Dedukcija i kondicionali: eksperiment s mentalnom kronometrijom

Mental model theory (Johnson-Laird & Byrne, 1991) is a dominant theory in human deductive reasoning studies. There are two other classes of theories that deal with human deduction: formal rules theories and content-specific rules theories. This paper begins with a few basic assumptions of the mental model theory, especially those according to which difficulty of inference is proportional to the number of mental models introduced to cognitive system. The aim was to find the function which would describe the inference effectiveness related to different number of mental models. Another aim was to check the justifiableness of response time use as dependent variable in this type of research, besides the usual use of valid answers proportion. Conditional tasks were used in the experiment. According to the theory, number of mental models needed for task solving varied from 1 to 5. All kinds of conditional inference (Modus ponens, Modus tollens) were included in tasks, even those valid in biconditional use but not in classical conditional use (negation of consequent, affirmation of antecedent). Furthermore, different conditionals types (classic conditional and biconditional) and different content conditionals (concrete and abstract) were used. Participants' task was to derive valid conclusion from premises as quickly as possible. Answer validity and response time were measured. Results confirmed some of the mental model theory assumptions. First, simple Modus ponens and Modus tollens comparison showed expected decrease in effectiveness for Modus tollens in both dependent variables. Then, strong participants' bias to perceive conditional as biconditional was observed. Furthermore, participants were more successful in solving concrete content tasks than abstract ones. Also, the effect of conditional directive ness was confirmed. This effect lowered performance when the inference direction was opposite to basic conditional premise. Finally, the main finding of this paper was the function that described decrease of inference effectiveness related to number of needed mental models. This function was almost linear for both dependent variables – proportion of valid answers and response time – and it heavily depended on number of mental models needed for completion of task. All obtained results were highly consistent with mental model theory implications. Key-words: reasoning, deduction, conditionals, mental models, mental chronometry, effectiveness of inference function Primljeno: 3. 07. 2006. PSIHOLOGIJSKE TEME 15 (2006), 1, 151-176