The Role of Task Context for Component Processes in Focus Switching

Summary The present study was aimed at differentiating processes involved in realizing a shift of the focus of active processing within working memory. In two experiments we used an event counting task similar to Garavan (1998) in which subjects had to retain two or three numbers and to perform an updating operation on one of them while maintaining the other ones. Such updating had to be performed in sequences of 10 up to 14 updating steps. To estimate focus switching costs we compared steps in which the same number as in the preceding step is updated to steps in which updating must be executed on one of the other memory elements. Additionally we manipulated four variables: difficulty of the updating operation, number word length, informational demand set by the updating operation to be executed, and memory set size. Results show that (1) maintenance rehearsal operates independently from focus shifting processes, (2) a one-step task-independent switching mechanism is not sufficient to account for changes in focus switching time, (3) serial transitions of attention may occur prior to access by retrieval, and (4) retrieval demands can be considered a main factor contributing to the time needed for a focus switch. Key words: focus, switch, updating, working memory, set size, retrieval Introduction: Working memory and management of memory contents Every day human cognitive activities such as language comprehension and production, mental arithmetic or simple "thinking of something" often involve multiple processing steps and maintenance and use of several intermediate results. Imagine for instance reading or listening to a story. Usually the action is sometimes interrupted by passages that give extra background information about situations or acting persons. To follow the action of the story while reading or listening, encoded contents must be maintained somehow to have the possibility to "switch" back to the point of interruption when it goes on. As a consequence it comes with little surprise that performance in complex tasks can be predicted by working memory measures (i.e., measures of the amount of information that can be maintained while simultaneously other mental operations have to be executed) much better than by measures of simple memory span that only require serial recall (e.g. Daneman & Carpenter, 1980; Shah & Myiake, 1996). A commonly accepted definition of working memory is that it is the set of cognitive processes that permit information to be transferred into and temporarily held in an accessible state, in service of ongoing mental task. That means that working memory capacity in general describes the ability to manage or to optimize the use of memory contents needed in ongoing and changing mental tasks we are faced with in everyday life, such as following a story, getting the point of a joke, or preparing a diner. In several models it is assumed that there is a capacity limited storage mechanism, often called focus of attention (Cowan, 1995; Engle, Lughlin, Tuholski, & Conway, 1999) representing memory contents that are associated with high retrieval speeds (Cowan, 1995) making them directly accessible for processing operations while there may be other memory information that is in an activated state but not object of current processing. In Baddeley's theory (Baddeley, 1996) memory contents in the focus of attention are those that are selected for processing by the central executive. It follows that holding selected memory elements accessible (within the focus of attention) and switching this focus to elements that must be accessed next is an important executive function of working memory (Smith & Jonides, 1999; Baddeley, 1996; Monsell, 1996). Up to now, detailed analysis of such switching within working memory has proceeded especially in reference to switching of tasks or task sets, respectively (Monsell, 1996; Allport, Styles, & Hsieh, 1994; Kluwe, 1997; Mayr & Keele, 2000). …