A new experimental approach to testing quantum models of conjunction error.

Summary Classical probability theory requires that the probability of the conjunction of two events be lower than the probability of one of the events alone. However, empirically, subjects do not always judge this way: this is the conjunction error. One of the currently promising explanations of this paradox is based on so-called "quantum" models, developed from the mathematical tools of quantum mechanics. But which versions of these models can precisely be used? In particular, can the simplest versions, called non-degenerate, be sufficient? This paper tests these non-degenerate versions through some of their experimental predictions, using an original experimental protocol. The results obtained in the laboratory suggest that non-degenerate models are not empirically adequate, and that future research concerning quantum models should be directed towards degenerate models.