The influence of complex working memory span task administration methods on prediction of higher level cognition and metacognitive control of response times

DAVID P. MCCABE 0 0 Colorado State University , Fort Collins, Colorado Participants between the ages of 18 and 80 were tested on a complex working memory span task that was administered either using a typical experimenter-paced method or using a method in which the processing component was presented at a fixed, limited-pace presentation rate. Path analyses revealed that even after controlling for individual differences in general processing speed, the limited-pace task predicted unique variance in episodic memory, executive functioning, and fluid intelligence, whereas the experimenter-paced task did not. For the experimenter-paced task, slower responses on the processing component of the task were associated with better recall, but only when individual differences in processing speed were controlled. These findings suggest that metacognitive control of response times affects recall from working memory span tasks, as well as the relationship between span task recall and high-level cognition. These results support resource-sharing explanations of working memory and suggest that limiting processing times using computer pacing of complex span tasks can be an effective way to efficiently measure working memory capacity. - The concept of working memory (WM) has become central to understanding complex cognition. Although researchers debate the specific details of models of WM, it can generally be defined as the cognitive system responsible for maintaining information or task goals in an active state over brief periods of time. Part of the reason for the success of WM as a model of short-term cognitive processing is that individual differences in the ability to maintain and manipulate information in WM, referred to as working memory capacity (WMC; Engle, Tuholski, Laughlin, & Conway, 1999), have been linked to various measures of higher level cognition, including reading comprehension (Daneman & Carpenter, 1980; TurleyAmes & Whitfield, 2003), episodic memory (McCabe & Smith, 2002; Oberauer, 2005; Park et al., 2002), executive function (Miyake, Friedman, Rettinger, Shah, & Hegarty, 2001), and general fluid intelligence (Engle et al., 1999; Kyllonen & Christal, 1990). Research on individual differences in WMC has primarily been based on examination of the relationship between performance on complex span tasks (e.g., reading span) and measures of higher level cognition. These complex span tasks typically require participants to encode and maintain several to-be-remembered items while completing some interpolated processing task, such as reading or arithmetic, that is intended to disrupt that maintenance. The participants are further required to recall the to-beremembered items in serial order. Complex span tasks can be contrasted with simple span tasks, which only require the maintenance and retrieval of to-be-remembered items in serial order. Recall is poorer for complex span tasks than for simple span tasks, and complex span tasks typically show stronger correlations with higher level cognition (e.g., Ackerman, Beier, & Boyle, 2005; but see Unsworth & Engle, 2007, for a detailed discussion of the similarities between simple and complex span tasks). Consequently, complex span tasks have become popular tools in the investigation of individual differences in many areas of psychology (see Engle & Kane, 2004, for a review). Some models of WM suggest that the reason that complex span tasks are more strongly related to complex cognition than are simple span tasks is that the former are more likely than the latter to engage the central executive component of WM (Baddeley, 2000; Engle et al., 1999). Indeed, individual differences in WMC have often been conceptualized as the efficiency of the central executive component of the WM system (Engle et al., 1999; McCabe, Roediger, McDaniel, Balota, & Hambrick, 2010), and this is the definition of WM capacity used in the present study. Of course, complex span tasks engage other abilities in addition to executive control processes. For example, Engles model of WMC (see Engle et al., 1999) indicates that task-specific strategies related to grouping or maintenance strategies, which are unrelated to attentional control (e.g., phonological rehearsal), are distinct from central executive functioning but may influence recall from span tasks. Methods of Administering Complex Span Tasks Complex span tasks are typically administered individually, and the experimenter influences the speed with which the processing component is completed (Conway et al., 2005). For example, in the reading span task, participants are typically asked to read sentences aloud (e.g., There are 7 days in every week), decide whether a sentence makes sense, and then remember the final word in the sentence (Daneman & Carpenter, 1980). Immediately after the participants have made their decision for a given sentence, the experimenter advances the screen to reveal the next sentence. Trials typically range in length between two and five sentences, and the participants are asked to complete the processing component of the task as quickly as possible while maintaining accuracy. Although the experimenter has some influence on the speed at which the processing component of the task is completed, by controlling when the screen advances, the participants are able to complete the processing tasks at their own pace. Thus, the participants have some control over the amount of time spent on the processing tasks. Previous research indicates that participants often take additional time to rehearse or refresh the to-be-remembered items during complex span task performance (Engle, Cantor, & Carullo, 1992; St. Clair-Thompson, 2007). Because the dependent measure for a span task is recall performance, considering the factors that influence recall performance is crucial to accurately estimating and understanding the WMC construct and its relation to higher level cognition. I will refer to the traditional administration method used for complex span tasks as experimenter-paced, because the experimenter controls when the screen advances to reveal the to-be-remembered item (cf. Friedman & Miyake, 2004; St. Clair-Thompson, 2007). This method can be contrasted with two other methods: a self-paced administration method, in which the participants completely control advancement of the screen following each processing item (e.g., by pressing the space bar when they have completed each processing item; Friedman & Miyake, 2004; Waters & Caplan, 1996), and a limited-pace administration method, in which the amount of time available for processing during the task is restricted in an effort to reduce the amount of processing and maintenance that the participants can engage in (Lpine, Barrouillet, & Camos, 2005; Oberauer, S, Schulze, Wilhelm, & Wittmann, 2000). Two primary issues were addressed in the present study, both associated with the administration method of complex WM span tasks. The first issue of int (...truncated)