This article outlines a theory of naive probability. According to the theory, individuals who are unfamiliar with the probability calculus can infer the probabilities of events in an extensional way: They construct mental models of what is true in the various possibilities. Each model represents an equiprobable alternative unless individuals have beliefs to the contrary, in which case some models will have higher probabilities than others. The probability of an event depends on the proportion of models in which it occurs. The theory predicts several phenomena of reasoning about absolute probabilities, including typical biases. It correctly predicts certain cognitive illusions in inferences about relative probabilities. It accommodates reasoning based on numerical premises, and it explains how naive reasoners can infer posterior probabilities without relying on Bayes's theorem. Finally, it dispels some common misconceptions of probabilistic reasoning.

An experimoiit was performed to determine whether the use of realistic materials mould improve performance in a deceptive reasoning problem. The task involved selecting from a set of envelopes those which, if they were turned ovor, could violate a given rule. The rulo concerned either a realistic relation ('if a letter is sealed, then it has a 50 lire stamp on it') or else an arbitrary relation between symbols ('if a letter has an A on one side, then i t has a 3 on the other side'). Twenty-two of the 24 subjects made at least one correct answer with the realistic material but only seven of them did so with the symbolic materials. The verbal formulation of the rule was also varied but yielded only a marginal interaction with tho main variable. It is argued that the critical factor is the intrinsic connexion betweon items rather than their specific nature.It is a w-ell-established fact that the content of a problem may have a significant effect upon insight into its underlying structure. Perhaps the clearest demonstration of this phenomenon in a purely deductive task is Wilkins's (1928) classic study of syllogistic inference. She discovered that problems with a familiar everyday content were generally easier than those with a purely symbolic or totally unfamiliar content : her subjects committed fewer fallacies even though the familiarity of the material provided no cue to what was, or was not, the valid conclusion. To anyone approaching thinking from a strictly formal point of view (such as the logically oriented psychologist or genetic epistemologist) this finding is both surprising and perplexing because in making a deduction the same mental operations are presumed to be carried out regardless of content. Hence, why should it be harder to execute them with one sort of material than with another? Our investigation was designed to re-examine the phenomenon and to try to answer this question.The particular deductive problem that we chose to study was one developed by Wason (1068). The subject is presented with the four cards shown in Fig. 1, together with the following rule : If a card has a n A on one side, then it has a 3 on the other side.He knows that each card has a letter on one side and a number on the other side; and his task is to choose just those cards which it is necessary to turn over in order to discover whether the rule is true or false. This is an extraordinarily deceptive problem even for the most intelligent of sub-

Reasoners succumb to predictable illusions in evaluating whether sets of assertions are consistent. We report two studies of this computationally intractable task of "satisfiability." The results show that as the number of possibilities compatible with the assertions increases, the difficulty of the task increases, and that reasoners represent what is true according to assertions, not what is false. This procedure avoids overloading memory, but it yields illusions of consistency and of inconsistency. These illusions modify our picture of human rationality.

No abstract

This paper discusses the insufficiency of approaches which consider reasoning processes in the light of logical models that neglect social context. Even when tasks are solved individually, the fact that subjects imagine the social context of the problem may be important. That is, they may imagine the goals of the promisor, promisee, and agent of the promise, when the promise, for example, is utilized as a rule in Wason's selection task. Theories such as that of Cosmides (1989), which explains content effects in terms of standard social contracts, do not seem to be sufficient ; they should be completed in order to explain differences like those obtained by Gigerenzer & Hug (1990) between what these authors call "unilateral" and "bilateral" cheating options. This work concludes with a general discussion of the relation between reasoning and social psychology.

No abstract