Judgments and decisions based on attempts to disambiguate the given information

2 Comments

• Hugo Mercier 11 June 2011 (18:21)

  Dear Hugo, Thank you for the very kind words and the very valuable in-depth comments. I think that within the literature on judgment and decision making and the literature in social psychology pragmatic inferences are (still) underrepresented; and in explaining particular phenomena with pragmatic inferences, not much is known about the nature of these processes. For example, it is rather unclear whether these processes are more automatic or controlled and under which conditions they operate. Generally, however, it is hard to believe that pragmatic inferences function mostly on a conscious level and that they predominantly involve controlled processing. As other inferences, also these pragmatic inferences should be largely unconscious and automatic as they are frequently used in everyday life. It seems that the Gricean maxims are subordinate to a more general relevance inference. That is, relevance is gleaned from what seems to be related to the ongoing conversation (thread), what is newly communicated information (information gain), what seems truthful (validity), and what was communicated clearly (reliability). That is, the Gricean maxims are meta-cognitive inference rules besides the many other meta-cognitive inference rules that help an individual to determine the relevance of any given information. I fully agree with you that Dan Sperber and Deirdre Wilson’s ‘relevance theory’ goes far beyond Grice’s original account in specifying the psychological processes that are involved in inferring relevance. Also other general models of information may be important to consider. For example, Arie Kruglanski’s epistemic ‘unimodel’ allows for an integration of a wide variety of meta-cognitive inference rules, including those that are linked to the conversational context. As you point out, in many studies on the influence of conversational processes on phenomena in judgment and decision making the conversational rules were discredited / changed via explicit messages (‘details were randomly selected form a pool of information’), presumably triggering intensive reasoning processes. Suddenly then, the effects were different than when these additional explicit messages or hints were not given. Strictly speaking, this does not tell us much about the level of consciousness and controlled processing when this information was not given. The explicit messages simply lead to a different interpretation of the conversational context – and how the person could and/or should use the available information when forming a judgment and making a decision. It is important to also consider that some studies include rather subtle changes of the conversational context / rules. What could this tell us? For example, in one study on conversational recency effects, we (Igou & Bless, 2003, Study 2) presented two-sided arguments (one pro and one con argument) by either one communicator or two communicators. The subtle difference of the conversational context (between subjects), affecting the inferred “relevance” of the first versus the second argument, leading to conversational recency effects only when the information was presented by one communicator. As outlined in that paper, conversational recency effects seem to be largely based on the maxim of quantity. Given that it seems rather unlikely that recipients elaborated on the reasons why arguments were presented by one or more communicators, it can be suggested that more or less implicit processes explain the conversational recency effect. This means that implicit processes seem to be associated with a conversational rule, here the maxim of quantity. In essence, there is good reason to believe that Gricean maxims guide information processing largely automatically just as many other inference rules. Future research on pragmatic inference in judgment and decision making should be more precise in explaining the nature of the inferences (e.g., whether they can be described by Gricean maxims, whether they require cognitive effort), and whether these particular inferences can be integrated into more general models of information processing (e.g., Sperber & Wilson, 1995; Kruglanski & Thompson, 1999; Kruglanski, Pierro, Mannetti, Erb, & Chun, 2007). References Igou, E. R., & Bless, H. (2003). Inferring the importance of arguments: Order effects and conversational rules. Journal of Experimental Social Psychology, 39, 91-99. Kruglanski, A. W., & Thompson, E. P. (1999). Persuasion by a single route: A view from the unimodel. Psychological Inquiry, 10, 83-109. Kruglanski, A. W., Pierro, A., Mannetti L., Erb, H.-P., & Chun, W. Y. (2007). On the parameters of Human Judgment. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 39, pp. 255-303). San Diego, CA: Academic Press. Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition (2nd ed). Cambridge, MA: Harvard University Press.

• Eric Igou 12 June 2011 (14:49)

  Thanks for a great review! I think the questions you ask at the end regarding the operation of the logic of conversation--"To which degree are they automatic (vs. controlled)? To which degree can they be conscious? Are pragmatic inferences relatively dependent on cognitive resources?"-- are crucial. My bet would be that the implicit processes that guide utterance understanding are much more adequately described by relevance theory than by Grice's maxims. By contrast, at the explicit level people may become aware of things like Grice's maxim, and that may guide their reasoning. It's also possible that what happens at the explicit level is indirectly caused by relevance theoretic principles, but does not rely directly on the maxims of conversation. For instance, in the case of framing effects, relevance mechanisms could prompt people to reason more about the answer in some cases--as you mention, if the situation is described in ethical rather than statistical terms for instance. But once reasoning has kicked on, its operation needs not rely on the same principle: people are likely to find reasons for or against a given answer in the text, which have little to do with the maxims of conversation. Maybe the case in which the distinction is the clearest is that of non-diagnostic information. When you tell participants that some information has been selected at random, I doubt that this impact their intuitive understanding of the utterances. After all, some information should still be relevant, so all of it has to be properly understood. By contrast, the explicit disclaimer could have an effect at the explicit level. It seems that non-diagnostic information is taken into account because people find it hard to justify ignoring it. This points to a reasoning problem, not an intuitive problem. When reasoning is given a good reason to ignore the non-diagnostic information--for instance because it could have been random--then it does so. In any case, this is a fascinating area of research, of which more psychologists should be aware since the vast majority of our experiments rely on verbal material, with all the problems that can create.