Wednesday, 15 June 2011

Mercier and Sperber on the origins of reasoning

I just wrote a post at NewAPPS on a very interesting article on the origins of reasoning by Hugo Mercier and Dan Sperber. As the post is a bit too heavy on the psychology side of things, I suspect it is not exactly within the remit of M-Phi, so I won't cross-post it. But perhaps some M-Phi'ers will be interested, so here is the link. It should be relevant for discussions on rationality and Bayesian epistemology, which many of us M-Phi'ers are interested in.

Mercier and Sperber argue for the bold view that the real function of reasoning is argumentative, and thus not to improve knowledge or make better decisions. Here is the abstract:

Reasoning is generally seen as a means to improve knowledge and make better decisions. However, much evidence shows that reasoning often leads to epistemic distortions and poor decisions. This suggests that the function of reasoning should be rethought. Our hypothesis is that the function of reasoning is argumentative. It is to devise and evaluate arguments intended to persuade. Reasoning so conceived is adaptive given the exceptional dependence of humans on communication and their vulnerability to misinformation. A wide range of evidence in the psychology of reasoning and decision making can be reinterpreted and better explained in the light of this hypothesis. Poor performance in standard reasoning tasks is explained by the lack of argumentative context. When the same problems are placed in a proper argumentative setting, people turn out to be skilled arguers. Skilled arguers, however, are not after the truth but after arguments supporting their views. This explains the notorious confirmation bias. This bias is apparent not only when people are actually arguing but also when they are reasoning proactively from the perspective of having to defend their opinions. Reasoning so motivated can distort evaluations and attitudes and allow erroneous beliefs to persist. Proactively used reasoning also favors decisions that are easy to justify but not necessarily better. In all these instances traditionally described as failures or flaws, reasoning does exactly what can be expected of an argumentative device: Look for arguments that support a given conclusion, and, ceteris paribus, favor conclusions for which arguments can be found.
I have a lot of sympathy for this approach, but ultimately I think it is flawed, as I argue in my post at NewAPPS.


  1. Thanks, Catarina - good stuff.

    "M-Phi'ers ..."

    M-Phistas? Es mejor, verdad?

  2. "M-Phistas? Es mejor, verdad?"

    Si, es mas poetico :)

    Yes, it's interesting stuff, and it has many commonalities with my own take on reasoning, logic and deduction. But their story is supposed to apply to *all* forms of reasoning, whereas I argue that deductive reasoning originates from a particular form of engaging in adversarial dialogues, which then gave rise to the deductive method as we know it.

  3. Thanks! There is an article on this in NYT:

  4. Florian, I linked the NYT article in my NewAPPS post, which you should go read now! :P It's all about the dialogical stuff we were talking about last week (btw, I'll be back in Sep., so we can continue talking then).
    In fact, I had known about the Mercier&Sperber article since last year, but seeing the NYT article yesterday gave me the idea of writing a blog post about it.

  5. Will do. Look forward to continuing our discussions in Sept.

  6. I could rant, but I won't. Just this quick comment: There seems to be some fuzziness/ambiguity surrounding the term 'function' here. I strongly believe that the function of reasoning is to determine what follows (in roughly Tarski's sense of follow) from various sets of premises. Of course, I have a very particular sense of 'function' in mind here. On this reading, however, all of the empirical evidence of the sort mentioned above is just irrelevant.

    Now, obviously, Mercier and Sperber have some other notion of function in mind. But I'm just not clear what notion that is.

    Okay, so it was a bit of a rant.

  7. What they mean by 'function' here is understood in evolutionary terms: what evolutionary advantage could have favored humans who could reason over those who could not, so that these would be selected for? The usual story is that reasoning is good for knowledge and decision-making (thus making those individuals 'fitter'), but here they are claiming that the real 'edge' that being able to reason gave to the individuals in question is to be able to win arguments and convince people.

    Personally, I think that the evolutionary angle is one of the weakest parts of the thesis. First, because I'm against ultra-adaptationist stories (reasoning may well be a byproduct of some other adaptive feature); second, because it seems to disregard that, more than winning arguments, humans are at their strongest when they collaborate. Relatedly, as I said in my NewAPPS post, I submit that the wide majority of our dialogical interactions are of the cooperative kind, and there are reasons why there would be much more evolutionary pressure to select for individuals who are able to engage in cooperative communication than in adversarial communication.

    But anyway, a question for you Roy: is there any aspect within the philosophy of logic and/or mathematics where empirical evidence *might* be relevant, according to you? Just so that I understand your general position better.

  8. Catarina,

    Okay, here comes the rant!

    First off, the main issue is not that the evidence that they are mobilizing is irrelevant to philosophical questions regarding logic or mathematics. On the contrary, I try to stay at least casually familiar with this sort of work for that very reason: Our philosophical account of correct reasoning would be worthless if it didn't connect in deep ways with empirical studies regarding how we do reason (and, in particular, what kinds of concepts we mobilize and what inferences we do accept and reject involving those concepts).

    That being said, I worry that there is often to much slippage between the sort of conclusions that evolutionary studies draw with regard to reasoning and the sort of conclusions that philosophers (or at least I) want to draw. The issue, again, is merely this: Scientists, including scientists who study reasoning, are asking different questions than we are (or, at least, than I am). As a result, we should be wary of thinking that the answers to their questions have any bearing on the answers to our questions, even if the data they collect is useful to us in terms of identifying central concepts, important inference patterns, etc. that we might want to examine philosophically (as well a providing a wealth of data helpful for generating intuitions and the like).

    I should point out that I have a very narrow view of what counts as a philosophical question, and what counts as philosophy more generally. To put it in overly simplistic terms, I take the hallmark of philosophical problems to be their involvement with issues that are either foundational or normative. If you are answering a question or defending a position that doesn't involve either (i) the foundations of some discipline or discourse or (ii) 'ought' statements of some kind, then you likely aren't doing philosophy. Of course, that doesn't mean that philosophers shouldn't be studying, worrying about, thinking about, or working with, other kinds of questions or accounts. But that just goes to show that philosophy is hard, and often answering philosophical questions and developing philosophical views involves a lot of work with non-philosophical stuff.

    So, to put all this in more concrete terms: When a scientist says something like:

    (1) "Evolution selected individuals who could reason, and in particular, who applied reasoning skills that were effective for persuasion, rather than skills effective for discovering truths or collaborative communication."

    then I will certainly pay attention and think that the claim is relevant to the questions I worry about (although I would have doubts about that particular statement, for reasons overlapping with your own). When a scientist says something like:

    (2) "A study of evolution demonstrates that reasoning skills that are effective for persuasion, rather than skills effective for discovering truths or collaborative communication, are the right (or best, or correct, etc.) reasoning skills"

    however, then my reaction is either "Wow, you're genuinely insane!" or, much more likely, "Well, clearly you mean something different than I do by "right" (or "best", or "correct")."

    Now, the article in question doesn't go this far, but I do often hear this sort of thing from philosophers and people who take this evolutionary stuff seriously (although I don't know why that worries me so much, given all the other crazy stuff I hear from philosophers!) The claim is usually supported by pointing out that the way we actually reason isn't just a matter of evolutionary luck, but that there are reasons why those skills were selected for, and that makes them the 'right' skills.

    (Continued below)

  9. (Continued from above)

    The point, I guess is this: Let's pretend that evolution selected for acceptance of the law of excluded middle (LEM) because that rule was immensely advantageous in argumentation and persuasion (since, although I have formulated it in a rather loose and silly way, this doesn't seem all that implausible). So, on the evolutionary understanding of 'right', LEM is 'right'. As interesting and thought provoking as that might be, and as useful as I might find that fact to be when trying to understand how language and logic works in general and where it came from, I just don't see it as being at all relevant to whether LEM is 'right' in the philosophical sense - that is, whether LEM is one of the rules of logic that necessarily preserves truth in virtue of logical form (or whatever). In fact, on my view LEM isn't 'right' (in the philosophical sense), and since this belief is based on pretty complicated a priori considerations involving knowability, determinacy, and their connection to logical constants and their meaning, I just don't see how any empirical evidence could ever sway me to change my verdict on LEM (unless the empirical evidence was so overwhelming that it forced me to revise basic, fundamental principles regarding meaning, knowledge, and determinacy).

    Anyway, you know it's time to stop when the website won't let you put everything in one comment. Maybe I'll write more later.

  10. Thanks for the rant, Roy! :)

    Lots I could say here, but let me focus on some of the main points.

    There are all kinds of uses of evolutionary arguments in psychology and philosophy, from very bad ones to very good ones. To name one of the bad ones: Cosmides' contention that we have an innate cheating mechanism on the basis of her results on a particular version of the Wason selection task. In fact, this holds of evolutionary arguments everywhere, including biology; many are just-so stories, others are among the most explanatory arguments one could hope for. So it doesn't make sense to discuss uses of evolutionary arguments in general; it needs to be looked into individually. And as I said before, I don't like the way Mercier and Sperber introduce evolutionary considerations in their argumentation.

    As for your conception of philosophy as having to do with 'origins' and 'normativity': although perhaps narrow, it strikes me as a very plausible one. But then, I think in both cases attention to empirical data can be very important. I think most of my work falls within the 'origins' category; I'm in the business of trying to understand where very basic concepts come from, in particular the tacit assumptions made along the way. But take necessary truth-preservation: where does the very idea of necessary truth-preservation come from? The experimental results tell us that untrained subjects typically do not seem to engage in anything that looks like the necessary truth-preservation game; in most cases, a strong component of non-monotonicity is present. But necessary truth-preservation IS an important concept, and attention to empirical data allows you here to give a negative answer to the question of its origin; no, it's not the kind of thing that untrained subjects seem to be doing, contrary to what a whole range of important people think (Kant, Piaget etc.).

    But then, where does it come from? My tentative answer is that it emerged in the context of rather specific adversarial dialogical practices in Ancient Greece, which then gave rise to a particular method of argumentation both in philosophy (logic) and math. Here the question of normativity becomes very pressing: what is the source of normativity for the canons of the deductive method? I've never come across any satisfactory, non question-begging answer to this question. Either way, in my story it's not going to be an evolutionary argument, because for me it's essentially a cultural process (which doesn't commit me to social constructivism either), and here again you could appeal to empirical data for a negative answer (say, if it turns out that reasoning according to the canons of deduction does *not* increase your fitness, as some computer simulations by B. Skyrms seem to suggest).

    But before discussing LEM or any specific case, we need to get clear on the very idea of necessary truth-preservation. And btw, I don't believe in 'in virtue of logical form' as the right explanation (see e.g. my forthcoming Synthese paper on logical hylomorphism), but my arguments are purely conceptual/historical, no appeal made to empirical data here :)

  11. Catarina,

    I agree with most of your points (and certainly agree quite strongly about your point that we need to judge these sorts of arguments individually, and not en masse, but of course the sort of answer I gave above kind of forced me into the latter, somewhat generalizing mode). Nevertheless, I do have one bone to pick.

    When I sketched my conception of philosophy, I mentioned normativity and foundations. When you glossed my view, however, you switched to normativity and origins. I think this switch is important. My feeling here is that the difference between my foundations and your origins is similar to the classic order-of-explanation vs. order-of-discovery thingamajiggy (that's a technical term!) When I talk of foundations what I have in mind is something like: whatever it is that ultimately justifies the way we use certain concepts or engage in certain practices, where this might be very different from - and might even be completely independent of - how those concepts and practices came about (which would be something like your origins?).

    A good example of this sort of thing is my take on logicism. I (among others) argue that arithmetic practice (both by professional mathematicians and everyday counters), and certain meta-attitudes we take towards arithmetic (that it is a priori, analytic, etc.), are justified based on the fact that we could stipulate Hume's principle as a sort of implicit definition of number, and then explicitly derive the second-order Peano axioms (and everything that follows from them). It is totally irrelevant to my story that no one outside of Minneapolis, Philadelphia, St Andrews, Aberdeen, Glasgow, Sheffield, or Stirling has ever actually carried out the suggested stipulation and re-derived their arithmetic from abstraction principles. Rather, our actual practices and attitudes (or at least, our best practices and attitudes, which will be judged in terms of how well they accord with the logicist foundation) are justified merely in terms of the possibility of arriving at these practices and attitudes via neo-logicist means. So in this particular case there is a clear, sharp distinction between origins and foundations.

    Of course, that doesn't mean that actual practice is not immensely important. Had our actual practices and attitudes (again, both those of the professional mathematician and those of the everyday counter) diverged substantially from those sanctioned by the logicist story, then the opponent of logicism would be completely justified in rejecting logicism along something like the following lines: "Maybe your logicist story explains some concept, but it ain't the concept actually used by mathematicians and counters-on-the-street. So go back to your office and come up with a foundational account of some concept that we are interested in!" In fact, Richard Heck pushed a line somewhat like this for a while, but I think it was because he wasn't (and, more importantly, we neo-logicists weren't) clear yet on the origins vs. foundations aspect of logicism (he would likely describe it differently, so if you are reading this, Richard, please chime in and correct this description!)

    To sum all this up bluntly: Insofar as we are talking about logicism, I don't care a whit about how mathematical practices and attitudes actually evolved or where they came from, so long as those practices and attitudes are ones that are in the end supported by the story given above. (I do, of course, care about these issues for other reasons, including the fact that the origins of these practices are interesting in and of themselves! That doesn't affect the point, however.)

    (continued next post)

  12. (continued from previous post)

    None of this should be taken to imply that I think that there is something wrong with studying the origins of various practices that philosophers are interested in. As I already noted, origins (whether cultural, historical, or evolutionary), at worst, certainly fall under the heading of non-philosophica-stuff-that-philosophers-need-to-know (just as mathematics isn't philosophy, but you aren't going to be a good philosopher of mathematics without knowing a good bit of mathematics!)

    There is a more important point, however, one I missed in previous posts. My view - that studying the origins of certain practices and attitudes isn't doing philosophy (although it might be important for doing philosophy) - is not just a function of my view that philosophy is foundations plus normativity. In addition, we need the additional premise that origins are not an integral part of either the foundational story or the normativity story. I accept the latter premise (at least, I do for those areas I do most of my work in), but this is because of the particular philosophical views I accept (logicism in the case of philosophy of mathematics, similarly old-fashioned approaches in philosophy of logic). If one adopts a different approach, then origins might be absolutely central and crucial. (For example: One can imagine a version of fictionalism where the origins of the practice are a critical component of explaining the practice, since the content of the fiction on this imagined view might be nothing more than the cumulative history of its use).

    Thus, I suspect some of the differences between us (whatever exactly those are - I think we are still figuring that out) is both that I have a very narrow working definition of philosophy (one which you seem at least sympathetic to) and that I have views that, when conjoined to that narrow conception, make origins somewhat irrelevant (and it is here where I suspect we differ dramatically, since your own views provide origins with a much more central role?)

    Anyway, I am not sure if this will help clarify my views, or is just a (hopefully interesting) tangent.

  13. Hi Roy,

    You are absolutely right when you point out that my move from your 'foundations' to my 'origins' was much too quick! Indeed, it says enough about how *I* view foundational enterprises generally speaking (Freudian slip, you might say).

    But I submit that even in YOUR conception of foundational work, attention to origins may be, and usually is, of crucial importance. So I'm not saying that the search for origins exhaust the foundational project, but that it is often an important component of it. One of the roles I ascribe to examining origins is that historical analyses of how a given conception or doctrine came about are often very useful to pinpoint and isolate assumptions that are made along the way. Now, isolating assumptions is, I'm sure you will agree, full-blown foundational work -- in particular when it turns out that some of these assumptions are contentious or rely on other assumptions that are contentious, which then suggests that the doctrine in question rests on shaky grounds. Again, I refer to my 'Reassessing logical hylomorphism' paper forthcoming in Synthese, where I try to expose the contentious assumptions underlying the idea that arguments are valid in virtue of their logical form and the whole business of demarcating logic by means of a demarcation of the class of logical constants.

    To think of the difference between foundations and origins only in terms of context of justification vs. context of discovery seems to me to disregard the contribution that the study of 'origins' can make to foundational analysis as you describe it. So in my project on deduction to be commenced shortly, I will try to show that a dialogical reconceptualization of deduction (which has been formulated on the basis of 'origins') can shed light on all kinds of purely philosophical puzzles concerning deduction (e.g. the so-called 'scandal of deduction').

  14. OK, can you send me the "Reassessing..." paper? You have convinced me that I might need to reassess my view (or, at the very least, assess whether I need to reassess!)

  15. "Reassessing..." is here:

    Open Access, courtesy of the Dutch government :)