Two halfs may be more than one whole: Category-split effects on frequency illusions.

Splitting an event category into smaller subcategories can increase the subjective frequency of events. Category-split effects are derived from an information-loss model that predicts regressive frequency estimates (rare events are overestimated and frequent events underestimated). The basic regression effect and the overestimation of very low frequencies resulting from category split should increase with information loss in memory. Experiments 1 and 2 supported these predictions for graphical stimuli. In Experiment 3, the split effect was used to eliminate group-related illusory correlations. Although stimulus split (different subtypes included in the list) and response split (separate judgments for each subtype) are essential, the illusion did not depend on attention to the split during encoding