Repetition priming of homographs and novel objects: Evidence for an item-specific locus

Two experiments examined the nature of the memory representations underlying repetition priming. In Expt 1, homographs (words with two meanings) were presented accompanied by bias words that demanded one particular interpretation of each. These homographs were then used in a test phase picture-word matching task. Repeated items accompanied by a picture chat demanded another interpretation of the word showed as much facilitation as repeated items accompanied by a picture that preserved the trained interpretation of the homograph. Under the same encoding conditions, participants were able to remember which bias words accompanied the homographs at training, yet there was no evidence of a role for the retrieval of interpretive encoding operations in producing the repetition priming observed; results were instead consistent with activation of item-specific representations underlying the effect. This locus of repetition priming was further investigated in Expt 2. Participants made same-different responses to pairs of simultaneously presented novel objects. There was no reduction in facilitation of responding following recombination of pairings of items between training and test, as compared to pairs repeated intact. These results are not compatible with accounts of the priming effect based only on the retrieval of prior processing episodes, or the reinstatement of prior processing demands. The results are consistent with a perceptual locus of the priming effect, based not on pre-existing representations or connections, but on representations of structure and form of individual stimuli.