Document Type


Degree Name

Master of Arts (MA)




Faculty of Science

First Advisor

Robert Gebotys

Advisor Role

Thesis Supervisor


Human revision of opinion has often been described as a list of heuristics and cognitive biases. The following study examined the possibility that one of these biases, conservatism, is a result of dissimilarities between the laboratory and the real world. In two experiments, participants ranked problems on a psychological scale of realism, solved several realistic and artificial problems and completed a questionnaire to assess their probability knowledge base. Results showed that participants classified problems into different groups according to their level of realism. Whereas their responses to the artificial problems were conservative, their responses to the realistic problems were very close to a Bayesian solution. Other aspects of the problem had a significant effect on participants’ performance as well. Mathematics were used more frequently when problems were artificial. Participants generally used base-rate/likelihood information when it was high, often neglecting to use this information when it was low. Experts’ solutions were not better than those made by novices; however, experts tended to solve the problems mathematically, whereas novices did not use mathematics as frequently. It is proposed that the process of solving realistic and artificial problems may be qualitatively different and therefore results cannot be generalized from one to the other.

Convocation Year


Convocation Season