In two experiments participants gave estimations of the possible outcomes for different types of conditionals. The estimates of conditionals with concrete content were closest to the normative logical representation of conditionals. The estimates of abstract content were more comparable to those of bi-conditionals. While reading the conditional If P, than Q, the initial mental model (p q) shows the relationship between p and q. If this model is sufficient for making logical conclusions, further models will not be constructed. This kind of conclusion is modus ponens (MP) and is performed with high level accuracy. However, the modus tollens (MT) type of conclusion requires the construction of three valid models (p q ; not-p q ; p not-q). This makes MT more difficult to solve, and more time consuming. Conditionals are propositions consisting of two elements (P and Q), with four possible outcomes. Some outcomes are also represented by mental models. However, because of cognitive parsimony and different biases, some mental models are more likely to be generated, and some less. The aim here was to examine the subjective possibility of different outcomes arising from these conditionals. Such probabilities were estimated for different types of conditionals. The first experiment (N=51) had a 3x3 factorial design in which the type of conditional (regular, bi-conditional and causal) and content (abstract, concrete and social contract) was manipulated. For each conditional, four possible outcomes were presented ; the participants’ task was to estimate the possibility for each outcome (0-100%). The order of the presentation of conditionals and corresponding outcomes was randomized. Results showed that outcome (p q) (the initial model in all situations) was highly represented. Furthermore, it seemed the abstract conditional appeared as a bi-conditional as the outcome (not-p q) was infrequent, i.e. not probable. However, by making conditional content more concrete, the probability of such outcomes increase ; the representation of this conditional becomes more like a normative representation based on rules of logic. In the second experiment (N=45) the effect of content was additionally manipulated through conditional plausibility. The experimental design included a two-factorial design (6x2), where the plausibility of the conditional was manipulated from highly implausible, to highly plausible. In addition to this, the conditional and bi-conditional were the two types of conditionals used. This procedure was similar to that in Experiment 1. The results showed lower estimates of possibility for valid outcomes (p q) and (not-p not-q) in implausible conditionals when compared to plausible ones, while the invalid outcome (p not-q) had higher estimates. These results demonstrate the importance of a known content in the construction of mental models. The most interesting result perhaps was obtained for the (not-p q) outcome. In the situation of implausible conditionals, participants do not make a difference in possibility estimates for this outcome of conditionals and bi-conditionals. As the plausibility of conditionals increased the estimate of this outcome increased in the case of the conditional, and decreased in the case of bi-conditionals. The results in this experiment were similar to those in Experiment 1. In conclusion, this research confirmed the idea that more plausible conditionals with concrete content have a higher probability of representation in accordance with the rules of normative logic. Abstract conditionals have a tendency to be represented by only one mental model, or in the form of bi-conditional. |