Moving on to discuss Chapter 2 of Catarina Dutilh Novaes’s The Dialogical Roots of Deduction …
CDN has argued in her opening chapter that “deduction remains a puzzling phenomenon. While a number of accounts have been proposed, none of them is entirely satisfactory.” So how to proceed? In §2.1, she proposes to adopt a “more encompassing perspective” than usual, using a wider range of “methodological approaches”. But what, more specifically, does that come to? She writes “the key idea of the project … was to go back to the roots of deduction. It is obviously inspired by Quine’s classic The Roots of Reference (Quine, 1974).”
Now, in principle, that is just fine by me: I’m a fan of Quine’s (underrated?) book, and I’d love to see a broadly comparable project to his, now focused on questions about deduction. But I’m not sure that’s really what we are going to get, as CDN immediately goes on to say that the way she “conceive[s] of the roots in question is broader in scope than Quine”, to the point where we are going to need “sustained engagement with the empirical literature in psychology, cognitive science, and education sciences”, and also “analysis of historical texts … but combined with a broader historical perspective taking into account developments outside philosophy”. I’m open to persuasion that casting the methodological net so very widely will give us in the end a coherently illuminating story: but the project certainly isn’t sounding very like Quine’s sharply focused project.
In fact, §2.2 is titled “The Different Roots of Reference”, and CDN says more about the variety of considerations she wants to bring to bear on her topic. I continue to have a question though — one I had already about Chapter 1 — of what exactly the topic is. She talks of ‘deductive reasoning’. And sometimes this seems to mean elaborated passages of multi-step reasoning (the sort of thing that’s indeed a bit of an acquired taste). For example, she writes “deductive reasoning emerged as a cognitive technology (though arguably, it remains restricted to circles of specialists), in a way similar to literacy.” If we are focusing on that — the sort of thing that, paradigmatically, mathematicians go on for — we’d all agree that there are will be an interesting “historical question: which cultural processes gave rise to the emergence of deductive reasoning [in that sense]” and an “ontogenetic question: how do deductive reasoning skills arise in a given individual?”. And again, when CDN writes “it is not clear that deductive reasoning abilities in fact confer survival advantages on those individuals”, maybe that case is indeed arguable for reasoning-as-extended-passages-of-argument.
But what about, say, the ability to form disjunctive beliefs, and then use such a belief in a step of disjunctive syllogism? What about, say, the ability to form conditional beliefs (in formulating plans), and then use such a belief in a step of modus ponens when we come to believe the antecedent? Are these cultural variables? Some might suppose, I guess, that a culturally contingent practice of extended deductive reasoning has its roots (to borrow a word!) in capactities for simple-minded deductive reasoning which are not so variable. (And the fact that there are some relatively simple deductive tasks we are not very good at — as the Wason test reminds us! — is no reason for thinking that being good at the ones we are good at doesn’t confer evolutionary advantage!) CDN writes that she is going to “survey the main findings on how deductive reasoning skills emerge and develop in individuals, drawing in particular from the literature on the psychology of reasoning and on mathematics education”. That may indeed be an interesting project — but we’ll have to see what it can tell us about deductive inference in general, as opposed more extended passages of reasoning in particular.
In §2.3, CDN links the project of looking for “roots” of deduction with the claim that the concept of deduction has significantly changed over time. Which again rather points up the non-Quinean aspects of her notion of “roots”. And this gets me wondering about the variety of philosophical enquiries that might be deemed illuminating enquiries into “roots” in some broad sense (and CDN’s usage is nothing if not broad). So before going on to comment on §2.4, where CDN first sketches her dialogical approach, I’m minded — just to get my own ideas straighter — to pause over this, and say something in my next post about the variety of enquiries into “roots” we find in books I greatly admire by Bennett, Hacking and Craig, as well as by Quine.
To be continued.
Thanks again Peter for these additional thoughts. As for the worry concerning the ‘wandering focus’ of the book: I am interested BOTH in simple one-step deductions like syllogisms (Chapter 6 discusses syllogisms extensively, and the experimental literature focuses mostly on these simple arguments) and in longer, more complex deductions like mathematical proofs. Moreover, obviously there is a connection between them: if it is shown that untrained people struggle *even* to perform simple deductions, then of course the longer ones will likely be a problem too. The examples you give of so-called simple deductions in real life can also be explained as instances of defeasible, non-monotonic reasoning on the assumption of normal conditions; often, to tease out truly deductive reasoning from reasoning that in simple cases will look like deductive reasoning, one needs to look at more complex cases. But even for the simple cases, there’s quite some experimental evidence suggesting that people are not reliably reasoning deductively, for example the ‘rose’ experiment by Stanovich et al. that I discuss later in the book (and more extensively in my 2012 book) which shows that people are significantly influenced by the plausibility of the conclusion in their assessments of validity. In this experiment, people deem an invalid syllogism as valid because it has a plausible conclusion, and another syllogism *of the same form* as invalid because it is formulated with made-up words that don’t elicit prior beliefs.
Re ‘what exactly the topic is’, let’s look at two claims from Chapter 2 quoted in your post:
(1) “deductive reasoning emerged as a cognitive technology (though arguably, it remains restricted to circles of specialists), in a way similar to literacy”
(2) “it is not clear that deductive reasoning abilities in fact confer survival advantages on those individuals”
You suggest, in effect, that the key to understanding them is to read ‘deductive reasoning’ as being about ‘elaborated passages of multi-step reasoning’. You even give (1) as an example of where that’s what it ‘seems to mean’.
If it’s only such extended reasoning that CDN has in mind, though, why isn’t that clearly stated? And why do her examples of deductive reasoning that some people find difficult, confusing, or nonsensical include simple inferences such as the “All cows are blue” syllogism on p 8 and A. R. Luria’s “What colour are bears (in Novaya Zemlya)?” experiment, later on the same page? More fully for those who don’t have the text:
(We’re to imagine trying to explain it “to a group of uninitiated, logically naïve interlocutors (say, high-school students)”, and CDN says “In most cases, the reaction will be of mild indignation that such a strange argument can be deemed ‘good’ in any sense whatsoever, given the absurdity of the sentences involved.”)
(CDN introduces this by saying “it has been observed that, in reasoning experiments, participants with little or no formal schooling often resist the very idea of reasoning on the basis of premises that they have no knowledge of.” A participant is quoted as replying “You’ve seen them – you know. I haven’t seen them, so how could I say!?”)
They’re from the part of Chapter 1 on bracketing belief, and it was immediately after describing Luria’s experiment that CDN wrote, “it seems that inferring conclusions from premises while disregarding one’s own doxastic attitudes toward premises and conclusions may require specific training. Yet, it is an integral component of deductive reasoning.”
So I think that’s part of the key to (1) and (2). Deductive reasoning involves bracketing belief, and (to many people at least) that doesn’t come naturally. Then in Chapter 2, she writes, “Recall that in Chapter 1 it was argued that deductive reasoning is by no means a ubiquitous phenomenon among humans, which prima facie suggests that the second option (a cultural, rather than evolutionary, development) is more plausible.”
(Another part of he picture may be that most human reasoning seems to be defeasible, and deductive reasoning isn’t.)
I won’t try to summarise her other arguments for (1) and (2); however, none of them seem to rely on ‘deductive reasoning’ being only about elaborated passages of multi-step reasoning.
As you point out
However, what CDN lists right after the Wason test is syllogistic reasoning. The Wason test seems relatively tricky to me. So I’d agree that not being good at that doesn’t mean we can’t be good at some other forms of deductive reasoning. But what are they, if we’re not even very good at belief-bracketing for syllogisms?
Perhaps modus ponens. But wait! What would evolution actually give us there? A fully deductive, indefeasible, belief-bracketing, likelihood- and familiarity-ignoring modus ponens? Or something more modest?
On what the topic is. I suppose my background worry is the CDN’s focus seems to wander a bit. Some of her claims seems much more plausible about passages of deductive argumentation; other claims she makes (as you say) are about simple one-step arguments. The changes of focus could/should be better signalled.
“Perhaps modus ponens. But wait! What would evolution actually give us there?” Good question. Exactly sort of thing CDN should be asking, I’d say. We’ll have to see how things go. (I haven’t read on at this stage, so that isn’t a loaded remark!)
A more personal footnote. My experience from back in the day when I was interested in the phil of mind and involved in some joint projects with psychologists was that psychologists tended to overinterpret their experimental results (e.g. about the mental capacities of children). By quite a bit! So I’m not well primed to take e.g. reports of psychological investigations like the supposed “bears in Novaya Zemlya” case as particularly weighty evidence of anything much … other than the participants’ suspicion of questions that might be aimed at making them look silly!
I don’t think that CDN is changing focus in quite the way you seem to have in mind. I think that when she says “deductive reasoning emerged as a cognitive technology … in a way similar to literacy” or “it is not clear that deductive reasoning abilities in fact confer survival advantages on those individuals”, she is talking about deductive reasoning in general, including simple forms, not only about elaborated passages of multi-step reasoning.
Some claims “are about simple one-step arguments” because she isn’t drawing a strong line between those cases and more complex ones, and because simple cases are what much of the experimental evidence is about.
I share your view that psychologists tend to overinterpret. Still, Luria isn’t just any old psychologist; there’s more even to his work with Uzbekistani peasants than the question on bears; other people have had similar results in other cases; and then there’s a great deal of other research on syllogistic reasoning and so on. In any case, however sceptical we might be, CDN is taking that evidence seriously.
Besides, claims about what cognitive abilities evolution gave us can be at least as questionable as psychologists’ overinterpretations.
In the end, though, I don’t think it matter greatly for CDN’s project whether biological evolution directly gave us simple forms of deductive reasoning or not.