The role of subjective linear orders in probabilistic inferences.

In probabilistic inferences concerning which of two objects has the larger criterion value (e.g., which of two cities has more inhabitants), participants may recognize both objects, only one, or neither. According to the mental-toolbox approach, different decision strategies exist for each of these cases, utilizing different probabilistic cues. Possibly, however, participants use these cues to build a subjective rank order that involves all objects, irrespective of their recognition status. The decision process then simply utilizes the distance between two objects in one's subjective order. We tested the role of such linear orders in reanalyses of existing data and in a new experiment. Participants' choices and decision times were determined both by subjective rank-order distances and by the recognition status of the compared objects. To integrate these theoretically inconsistent findings, we discuss the role of the evidential difference (or the degree of conflict) between two objects.