Priming and search processes in semantic memory: a test of three models of spreading activation

How does the strength of an association determine the speed with which the association can be retrieved? Three alternative classes of spreading activation models were developed and tested. All three models hypothesize a two-stage retrieval process consisting of an initial process of activation of associates of a concept node, followed by a process of selection of activated associations for evaluation. One class of models hypothesizes that strength influences the duration of the activation process and thereby the duration of the retrieval process; a second class of models claims that the speed of activation of an association is independent of strength, but that strong associations are activated to a higher level than weak associations and so are selected faster for further processing; a third class proposes effects of strength on both the duration of the activation process and the level of activation of an association. Three priming experiments were conducted to discriminate the models. In all three experiments, the strength of the prime-probe relation and the interval between prime and probe presentation (SOA) were varied. RT to make a positive response in a categorization task was the dependent variable of interest in Experiment 1; Experiments 2 and 3 examined priming effects in a naming task. The effects of SOA and strength were independent in all three experiments. The findings were interpreted as supporting those models which claim that strength influences only the level of activation of an association.