis and chaos. By contrast, the defenders of the Constitution, above all Alexander Hamilton, thought that diversity could be a creative force and that "the jarring of parties" could be productive, because it would "promote deliberation."
Political conventions are occasions for group polarization. This is inevitable and by design. But in the best cases, political campaigns get people to escape from their information cocoons — not merely because competing perspectives are available, but because citizens are really listening.
When escape proves difficult, it helps to insist on the importance of respecting technical expertise. In politics and government, a healthy respect for the technical expertise of scientists, lawyers and economists usually helps to anchor discussion — and to avoid a crippled epistemology.
Many of our political convictions are intensely held, especially in an election season.
Some of us are undoubtedly right. But an appreciation of how we know what we know should help to engender a healthy dose of humility, making political campaigns far more productive and sensible governance far more likely.
Cass R. Sunstein, the Felix Frankfurter professor of law at Harvard University, is a Bloomberg View columnist, former administrator of the White House Office of Information and Regulatory Affairs and author of "On Rumors: How Falsehoods Spread."