Introduction – Reasoning as a social device

Introduction — Reasoning as a social device (link to the article)

Hugo Mercier

The social context exerts a very important influence on our decisions, which has not been considered in its full extent by research in decision making. Several strategies are available to take social factors into consideration as much as they deserve. The first is to add a minimal layer of social information and social motivation to the typical methods of decision making. The second is to postulate new mechanisms that pertain only to social aspects of decisions, with their attending biases. The third solution is to reexamine some of the mechanisms classically studied in decision making as social devices. This more radical solution is illustrated here with the case of reasoning. It is suggested that reasoning in fact has a social, argumentative function. An argumentative theory of reasoning makes sense of many puzzling findings from decision making and other areas of psychology. It also provides different, more efficient ideas for debiasing. The third strategy may be usefully implemented for other cognitive mechanisms.

Please post your comments on the paper below.





  • Moin Rahman 15 June 2011 (16:01)

    I am troubled by the use of the term “rationality” to describe how one person wins over the other. Yes, it is rational to the extent that one agent assumes that the conspecific (other agent) is rational and has the capacity to follow the former’s train of reasoning, albeit motivated (reasoning). To that extent, Mercier & Sperber are correct. (I have cited them in my most recent paper on high stakes social interactions.) However, they are wrong to confuse it with the kind of “logical reasoning” that is required to maximize the expected utility. (Human decision making process is riddled with heuristics & biases (Kahneman & Tversky), where some are good and others are sub-optimal — thus, humans do not make perfect logical reasoners. However, overall, humans are best in “satisficing” (Simon) — i.e., go for a good enough (but not necessarily best) outcome, when operating under time pressure and uncertain/lack of info conditions — within the scope of bounded rationality; and in some cases “robust satisficing” (Schwartz, et al.); where a strategy of robust satisficing provides a good enough outcome in conditions of radical uncertainty (Non-equilibrium and volatile system states). To make this long story short, the “Motivated Reasoning” that emerges through argument does not equal to “Rationality” of the kind which seeks to maximize utility in game theoretic or real world situations*. *For example, taking an inordinate time to research the pros and cons of every car through consumer reports, independent reviews, test drives, historical records, etc., etc., and then buy a car. That is impractical but highly logical! The car example I end with is a case of multi attribute utility analysis. Where every attribute (e.g., mileage, comfort, reliability, # of cup holders for some 🙂 is weighted, scaled and rated. And summed scores thus derived are compared across cars.”

  • Hugo Mercier
    Hugo Mercier 17 June 2011 (16:59)

    Thank you for your comment. I’m not sure what it is you are referring to, as I don’t remember using the term ‘rational’ or ‘rationality’ in that context. Other than that, we seem to be in broad agreement.