Elicitation of Subjective Probabilities in the Light of Signal Detection Theory

Most theories of decision-making under uncertainty are based on the idea that individuals weight events by some (transformation of) subjective probabilities. Elicitation of these subjective probabilities has thus became a major concern for evaluating and applying these models. In order to do so, one needs: (i) to be sure that such subjective probabilities actually exist in individuals’ minds and, (ii) to be able to infer these subjective probabilities from behavioral data. Cognitive sciences demonstrate that decisions in perceptive tasks are based on probabilities encoded at the neuronal level. Moreover, Signal Detection Theory (SDT) provides a theoretical model, confirmed by experimental neuronal data, that relates these probabilities to behavioral data. Here we use three different elicitation rules to measure individuals’ confidence in a perceptive task, and compare the results with the theoretical predictions based on SDT. We find that subjective probabilities elicited by a specific rule, the Matching Probabilities, fit very well the theoretical predictions. We also show that these results are consistent with those obtained for the same subjects in a non-perceptive task (a knowledge and logic quiz). This paves the way for extending our findings to non-perceptive tasks. ∗Financial support from ANR grants is acknowledged (Riskemotion ANR-08-RISKNAT-007-01; Feeling of control BLAN07-2 192879). †GREQAM, CNRS, Aix-Marseille University. [email protected]. ‡Corresponding author: [email protected]. CES University of Paris 1, Paris School of Economics. Address: 106-112 boulevard de L’Hôpital, 75647 Paris Cedex 13 France. §CES University of Paris 1, CNRS. [email protected].