Frequency judgments: The problem of defining a perceptual event

In four experiments the conditions under which frequency judgments reflect he relative frequency of complex perceptual events were explored. Subjects viewed a series of 4 x 4 grids each containing seven items, which were letters and numbers in one of four typefaces. Later judgments of the relative frequency with which particular letters appeared in particular typefaces were unaffected by a warning about an upcoming frequency judgment task, but were affected by both the time available for processing the stimuli and the nature of the cover task subjects engaged in while viewing the grids. Frequency judgments were poor when exposure durations were less than 2 s and when the cover task directed subjects ' attention merely to the locations of the items within the grids. Frequency judgments improved when the cover task directed subjects ' attention to the identity of the stimuli, especially to the conjunction of letter and typeface. The results suggest that frequency estimation of complex stimuli may be possible only for stimuli that have been processed as phenomenal objects. Largely prompted by Hasher and Zack's (1979, 1984) pro-posal that certain attributes of events are automatically en-coded, there has been considerable r cent interest in people's ability to judge the relative frequency of events (e.g., Barsalo