Hugo Mercier sent me this response (below) to my blogpost The invisible hand of argumentative reasoning doesn't work so well - so what can we do about it? Thanks to Hugo for this response!
Argumentation gets a bad press. It’s often portrayed as futile: people are so ridden with cognitive biases—less technically, they are pigheaded—that they barely ever change their mind, even in the face of strong arguments. In her last post, Helen points to some successes of argumentation in laboratory experiments with logical tasks, but she doubts whether these successes would extend to other domains such as politics or morality.
I think this view of argumentation is unduly pessimistic: argumentation works much better than people generally give it credit for. Moreover, even when argumentation fails to meet some standards, the problem might lay more with the standards than with argumentation. Here are some arguments in support of a view that is both more realistic in its aspirations and more optimistic in its depiction of argumentation—we’ll see if these arguments can change Helen’s mind about the power of arguments.
As Helen pointed out, the argumentative theory of reasoning that Dan Sperber and I developed explains why reasoning is biased—in particular, why it tends to look for reasons that support the reasoner’s point of view (myside bias or confirmation bias). According to this argumentative theory, when reasoning produces arguments, its function is to convince others, not to check the validity of the reasoner’s position, and so finding arguments that support the reasoner’s position makes sense.
By contrast, when reasoning evaluates arguments, its main function is to let the reasoner be convinced by good enough arguments. Argument evaluation should be as objective as possible, so that people can accept strong enough arguments even if they challenge the reasoner’s prior beliefs, or if they come from untrustworthy sources.
Helen cites some research purportedly showing that argument evaluation is in fact heavily biased. In these experiments, participants are given an argument—in the interesting condition, an argument that challenges their beliefs—and they are then asked to evaluate the argument. What typically happens is that people read the argument, but they do not find it strong enough to simply accept its conclusion. As a result, they start looking for arguments to support their rejection of the conclusion, as one would in a dialogic setting. Many of these arguments will point to flaws in the argument to be evaluated, making for a particularly critical evaluation.
If this interpretation is correct, then the only thing these experiments reveal is that the production of argument is biased (as expected), not that the evaluation per se—the brief phase that takes place when people read the argument—is biased. As it happens, I’m working on a paper that defends this interpretation with new experiments revealing that when a more appropriate methodology is used, argument evaluation does not seem to be biased.
Even if the reader was to trust me regarding the interpretation of these experiments, the objectivity of argument evaluation might seem hard to reconcile with the perceived failures of real life argumentation. If argument evaluation is so objective, why is it supposedly so hard to change people’s minds on so many issues?
If it is to be adaptive, argument evaluation should lead people to accept beliefs and decisions that are, on the whole, adaptive. This means that argumentation does its job when it rejects an argument that would lead to potentially less adaptive beliefs or decisions. This is true even if other people think the argument is strong, and if the reasoner cannot adequately defend her rejection of the argument. I would guess that this explains many so called failure of argumentation in politics, religion, and morality.
Take politics for instance. Weeden and Kurzban have recently argued that contrary to some political science received wisdom, people’s opinions often track their interests fairly well—as one gets rich, one tends to want lower taxes for instance. If they are right, then political argumentation should not be expected to make a massive difference. If it did, it would lead people to vote against their interests.
Weeden and Kurzban’s analysis also applies to some moral opinions. They suggest, for instance, that opinions about promiscuous sex are partly self-interested and depend on one’s life choices. Again, to the extent that Weeden and Kurzban are correct, then one should not expect argumentation to have so much effect on these opinions.
To many this standard for sound argument evaluation might seem depressing, as it means that it’s going to be hard to convince everyone of whatever political and moral beliefs we tend to favor. I, on the contrary, find this quite reassuring. I’d rather live in a world in which it’s too hard rather than too easy to convince people to act against their interests.
This being said, argumentation is still quite powerful. Even if sometimes the changes it leads to are small (at least, smaller than some would like them to be), they tend to be in the right direction. This is true not only when people tackle logical tasks in the lab, but also:
- When they tackle a wide range of others tasks in the lab
- When they tackle a wide range of work problems
- When they tackle a wide range of school problems
- When they discuss moral issues
- When they discuss political issues
- When they tackle scientific problems
I’m not saying that argumentation is all-powerful—I could write another post on some of the reasons it sometime fails to change people’s mind even when it should. But it is vastly more effective than many people give it credit