Recovery of unavailable perceptual input

Abstract Two experiments on the “recovery effect” are reported. In Exp. I, Ss' recall of a briefly flashed stimulus was tested before and after fantasy generation. Postfantasy recall was greater than prefantasy recall to an extent that significantly exceeded corresponding increases in recall obtained by control Ss. One group of control Ss did not generate fantasy, while the other produced fantasy but copied an experimental S's first recall attempt instead of seeing the stimulus. When the number of responses emitted in postfantasy recall was equalized for all groups, however, the difference in recall increments disappeared between the fantasy and nonfantasy group, suggesting that fantasy augments response rates rather than sensitivity to the stimulus trace. Exp. II confirmed this inference. A recognition indicator with confidence ratings was employed, from which ROC functions were extracted, allowing direct measures of pre- and postfantasy sensitivity. No sensitivity increments were found in either the fantasy or the nonfantasy group, though fantasy affected confidence ratings and, therefore, hit and false alarm rates.