Memory recruitment: A backward idea about masked priming.

Accounts of priming typically assume that primes activate existing representations, thus prospectively altering target processing. In contrast, Bodner and Masson (1997) described a memory-recruitment account of priming in which the processing operations applied to primes-even masked primes-are encoded into a new memory instance that can be retrospectively recruited to contribute to target processing if the context and task conditions support recruitment. Our chapter updates three lines of research we have pursued since our last update (Masson & Bodner, 2003) in attempting to contrast various priming accounts. The first line examines when and why nonwords, which do not have existing representations, show masked priming. The second line examines when and why masked priming is sensitive to prime-proportion manipulations. The third line examines whether masked priming effects can be lasting. After summarizing some of the successes and failures of our backward account, we acknowledge that we failed to provide unequivocal evidence for it. We end with some suggestions for moving accounts of masked priming forward.