In their book The Enigma of Reason, out last April, Hugo Mercier and Dan Sperber have written an important book on an important but neglected topic. They argue first that humans, and only humans, have a brain module that handles abstract reasoning:
How do you and your wife make key decisions?
Why did we gradually arrive at more accurate beliefs, if that is not the goal of abstract reasoning?
This is actually quite clear in the book. Arguments for our positions aren't geared to truth, but the counter-argumenters of listeners are. And good counter-argumenters improve the quality of subsequent arguments. Listeners being better than talkers, the asymmetry that Robin contests, isn't really the product of overlooking anything; it is at the center of their argument.
[Less Wrong, too, underestimates its divergence from Mercier and Sperber. First, they demolish Kahneman's two-factor theory, which is a basis for Rationalist thinking. According to the Rationalists, we can improve our thinking by learning to reason better. But Mercier and Sperber show that great thinkers were not great reasoners and were not independently rational.]
I suspect the true account is a hybrid account, in which reason serves both a truth-seeking function and more cynical functions, with the former providing cover for the latter. If reason has zero truth-seeking function, it seems like doing an impressive dance or something could serve the same role as an impressive argument.
It seems to me that some of this could be clarified by using the framework of Trivers' theory of the subconscious (as published in the forward to The Selfish Gene, 1976): There is a reasoning capability whose purpose is to assess the correct actions to take. This capability is not permitted to speak, and so we call it "subconscious". There is another reasoning capability whose purpose is to speak, and thus engage in all of the social acts involving speaking. We call it "conscious". For optimal performance, the conscious capability is not always granted access to the complete information that the subconscious capability possesses.
In the narrower sense of delivering abstract argumentation, it seems like we are dealing with the conscious capability, and then in a fairly narrow part of its activity in most humans (other than professors, op-ed writers, and other members of the chattering class). Indeed, in many humans, that behavior pattern is probably used very little. There's no reason to expect the typical peasant to be well-practiced at it or to be motivated to do it. A real test would be to put the peasant in a situation where getting the right answer will ensure their survival. I'll bet they suddenly become far more definitive in their analysis!
There is also the complication that carefully assessing an argument is a time-consuming activity, and time is scarce. So people often use much cruder filters for first deciding whether to pay attention to an argument at all.
You write, "Mercier and Sperber suggest that arguing used to be different, and better: [...] 'When the overriding concern of people who disagree is to get things right, argumentation should not only make them change their mind, it should make them change their mind for the best.'"
I don't exactly follow this thread. It's true that when the overriding concern is to discern the facts of a situation, people will be receptive to argumentation. And that seems to be the case in practice. But in the majority of situations, the central question is not what the facts are but how to divide resources between different claimants. E.g., the placing of the border between the United States and the Iroquois was never one of objective facts but rather one of relative power. You can see this distinction in current politics; in the few situations where there is broad agreement on what the goals are (because everybody thinks their interests are closely aligned), the matter is handed over to "technocrats". E.g., manipulating the US dollar money supply. In matters where people seem to hold dearly to false information, invariably it is a situation where that "information" justifies a decision which favors the its believers over some other faction.
Interesting. I haven't read Mercier and Sperber's book. I suppose it largely overlaps with this long article of theirs though (which I have read). I recommend the comments by other scholars.
The interactionist account explains why "throughout the ages, smart physicians felt justified in making decisions that cost patients their lives" well, whereas that fact may be a problem for the intellectualist account. However, prima facie, it might seem that the interactionist account would have problem explaining intellectual progress. Why did we gradually arrive at more accurate beliefs, if that is not the goal of abstract reasoning?
I'd be interested in whether the authors discuss that question, or whether you have thoughts on that.
The question appears somewhat analogous to creationist vs vs evolutionary accounts of the development of life. Evolutionary theory can explain inefficiencies better than creationist accounts. However, prima facie, it seemed that evolutionary theory could not explain why many features of animals and plants are so highly functional.
Of course, it turned out that evolutionary theory could explain the latter. So, I'm not saying the interactionist account could not explain intellectual progress. But I'd be interested in a fine-grained explanation of why it is possible on the interactionist account - in line with the fine-grained evolutionary explanations of functional features of animals and plants.
In the last quote it looks like Mercier and Sperber are saying, not that arguing used to be different and better, but that it gets used more in the face of big collective decisions in small-scale societies (which more of us belonged to in the past) than in large-scale ones.
It seems like what we consider good, high-status reasoning is communication in "far" mode. It's a way of sending a message to an unknown audience in far-off places and times, where we can't rely so much on local culture and reputation. To cross the gap, there is a greater attempt to find universal arguments. In the limit we find mathematics, which as far as we know is independent even of the peculiarities of human thought, or so we hope.
The incentives to write in "far mode" are often weak, relative to more immediate, local communication. It might not happen at all unless the local culture encourages it. Survivor bias might cause us to think "far mode" arguments were more common in the past than they are today.
How much persuasion is self persuasion and how much testable and tested? How much questioning is self questioning and how much experienced and shared? How much is accepted as consistent with beliefs and experience and confirmation of priors or ignored as unimportant or irrelevant?