More modeling but still no stages: Reply to Borowsky and Besner.

Abstract

The study of semantic priming in lexical tasks has yielded rich and highly intricate patterns of results that have provided important insights into the nature of lexical and semantic processing (see McNamara, 2005; Neely, 1991, for reviews). As various theories have grappled with the complexities of these findings, though, they have been led to augment basic processing mechanisms (e.g., spreading activation) with additional, seemingly independent mechanisms (e.g., expectancy-list generation, retrospective matching, resource limitations), many of which seem directly tailored to the experimental priming manipulation itself. In this way, researchers have begun offering theories of priming rather than theories of lexical processing that give rise to priming. In Plaut and Booth (2000), we offered an account of lexical processing, grounded in general (connectionist) principles, that we claimed could account for certain aspects of the relevant empirical phenomena, including some findings thought to implicate additional mechanisms. We supported this account by a specific computational implementation that, although necessarily limited and intentionally simplified, instantiated the core aspects of the theory. To enable the quantitative adequacy of the implemented model to be evaluated in detail, we developed it in the context of three specific empirical studies: adults and children tested on primed lexical decision at a long (800 ms) stimulus onset asynchrony (SOA) and adults tested at a short (200 ms) SOA. Previous work had found greater semantic priming (i.e., faster reaction times [RTs] following related vs. unrelated primes) for low- versus high-frequency target words and inhibition (i.e., slower RTs following unrelated vs. neutral primes) at a long, but not short, SOA (see Neely, 1991). We examined the extent to which these effects depended on individual differences among participants in age or perceptual ability. In brief, we found that greater priming for low-frequency targets was exhibited only by participants with high perceptual ability and that adults but not children exhibited inhibition at the long SOA. We went on to show that our implemented model behaved similarly. Here, we present a brief review of the Plaut and Booth model before addressing recent challenges to it.