Tests of strategy differences between the two subject-groups involved examination of differences in neural activity between memory-load conditions across individual subjects, in each task period. To examine the relationship between PFC activation and memory-load in each of the age-groups, we determined, for each subject, the extent of activation change in each task period, by determining the regional mean parameter estimate in each memory-load condition. Figure 17.2 shows the regional mean parameter estimates of younger and older subjects plotted against memory-load in each task period.

Dorsal PFC

Activation in each memory-load condition and task-period, for each group in dorsal PFC, (based on median parameter estimates) can be observed in Figure 17.2. First, we performed tests for monotonic changes in this region with increasing memory-load across all subjects. No significant changes were found in the Encoding period (p = 0.26) but significant increases in activation with increasing memory-load were found in the first Delay period (p = 0.03), in the second Delay period, (p = 0.003) and in the Retrieval period (p = 0.003). Next, we performed tests to assess whether the pattern of changes in PFC activation across different memory loads differed between age groups. Visual inspection of the slope functions plotted in Figure 17.2 indicated minimal age-related differences. Formal testing confirmed this observation; K–S tests were non-significant in the Encoding (p = 0.19), the first Delay period (p = 0.86); the second Delay period (p = 0.50), and in the Response period (p = 0.99).

Ventral PFC

Activation in each memory-load condition and task-period, for each group in ventral PFC, (based on median parameter estimates) can be observed in Figure 17.2. Tests of monotonic activation changes in this region with increasing memory-load across all subjects were significant in all task periods. In the Encoding period there were significant decreases in activation with increasing memory load (p = 0.007) whereas in the first Delay period (p = 0.05), second Delay period (p = 0.03) and Retrieval period (p = 0.003) there were significant increases. Visual inspection of the slope functions plotted in Figure 17.2 indicated minimal age-related differences. These observations were supported by K–S tests in the Encoding period (p = 0.27), the first Delay period (p = 0.94), second Delay period (p = 0.50), and in the Response period (p = 0.76).

Tests of age-related efficiency differences

Tests of efficiency differences between the two subject groups involved examination of differences in activation-performance relationships between individual subjects across memory-load conditions,(p.290)

(p.291)
in each task period. To characterize the performance of individual subjects, we computed a performance composite score for each subject that was comprised of the subject’s z-standardized RT subtracted from their standardized accuracy scores. Thus, high performance composite scores corresponded to high accuracy and low RT. Table 17.1 shows subjects’ performance scores by rank. To characterize each subject’s activation level we determined, for each subject, the extent of activation in each task period, independent of memory-load, by determining the regional mean parameter estimate averaged over memory load. We could infer age-related differences in neural efficiency to the extent that we could observe interactions between Age-group, Performance score on activation. Thus, to test for age-differences in performance activation relationships, we performed analyses of variance in each task period, with age-group, performance score and activation as between-subjects factors.

Dorsal prefrontal cortex

Table 17.2 shows the standardized regression coefficients that characterize the relationship between performance-composite scores and regional activation in dorsal PFC, and the results of the F-test for the interaction of Age-group and Performance-score on activation. Examination of Table 17.2indicates that the regression coefficients for younger subjects were negative, indicating decreases in activation with increasing performance-composite scores whereas regression coefficients for older subjects were positive, indicating increases in activation with increasing performance-composite scores for older adults. The observation of opposite linear trends between the younger and older subjects was confirmed, in the Encoding and Response periods, by significant interactions of Age-group and Performance score in these task periods (Encoding: F(1,10) = 10.88, p ‹0.008, MSe = 0.003; Response: F(1,10) = 5.30,p ‹0.04, MSe = 0.005). In general, more modest regression slopes were observed in the Delay periods. No other main effects or interactions were significant.

Ventral prefrontal cortex

Table 17.3 shows the standardized regression coefficients that characterize the relationship between performance-composite scores and regional activation in ventral PFC, and the results of the F-test for the interaction of Age-group and Performance score on activation. Examination of Table 17.3 indicates that the regression coefficients for younger subjects were negative, indicating decreases in activation with increasing performance-composite scores whereas regression (p.292)