Is associative priming a valid method to differentiate the serial and parallel models of face identification?

Abstract

One current controversy in face identification is whether names are accessed after or in parallel to semantic information. In prior research, Schweinberger, Burton, and Kelly (2001) have shown that phonological decisions to the names of famous faces were facilitated by name but not by semantic primes, while semantic decisions were facilitated by semantic but not by name primes. They proposed a parallel rather than a serial model of face identification. We replicated these experiments by (1) adding specific semantic primes in order to examine the effect of the uniqueness of the relation between semantic or phonological information and a face, and (2) adding a neutral baseline to differentiate facilitation from inhibition effects when primes and targets were related or unrelated. Our study revealed that uniqueness plays an important role in associative priming, with the specific primes (whether lexical or semantic) producing the greatest effects. However, it was also revealed that the same primes generated different effects according to the difficulty of the task. Our results highlight the importance, in associative priming, of the specificity of the information and of the relation between the primes and the task, and raise the question of the validity of such a method to test cognitive models of face identification.