Intuitive and reflective inferences

vs. argumentative contexts If our approach is right, reasoning should be more easily triggered in argumentative situations. We should therefore expect better performances on reasoning tasks where the participants are placed in such situations. Standard theories make no such prediction. If anything, argumentative contexts should increase the cognitive load since they involve taking into account different opinions. It is now well established that performances on computationally trivial logical problems can be dismal. As Evans states in his review of the literature on deductive reasoning: ―it must be said that logical performance in abstract reasoning tasks is generally quite poor‖ (Evans, 2002, p. 981). The simplest way to compare the abstract context of a being communicated or revising his trust in the speaker. He will tend to choose the solution that brings less incoherence, and this will often be to lower trust in the speaker. classical reasoning experiment with an argumentative context is to get the participants to discuss the problem in groups. Among the great many studies on group decision making, the most relevant are those bearing on problems that have a demonstrably correct answer—and are thus analogous to the tasks used in most reasoning experiments. It has now been repeatedly shown that, provided certain minimal conditions are met (the good answer must be accessible to at least one of the participants for instance), what is observed is that in such contexts, if one of the participants has the correct answer, then the other members will get to it too. This has been shown for mathematical tasks (Laughlin & Ellis, 1986; Stasson, Kameda, Parks, Zimmerman, & Davis, 1991), ̳Eureka‘ problems (in which the correct solution seems obvious in retrospect— Laughlin, Kerr, Davis, Halff, & Marciniak, 1975), and Mastermind problems (from the board game—Bonner, Baumann, & Dalal, 2002). In all these cases the performance of groups tends to be at the level of the best participants taken individually. The experiments carried out by Moshman and Geil (1998) illustrate this point dramatically. The experimenters had participants solve the Wason selection task, either first individually and then in groups, or directly in groups. In both cases, the performance of the groups was impressively higher than that of the participants who were solving the problem individually: 75% of the groups found the right answer, compared to 14% in the solitary condition 6 . According to the theory advocated here, this dramatic improvement is due to the fact that when they have to solve the problem in groups, participants have to argue and debate, and that this activates their reasoning abilities in such a manner that they are able either to come up with the correct solution, or at least to accept it and reject the incorrect ones. Of course, this is not the only possible interpretation of these results. An alternative interpretation might be that the smartest participant gets it right and the others recognize her competence and accept her answer without reasoning (explanation hinted at by (Oaksford, Chater, & Grainger, 1999). Another possible interpretation is that the participants are simply sharing information, and not reasoning. These explanations are hard to reconcile with the following facts. First of all, information sharing is often insufficient to solve the task. For example, in the Wason selection task, it will often be the case that a participant has wrongly selected a card and another has rightly rejected it. In that case, sharing information won‘t do the trick: participants have conflicting pieces of information, and they have to pick the correct one. This means that conflicts and debate should occur. An analysis of the transcripts of such 6 A similar effect—if a bit less dramatic—was observed by (Maciejovsky & Budescu, 2007). experiments will show that such is indeed the case (Moshman & Geil, 1998; Trognon, 1993), and there is a large literature showing that conflict is often the crucial factor that allows groups to outperform individuals (see the references in Schulz-Hardt, Brodbeck, Mojzisch, Kerschreiter, & Frey, 2006). In some cases, conflicts will even lead a group in which no individual had the correct answer towards it—provided that not everyone makes the same mistake to start with (this happened in some of the groups studied by Moshman and Geil and is known in developmental psychology as ―two wrongs make a right‖ (Glachan & Light, 1982; Schwarz, Neuman, & Biezuner, 2000) and as the ―assembly bonus effect‖ in social psychology (Kerr, Maccoun, & Kramer, 1996). The explanation based on the recognition of an expert is also hard to reconcile with the presence and importance of such conflicts. One could even argue that the opposite in fact happens: a person is recognised as an expert because she uses good arguments—so participants must use reasoning to discern good arguments in the first place (see Littlepage & Mueller, 1997 for evidence in that direction). Finally, we can also rule out an explanation based on general motivation: one might think that participants are more motivated—will make greater effort—to solve any task in group. This would be quite surprising, however, given the importance of social loafing in groups (Karau & Williams, 1993). Moreover, if motivation was the problem, it should be alleviated by monetary incentives. However, in line with the general observation that money tends to have no effect on performances in decision making tasks (Camerer & Hogarth, 1999), it has been shown that monetary incentives do not increase the performance in the Wason selection task (Johnson-Laird & Byrne, 2002; Jones & Sugden, 2001)—a result in sharp contrast with the dramatic improvement in group settings.