NANC 14(6).book(NANC_A_198667.fm)

This study was conducted to investigate the relative roles of working memory updating (updating) and processing speed in mediating age-related differences in fluid intelligence. A sample of 142 normal adults between 18 and 85 years of age performed a set of updating, processing speed, and fluid intelligence tasks. Hierarchical regression analyses indicated that the proportion of unique age-related variance in updating measures was related to the complexity of speed measures. There was a larger proportion of unique age-related variance in updating measures after controlling for the variance in simpler speed measures. Moreover, structural equation modeling showed that updating mediated almost all the age-related effects on fluid intelligence. These results suggest that updating, but not speed, is the critical mediator between age and fluid intelligence. In addition, the speed mediation of age-related differences in fluid intelligence as indicated by previous studies is at least partially derived from the executive component of speed measures. For a long time, processing speed theory (see Salthouse, 1996, for a review) has been one of the dominant theories in the field of cognitive aging research. This theory emphasizes the fact that the processing speed is generally slower in older adults than in younger adults, and postulates that this reduction in speed leads to impairments in higher-order or fluid cognition. However, in recent years, the attention of cognitive aging research appears to be drawn towards the concept of executive function, which is broadly defined as control processes responsible for the monitoring, controlling, and integration of other cognitive operations. Note that executive functions are Address correspondence to: Deming Li, Institute of Psychology, Chinese Academy of Sciences, Beijing, China. E-mail: [email protected] 632 TIANYONG CHEN AND DEMING LI often associated with the frontal lobes, and there is sufficient neuroanatomical evidence indicating that frontal lobes are more vulnerable to age-related deterioration compared to other brain areas (Raz, 2000, 2004), some researchers proposed frontal (or executive) decline hypothesis of cognitive aging, which predicts that increasing age is associated with a greater decline in measures of executive functions than with declines in general cognitive abilities. Furthermore, declines in executive functions may account for agerelated differences in higher-order performance (Rabbitt et al., 2001; West, 1996; Woodruff-Pak, 1997). Working memory updating (updating) is one of the most frequently postulated executive functions in recent literature (e.g., Collette & Van der Linden, 2002; Friedman et al., 2006; Miyake et al., 2000; Salthouse et al., 2003). This updating function involves constantly monitoring incoming information relevance for the task at hand, and then appropriately modifying the content of working memory by replacing old, no longer relevant information with newer, more relevant information (Morris & Jones, 1990). Previous research has classically explored the updating process with running memory task, which requires participants to watch strings of items (letters or digits) of unknown length, and then to recall a specific number of recent items in sequence. Morris and Jones (1990) have shown that updating is not the passive maintenance of memory loads, but the dynamic aspects of realtime processing in working memory. That is to say, updating belongs to the executive component of working memory. Previous studies have suggested that working memory, especially its executive component, has a strong relationship to fluid intelligence (Conway et al., 2002; Engle et al., 1999; Kane et al., 2005; Kyllonen & Christal, 1990). Furthermore, a recent paper directly examined the relations between several executive functions (including updating) and intelligence, and found only updating was highly correlated with intelligence measures (Friedman et al., 2006). Engle et al. (1999) and Friedman et al. (2006) suggested that the link between working memory (or updating) and fluid intelligence is the demand for controlled attention to maintain goal-relevant information in the face of concurrent processing and distraction. In the aging research field, it has been found that updating operations are more vulnerable to age than storage capacity when performing a running memory task (Van der Linden et al., 1994). Altogether, because working memory (especially its executive component, updating) has a verified strong relationship to both age and fluid intelligence, it is reasonable to assume that age-related decline on fluid intelligence may be primarily due to updating impairment, and this assumption is indubitably consistent with the executive decline hypothesis of cognitive aging. However, it is regrettable that, as a good indicator of executive component of working memory, updating has been investigated relatively rarely in relation to fluid intelligence and in the field of cognitive aging research. An WORKING MEMORY UPDATING 633 exception is the study of Salthouse et al. (2003), which found various executive functions, especially updating, mediated almost all the age-related effects on fluid intelligence. However, in that study, they did not directly compare the relative roles of updating and processing speed in mediating age-related differences in fluid intelligence. In previous studies, Salthouse preferred to adopt general slowing to explain age-related decline in cognition, and assumed that processing speed is a primitive construct that influence the cognitive system without themselves being reducible to other psychological constructs (Salthouse, 1996; Verhaeghen & Salthouse, 1997). That is to say, when processing speed is considered, the mediation of agerelated effects on fluid abilities through executive construct are much smaller than those through speed construct (e.g., Salthouse & Miles, 2002), or most of the relations between executive construct and both age and other cognitive abilities are shared with other variables, especially speed variables (e.g., Salthouse et al., 1998). These studies, together with others (e.g., Bryan et al., 1999; Fisk & Warr, 1996), induced us to feel that there has been little evidence to support executive decline hypothesis in cognitive behavior studies, especially where processing speed is considered. While numerous empirical reports have been found to support the processing speed theory, it is important to note that some studies have suggested that the relations between processing speed and fluid intelligence, and the speed mediation of age-related variance in other cognitive abilities, were varied when choosing different speed measures. For instance, Conway et al. (2002) showed that when choosing speed tasks requiring minimal demands on working memory, they did not find a significant unique relation between processing speed and fluid intelligence. Verhaeghen (1999) found that agerelated difference of recognition performance could be well accounted for by perceptual speed, but to a lesser extent by inspection time. Different types of measures have been used to assess the processing speed of an individual (see Salthouse, 2000, for a review). For instance, reaction time (e.g., choice reaction time) is one of the most frequently used speed measures, which is generally assumed to require four processing stages (i.e., stimulus encoding, stimulus identification, response selection, and motor execution). Psychophysical speed (e.g., inspection time) is supposed as an even simpler or purer measure of central processing speed than reaction time (Nettelbeck & Wilson, 1985), which normally involves only two processing stages (i.e., stimulus encoding and stimulus identification). Compared with reaction time and psychophysical speed, perceptual/motor speed measures (usually on paper-and-pencil tests, e.g., digit–symbol substitution test) are more complex, which generally involve some other cognitive processes. For instance, a digit–symbol substitution test typically involves visual searching, attention switching, and working memory resources (i.e., the more digit–symbol pairs were maintained internally, the more items 634 TIANYONG CHEN AND DEMING LI were completed in a given time). Therefore, we assume that some widely used speed measures (i.e., perceptual/motor speed) in cognitive aging studies are not pure measures of processing speed, and involve some executive operations. The purpose of this study was to investigate the relative roles of updating and processing speed in mediating age-related differences in fluid intelligence, and to find empirical evidence for the executive decline hypothesis. Different from previous studies, which usually take processing speed as cognitive primitives, this study will adopt three types of speed measures (i.e., inspection time, reaction time, and perceptual/motor speed), which is assumed to involve different level of working memory resources.