Academic journal article Canadian Journal of Experimental Psychology

Increasing the Difficulty of Response Selection Does Not Increase the Switch Cost

Academic journal article Canadian Journal of Experimental Psychology

Increasing the Difficulty of Response Selection Does Not Increase the Switch Cost

Article excerpt

Several theories of task switching assume that basic task processes such as stimulus identification and response selection do not contribute to task-switch costs. This conclusion is mainly based on the finding that stimulus-identification manipulations have no influence on the size of the switch cost. The present study tested the influence of response-selection manipulations on the size of the switch cost. The authors manipulated the difficulty of response selection by using a semantically based response-side effect that is associated with numerical-judgement tasks, namely the Spatial Numerical Association of Response Codes (SNARC) effect. The authors observed a SNARC effect and a switch cost, but no interaction between the two: the task-switch cost did not differ between SNARC-compatible and -incompatible responses. The authors conclude that response selection does not contribute to the switch cost on the current trial, which provides further support for the idea that basic task processes and task-switch processes are separate.

Keywords: executive control, task switching, response selection

Many researchers assume that cognitive flexibility depends on executive processes that coordinate and schedule routine actions (e.g., Norman & Shallice, 1986). Routine actions involve simple cognitive abilities that are performed when triggered by the environment. While routine actions can account for basic task performance, they are insufficient when facing novel situations and situations in which more than one task can be performed. In such situations, higher-level control processes have to intervene to regulate behaviour in an adequate way (Baddeley, 1996).

The distinction between (more basic) task processes and control processes is prominent in task switching. When switching between tasks, task switches are typically slower and less accurate than performance on task repetitions (i.e., the switch cost). The common idea is that when subjects switch from one task to another, additional control processes are required to configure and protect the cognitive system such that the new task can be performed (e.g., Altmann & Gray, 2008; Logan & Gordon, 2001; Mayr & Kliegl, 2000, 2003; Rogers & Monsell, 1995; Rubinstein, Meyer, & Evans, 2001; Sohn & Anderson, 2001). For Instance, Rubinstein et al. (2001) proposed that basic task processes comprise information-processing stages such as stimulus identification, response selection, and response execution. On task repetitions, basic task processes are sufficient for successful behaviour. However, on task switches, additional processes are required: When switching tasks, a goal shift is made prior to stimulus identification and task rules are activated prior to response selection (Rubinstein et al., 2001). These additional control processes support task switching, but are time consuming and entail switch costs. In a similar vein, Mayr and Kliegl (2000, 2003) proposed that only one set of task rules can be maintained in working memory. On task repetitions, the same set of task rules is reused, and no additional control processes are needed. By contrast, on task switches additional control is required to retrieve the now-relevant task rules from long-term memory (LTM).

The aforementioned accounts thus assume that the switch cost reflects the duration of additional control involved in task switching and is not influenced by basic task processes. Consistent with this idea, manipulations of switch-related control processes such as goal shifting, rule activation, and rule retrieval influence the switch cost, whereas manipulations of stimulus identification do not influence the switch cost. Rubinstein et al. (2001) observed larger switch costs for tasks requiring complex categorization rules than for tasks requiring simple categorization rules. Similarly, Mayr and Kliegl (2000) observed larger switch costs for tasks with higher LTM-retrieval demands. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.