A mental model theoretical explanation of facilitation in probabilistic reasoning

We present an experiment in probabilistic reasoning conducted with 200 participants, whose task was to forecast the outcome of two independent events in a probability problem. We evaluated predictions of the Mental Model Theory of Extensional Reasoning based on Johnson-Laird, Legrenzi, Girotto, Legrenzi and Caverni (1999), 'Naive Probability', according to which people construct the probability of an event from the different possible ways in which the event could occur, and use the proportion amongst mental models to determine the probability of the event. We observed that the formulation of questions with a single change in content has a direct impact on performance. Formulating the premise of a coin tossing problem with a distinguishing attribute such as gold versus silver coin, improves the performance significantly from 64% to 79% correct. To explain the resulting facilitation effect, we apply the Mental Model Theory of Extensional Reasoning and the 'tagging' of mental representations. Based on our observation, we adapted the theory to incorporate the 'tagging' of mental models with attributes. Our explanation is that the mental models are tagged with the additional attribute, which in turn facilitates the distinction between otherwise confounded mental models and thus improves combinatorial reasoning performance. This facilitation with distinguishing attributes also significantly reduces the equiprobability bias noted in the control condition. The Mental Model Theory of Extensional Reasoning is a more precise and algorithmic description of the process to elaborate mental representations which define the sample space in probabilistic tasks.status: publishe