Stimulus-stimulus and stimulus-response compatibility: An interactive activation model.

Abstract

Every day we receive constant stimulation through our senses and take quick and accurate action. Human behavior in these situations is greatly influenced by compatibility (match vs. non-match among stimuli and responses). In this dissertation, I investigate three types of compatibility effects: one introduced by similarity between relevant stimulus and response dimensions (relevant S-R overlap), another by similarity between irrelevant stimulus and response dimensions (irrelevant S-R overlap), and a third by similarity between two stimulus dimensions (S-S overlap). In experiments 1 and 2, I employed four-choice Stroop tasks where all three types of overlap were present and manipulated. In order to measure the principal effects of these different types of overlap, additional control tasks were used where only one type of overlap was present and manipulated in a given task. From these experiments, I found that all three types of overlap affect reaction time (RT) performance in the Stroop tasks. This finding refutes many theories that attribute the Stroop effect exclusively to either stimulus-processing (S-S overlap) or response production (S-R overlap). It is also found that the effects of various types of overlap are interactive rather than linearly additive. In experiment 3, the irrelevant stimulus preceded the relevant stimulus with a time interval (stimulus-onset-asynchrony, or SOA). I found different time courses for the S-S overlap and for the irrelevant S-R overlap. All these experimental findings are in qualitative agreement with the dimensional overlap model (Kornblum, 1992; Kornblum, Hasbroucq, & Osman, 1990). Based on the dimensional overlap model, I advanced a quantitative, interactive activation model, which assumes (1) a feed-forward, layered (input-intermediate-output) network with lateral inhibition, (2) a convergence of input layers onto an intermediate layer implementing the S-S overlap, and (3) the presence of the automatic and controlled processes in the determination of outputs implementing the S-R overlap. This model can account for the results from the preceding experiments as well as many major findings in the literature.