Doctor of Philosophy (PhD)
Faculty of Science
Priming is the benefit that an event receives when its processing has been preceded by the processing of a related or identical event. Context effects on priming are evident when priming changes as a function of some feature of experimental trials. The most commonly explored context effect is that of relatedness proportion (RP), where it has often been shown that the magnitude of priming (semantic or repetition) is directly related to the proportion of related trials: Increasing the related trials results in greater priming. Although previously thought to depend on strategic processing, recent evidence of context effects from designs using masked primes and short stimulus onset asynchronies (SOAs; less than 250 ms) refutes this strategy view because it should not be possible to enact strategies in such brief intervals. In addition, such findings provide evidence against the dominant view that masked priming with short SOAs results from automatic spreading activation because automatic spreading activation should not be influenced by contextual factors. The current set of experiments was designed to provide converging evidence for the notion that episodic accounts may best account for priming by exploring whether differences in another context variable—list composition—would alter semantic priming in a lexical decision task (LDT) for masked, short SOA (67 ms) primes. In 3 experiments, list composition was manipulated by presenting experimental trials that contained either 3 prime types (nominally related, semantically related, or unrelated) or 2 prime types (semantically related or unrelated) in either a between-subjects design (Experiments 1 and 2) or a within-subject design (Experiment 3). List composition effects were found: Response times (RTs) to semantically related primes were associated with a response cost in the 3-Prime-Type condition but were associated with a response benefit (facilitation) in the 2-Prime-Type condition. Episodic accounts can best account for these results, whereby to best facilitate target identification, the cognitive system is biased to detect primes containing features that are most transfer appropriate with respect to targets.
Major, Jennifer C., "List Composition Effects for Masked Semantic Primes: Evidence Inconsistent with Activation Accounts" (2008). Theses and Dissertations (Comprehensive). 1059.