Carolyn Backer Cave

Many results implicate perceptual processing in repetition priming, but little is known of potential mechanisms for priming. A new method was used to help determine the processing stage at which priming occurs. Priming pictures were presented under dominance or suppression generated by binocular rivalry. Although low-level, sensory attributes can be processed under rivalry suppression, there is no evidence that repetition priming can be supported by such low-level processing. Priming was found only for stimuli that were processed sufficiently to be identified in the priming stage. The results demonstrate that repetition priming requires processing of stimulus attributes into relatively high-level representations.

A form of long-lasting repetition priming has been demonstrated by faster naming of pictures repeated from a prior exposure than new pictures This facilitation is independent of explicit memory in both normal subjects (Mitchell & Brown, 1988) and amnesic patients (Cave & Squire, 1992) Using picture naming, the current study demonstrated that priming could be detected with delays of between 6 and 48 weeks between the initial exposure and the priming test Recognition memory was also above chance at these delays, but performance on the two tests appeared independent These results show that a single stimulus exposure can have very long-lasting effects Accounts of repetition priming as a form of implicit memory will have to accommodate long-lasting changes in stimulus processing based on a single exposure.

The degree to which repetition priming is perceptually specific is informative about the mechanisms of implicit memory as well as of perceptual processing. In 2 sets of experiments with pictures as stimuli, we tested the effects of color and pattern manipulations between study and test on implicit memory (i.e., naming facilitation) and explicit memory (i.e., 2 forms of recognition). These manipulations did not affect priming. However, participants were able to explicitly detect stimulus changes at above-chance levels. Changes in color also produced small decrements in participants' ability to judge that repeated stimuli were old on a recognition test. Experiment 2 showed diminished priming with changes in the stimulus exemplar (i.e., a different picture of the same named object) from study to test, which demonstrated that the picture-naming paradigm is sensitive to changes in physical attributes. The results suggest that physical attributes that are not essential to the formation of a shape representation do not influence repetition priming in a basic identification paradigm. Suggestions for how priming may be mediated are discussed.