Pulled in the Wrong Direction! Rethinking Response Control in Computational Models of Conflict Tasks
In daily life, we constantly make decisions that require us to navigate multiple sources of information. For example, when deciding whether to push or pull to open a door, the design of the door handle may trigger the automatic response tendency to pull, while a “push” sign indicates the opposite response. This mismatch illustrates how various informational cues can signal different response options that compete for activation, thereby placing demands on our ability to control responses. Such situations have sparked a large body of research to study how well individuals can flexibly control their responses during cognitive conflict.
The central problem identified and addressed in this Master Thesis is that current conceptualizations of response control lack the specificity to provide a complete picture of this control ability. Traditionally, researchers have relied on summary statistics that only provide a coarse description of response control. The move toward computational modeling, with the Diffusion Model for Conflict tasks (DMC), allows response control to be studied in terms of latent processes that give rise to behavior. Despite these advances, I argue that a critical oversight remains: current model-based measures of response control neglect the full state of the decision-maker. Specifically, distractor strength, as quantified by the model’s “peak amplitude” parameter, is used as a proxy for response control. In other words, the notion of response control is often approached in a narrow and isolated manner. To address this oversight, the present thesis aims to refine the conceptualization of response control by taking the broader decision context into account. The first part therefore investigates how peak amplitude interacts with other DMC parameters, using a simulation-based approach. Based on these findings, alternative measures of response control are developed and systematically evaluated against comparative criteria. In the final part of the present work, the most robust measure is applied on empirical data in relation to the existence of a common response control mechanism underlying performance in different conflict tasks.
Through theoretical simulations, the results show that peak amplitude should not be considered in isolation to study response control. Rather, the effect of distractor strength (peak amplitude) on behavior depends on the amount of evidence one accumulates (decision boundary) and efficiency of evidence accumulation (drift rate). This has important implications: two individuals with the same peak amplitude may nevertheless differ greatly in how cognitive conflict impacts their decision formation. Therefore, I provide a novel conceptualization of response control that accounts for the dynamic interplay between automatic and controlled activation. Specifically, Triple R emerges as the most robust model-based measure of response control among the set of candidates outlined in this thesis. Furthermore, when applied to empirical data, the significant positive correlation of Triple R between different conflict tasks provides evidence for a common response control mechanism across contexts. This finding speaks directly to the long-standing debate on the domain-generality of response control and demonstrates the value of Triple R as a context-sensitive measure of response control.
Meer lezen