Backward associative priming relies on an automatic semantic matching process

Abstract

Backward (BA) priming, which is the facilitated recognition of targets that have a backward association with the prime (e.g., baby-stork), is said to occur due to a semantic matching process that is only engaged in the LDT at long SOAs (Neely, Keefe, & Ross, 1989). However, BA priming occurs at short SOAs and in other tasks (Kahan, Neely, & Forsythe, 1999), suggesting that it may rely on an automatic process. A lexical decision task was administered in which a nonword relatedness proportion (NWRP) was created, such that 50% of nonwords were related to their primes (e.g., boy-girk), and participants were warned that checking for a relation will not be helpful for task performance. For unmasked (but not masked) primes, BA priming occurred even when conditions (i.e., high NWRP and warning) decreased the utility of a semantic matching strategy. This suggests that BA priming relies on an automatic semantic matching process that requires a conscious prime. Furthermore, analyses of integrative priming (e.g., log-house) suggest that INT priming relies on a hybrid prospective/retrospective process.