Cognitive control and response modality in the Stroop task
The Stroop interference effect is one of the most robust findings in cognitive psychology and is widely used to study both the automaticity of reading and attentional control (how attention is used to maintain focus on a task goal in the face of distractions). The aim of the thesis is to provide insight into these processes by investigating which components of the reading process conflict to produce the interference and how attentional control can be strategically deployed to maintain focus on the task goal. To that end, we conducted five empirical studies that all featured a comparison of two types of Stroop tasks: an oral Stroop task where the task goal is to respond aloud to the display colour, and a manual Stroop task where the task goal is to categorise colours using manual button presses. We also went beyond the analysis of mean RT and accuracy data by using diffusion modelling (Ratcliff, 1978), which accounts for both the accuracy of responses and the whole distribution of associated RTs. In Chapter 1, Mills et al. (2019) investigated the negative priming effect and Kinoshita & Mills (2020) used a picture variant of the Stroop task where participants orally named, or manually identified via button presses, line drawings of animals. In Chapter 2 we investigated whether the semantic Stroop effect (Kinoshita et al., 2018) and the word interference effect (Mills et al., 2021a) components of the Stroop interference effect could be modulated by neutral proportion (i.e., the proportion of conflict free neutral hash symbol trials relative to word trials). In Chapter 3, Mills et al. (2021b), broke the Stroop interference effect down into its components (response set effect, semantic Stroop effect and the word interference effect). In the first study of Chapter 1 we challenged the commonly held view that negative priming is a ubiquitous effect by showing that the negative priming effect is present in the oral task but absent in the manual task at both the level of mean RT and the whole RT distribution. In the picture variant Stroop task used in study two the major finding was that pseudo-word distractors that shared the onset segment with the picture name facilitated picture naming relative to pseudo-word distractors without onset overlap, regardless of orthographic overlap, (e.g., CUST–camel = KUST–camel < NUST–camel); in contrast in the manual task there was no difference between the pseudoword conditions. In Chapter 2 which investigated the neutral proportion effect, we found that both the semantic Stroop effect and word interference effect components were of larger magnitude when neutral proportion was high than when it was low, particularly in the oral task. We discuss two alternative explanations for the neutral proportion effect: attentional control of conflict or a trial frequency effect. The major findings of Chapter 3, which looked at the components of the Stroop interference effect, is that at the level of mean RT, the overall Stroop interference effect was larger in the oral than the manual task, which was also reflected in all the component effects (semantic Stroop effect and word interference effect), except the response set effect. At the level of the diffusion model, the drift rate parameter was modulated by the Stroop interference effect, and the semantic Stroop effect component, particularly in the oral task. However, the attentional control interpretation of the neutral proportion effect was contradicted by the finding that neutral proportion did not modulate the drift rate. Overall, in all five studies we found strong evidence that the interference due to reading is larger in the oral task than in the manual task. We will use these results to make the case that conflict in the Stroop task is not an automatic or invariant process, but is instead driven by top down, strategic processes that are modulated by the task goal (colour naming in the oral task and colour categorisation via button presses in the manual task).