Processing Speed, Attentional Capacity, and Age-Related Memory Change

Abstract

This study compared the relative importance (i.e., proportion of shared variance) of attentional capacity and processing speed accounts of cognitive aging to predict age differences in episodic and working memory performance. Right-handed adults (n = 100), 18 to 88 years of age, completed measures of attentional capacity (divided attention), processing speed, and episodic and working memory. The results provide little support for the predictive utility of the attentional capacity construct, independent of processing speed ability in accounting for age-specific episodic memory relations. The results are, however, consistent with the notion that attentional capacity mediates aspects of age-related working memory change.

The authors would like to thank Jamie Campbell, Neil Charness, Laurie Sippola, Norma Stewart, and two anonymous reviewers for comments and helpful suggestions on earlier versions of the manuscript.

Notes

Note. Education refers to years of formal education completed. NART refers to the mean score (/50) obtained on the Nation Adult Reading Test. Health represents a self-assessment on a 5-point scale ranging from 1 = excellent to 5 = poor. Limitation of daily activities is on a 5-point scale. Cardio/blood pres. refers to summed score of surgical intervention for cardiovascular intervention and whether or not blood pressure medication is being taken (yes/no responses).

¹Letter fluency has been used by a number of researchers as an index of executive performance (e.g., Henry, Crawford, & Phillips, 2005). In order to test the degree to which our sample was at or above age norms for executive functioning, we compared the obtained letter fluency scores from our sample to a published normative data set of 1300 participants ranging from 16 to 95 years (Tombaugh, Kozak, & Rees, 1999). Tombaugh et al. asked participants to generate as many words as they could beginning with the letters F, A, and S. Number of words generated ranged from a high of 43 for the 30s and 40s age decades and declined to 29 for the 80s decade. This represents a 33% drop in performance from the 30s decade to the 80s decade. As noted in Methods, the task incorporated in the current study differed from that of Tombaugh et al. in that participants were given one letter at a time for 20s. An analysis of the averaged responses for the two control letter fluency conditions (see Appendices C and D) reveals that letter fluency scores drop from a high of 8.4 for the 30s decade to 6.6 for the 80s decade. This represents a 23% drop in performance. These data suggest that the older sample may be slightly superior on measures of executive functioning compared to published norms.

Note. Dual = concurrent task performance rate; residual = residual score from regression equation predicting dual-task performance from single-task performance.

∗Significant at the .01 level (two-tailed).

Note. Dual = concurrent task performance rate; residual = residual score from regression equation predicting dual-task performance from single-task performance.

Numbers in parentheses refer to percentage of shared age-related variance among the four simple composites within each calculation metric.

∗Significant at the .01 level (two-tailed).

Note. I = first recall; II = second recall; processing speed composite = average of Letter Comparison and Pattern Comparison; single-task composite = composite of performance on control task measures; dual-task composite = composite of proportional ([single − dual]/single) decrement scores; episodic memory = composite of I and II trials of Delayed Word Recall, Logical Memory, Difficult Paired Associates, and Rey Complex Figure Test; working memory composite = composite of Listening Span and Computation Span; NART = NationalAdult Reading Test.

^a Estimated test-retest reliability derived by using Pearson correlation.

^b Value from Snow, Tierney, Zorzitto, Fisher, and Reid (1989).

^c Estimated intercorrelation of scores using Cronbach's Alpha.

^d Value from Wechsler (1987).

^e Value from Salthouse and Babcock (1991).

^f Value from Crawford, Parker, Stewart, Besson, and De Lacey (1989).

∗Significant at the .01 level (two-tailed).— = no scores available.

Note. I = first recall; II = second recall; single task = composite of performance on control task measures; processing speed = average of Letter Comparison and Pattern Comparison; Dual task = composite of proportional ([single − dual]/single) decrement scores; episodic memory = compositeof I and II trials of Delayed Word Recall, Logical Memory, Difficult Paired Associates, and Rey Complex Figure Test; working memory = composite of Listening Span and Computation Span.

Note. Single task = composite of performance on control task measures; dual task = composite of proportional ([single − dual]/single) decrement scores; Speed = averageof Letter Comparison and Pattern Comparison; I = first recall; II = second recall; episodic memory = composite of I and II trials of Delayed Word Recall, Logical Memory, Difficult Paired Associates, and Rey Complex Figure Test; working memory = composite of Listening Span and Computation Span.

Note. Dual task = composite of proportional ([single − dual]/single) decrement scores; speed = average of Letter Comparison and Pattern Comparison; I = first recall; II = second recall; episodic memory = composite of I and II trials of Delayed Word Recall, Logical Memory, Difficult Paired Associates, and Rey Complex Figure Test; working memory = composite of Listening Span and Computation Span. M = mean value.

²We also examined the degree to which the criterion and mediator variables interacted with age. To do this, we created four interaction terms that represented the interaction between working memory and dual-task performance, working memory and processing speed, episodic memory and dual task performance, and episodic memory and processing speed and regressed these interaction terms on age. These additional analyses revealed that the interactions between the mediator and the criterion variables did not account for any additional variance over and above that of the criterion and mediators alone (largest t = 1.01, p = .315). These additional analyses demonstrate that the criterion and mediator relationship was homogenous as a function of age.

Note. Speed = average of Letter Comparison and Pattern Comparison; dual task = composite of proportional ([single − dual]/single) decrement scores.

∗Significant at the 0.01 level (two-tailed). Numbers in parentheses refer to the proportion of shared age-related variance among mediating and criterion variables.

Note. Control age R ² = .640 (p < .01).

Note. Control age R ² = .245 (p < .01).

Note. Control age R ² = .036 (p = .06).

Note. Control age R ² = .027 (p = .10).

Note. Control age R ² = .090 (p < .01).

Note. Control age R ² = .010 (p < .01).