Aging, Neuropsychology, and Cognition 1998, Vol. 5, No. 2, pp. 129-146 1382-5585/98/0502-129$12.00 © Swets & Zeitlinger Differential Decline of Verbal and Visuospatial Processing Speed Across the Adult Life Span* Bonnie Lawrence, Joel Myerson, and Sandra Hale Washington University Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 ABSTRACT The present study compared the age-related decline in verbal and visuospatial processing speed in 131 participants aged 18 to 90 years. Participants performed four verbal and four visuospatial tasks. Age differences in processing speed were compared at the group and individual levels. For the group-level analyses, participants were divided into a young adult group and six older groups subdivided by decade. The mean verbal and visuospatial response times (RTs) for each group were regressed on the corresponding RTs of the young adult group. The slope of the visuospatial regression was greater than that for the verbal regression at all ages, and the difference between the visuospatial and verbal slopes increased with each decade. For the individual-level analyses, a verbal and a visuospatial processing-time coefficient (Hale & Jansen, 1994) was obtained for each individual, and these values were then regressed on age. Verbal processing time increased linearly by approximately 50% while visuospatial processing time increased exponentially by approximately 500% from 18 to 90 years. Taken together, the results at both the group and the individual level demonstrate that aging affects visuospatial processing to a much greater extent than verbal processing. The differential decline of scores on tests involving verbal and visuospatial information processing is well documented. Most notably, the age-related differential decline on the verbal and visuospatial subtests of the Wechsler Adult Intelligence Scale – Revised (WAIS-R) appears to begin as early as age 40, with visuospatial measures (e.g., block design) declining to a much greater extent than verbal measures (e.g., vocabulary) thereafter (Wechsler, 1981). However, the proper interpretation of this pattern of decline has long been a matter of dispute. This is because a direct comparison may be confounded by the speeded nature and greater novelty of the more visuospatial subtests and the nonspeeded nature and greater reliance on previously learned information of the more verbal subtests. More recently, several studies have compared the effects of age on verbal and visuospatial pro* cessing using experimental designs that do not confound the nature of the task with the type of information to be processed (Hale & Myerson, 1996; Lima, Hale, & Myerson, 1991; Myerson, Hale, Rhee, & Jenkins,1998; Tubi & Calev, 1989). The results of these studies are consistent with an interpretation that emphasizes the importance of the type of information (i.e., verbal versus visuospatial) as a determinant of the magnitude of the age difference. However, these studies typically have compared verbal and visuospatial processing in one younger and one older group (e.g., 20-year-olds and 70-yearolds) and have not examined the development of such age differences either longitudinally or cross-sectionally. The goal of the present study was to examine the development of differences between verbal and visuospatial informationprocessing speed across the adult life span using a cross-sectional design. Address correspondence to: Bonnie Lawrence, Department of Psychology, Washington University, Campus Box 1125, St. Louis, MO 63130, USA. E-mail: [email protected]. Accepted for publication: May 20, 1998. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 130 BONNIE LAWRENCE ET AL. Reports of age-related declines in processing speed are ubiquitous in the literature (Birren, 1965; Cerella, 1990; Cerella, Poon, & Williams, 1980; Salthouse, 1985). Evidence suggesting that aging is accompanied by some form of general slowing was provided originally by analyses in which the mean response times (RTs) of older adults were plotted as a function of the mean RTs of young adults on the same tasks. This method of inspection, known as the Brinley plot (Brinley, 1965), typically yields slope coefficients greater than 1.0 with single linear functions accounting for over 90% of the variance in older adults’ RTs. The slope coefficients have been assumed to measure task-independent aspects of cognitive speed (Cerella, 1985; Hale & Jansen, 1994; Lima et al., 1991) and the large proportion of variance accounted for is often interpreted as consistent with the general slowing hypothesis. According to this hypothesis, age-related slowing is not localized within any specific cognitive process but affects all processes to approximately the same degree. Rather than being completely global, Hale and Myerson and their colleagues (Hale & Myerson, 1996; Lima et al., 1991) have suggested that slowing may be general within particular cognitive domains (e.g., verbal and visuospatial). Based on analyses of both experimental and meta-analytic data, Hale and colleagues concluded that the relationships between young and older adults’ RTs on lexical (i.e., verbal) and nonlexical (i.e., visuospatial) information-processing tasks were more accurately described by separate, domain-specific regression lines than by a single, general regression line. More specifically, according to the Lima et al. meta-analysis, the slope of the function describing the verbal data was approximately 1.5, whereas the slope of the function describing the visuospatial data was approximately 2.0. Importantly, slowing within each domain appears to be general as the RTs from each domain are well fit by a single regression line. Slope coefficients for verbal and visuospatial RTs similar to those reported by Hale and colleagues have been reported elsewhere in the literature (e.g., Madden, 1992; Sliwinski & Hall, 1998), and evidence for additional domains involving short-term or working memory processes has also been described (Mayr & Kliegl, 1993; Sliwinski & Hall, 1998; Swearer & Kane, 1996). Some previous cross-sectional research suggests that aging is accompanied by linear increases in verbal RTs (Madden, 1992), whereas other research suggests that both verbal and visuospatial RTs increase in an exponential manner (Cerella & Hale, 1994). Cerella and Hale recently proposed a mathematical model of changes in the speed of information processing across the life span. According to their model, a negative exponential function describes processing speed changes throughout childhood and a positive exponential function describes processing speed changes throughout adulthood. Combining the two exponentials into one U-shaped function, Cerella and Hale successfully characterized RT data taken from various studies that examined changes in processing speed across the life span. There are limits to the conclusions that may be drawn from directly comparing the results of the Madden (1992) and Cerella and Hale (1994) studies with respect to the effects of aging on verbal and visuospatial processing speed. One potential problem is that although Madden used lexical decision tasks, a subset of the conditions used visually degraded stimuli and thus may have required visuospatial as well as verbal processing. Another potential problem is that although Cerella and Hale analyzed data from both lexical and nonlexical tasks, they did not test to see if the rates of change differed for the two types of tasks. Moreover, an attempt to do so based on published data would be complicated by the fact that these data sets are from different samples, making it difficult to determine whether the results reflect differences in the effects of aging on verbal and visuospatial processing or sampling differences between the studies. Therefore, the present study tested the same participants on both verbal and visuospatial tasks in order to compare the form (i.e., linear or exponential) and the rate of the change in verbal and visuospatial information processing across the adult life span. 131 VERBAL AND VISUOSPATIAL PROCESSING SPEED Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 METHOD Participants The young adult control participants were 24 undergraduate students attending Washington University (M age = 19.7 years, SD = 0.95) who participated in the Hale and Myerson (1996) study. Ninety-three adult participants were recruited from a pool of volunteers maintained by the Aging and Development Program at Washington University. Seven of the 93 adults (M age = 66.6 years) were excluded from analyses because of high error rates (see Results) and an additional 3 participants were excluded because of their health profiles (e.g., poor vision, childhood diabetes; M age = 66.7). The data from 24 older adults (M age = 69.7 years, SD = 3.5) who participated in the Hale and Myerson (1996) study were combined with the adult data collected for the current study, thereby enabling us to obtain a large sample of older adults (n = 107). A breakdown of participant characteristics is provided for each decade in Table 1. All participants were screened for vision problems and completed a health questionnaire. Overall, participants reported being in good to excellent health. Verbal ability was assessed using the Shipley Vocabulary test from the Shipley Institute of Living Scale. Raw scores on the Shipley Vocabulary were regressed on age, revealing no significant age-related trend in performance, t(128) = 1.53, ns. Materials Stimuli were presented on a NEC MultiSynch 2A monitor controlled by an IBM compatible 286 computer. Stimulus presentation was synchronized with the refresh cycle for the monitor display, and RTs were recorded with 1 ms accuracy. Participants used a response panel with three buttons arranged in an inverted triangle to indicate their decisions (the upper two buttons) and to initiate trials (the lower button). Once a trial was initiated, the stimulus remained on the screen until a response occurred. Design and Procedure Four verbal tasks (single lexical decision, double lexical decision, category judgment, and synonymantonym judgment) and four visuospatial tasks (line-length discrimination, shape classification, visual search, and abstract matching) were completed by all participants. These tasks were the same as those completed by young adult and older adult participants in the Hale and Myerson (1996) study. For the purpose of completeness, an abbreviated description of the details of each task is provided below. Additional details are available in Hale and Myerson. Each participant received the same order of task administration so that correlational analyses and comparisons of subgroups of different ability levels would not be confounded by order differences. The order of task administration (line-length discrimination, single lexical decision, visual search, category judgment, shape classification, synonymantonym judgment, abstract matching, and double lexical decision) distributed tasks from the two domains evenly across the sessions. Additionally, the order of administration of the tasks neither proceeded from easiest to most difficult nor vice versa, so as not to confound either practice or fatigue with task complexity. Each task began with six practice trials, followed by the experimental trials in which stimuli from all of the conditions within a task were interleaved. The first two experimental trials were ‘‘buffer’’ trials (i.e., the RTs on these trials were not included in the analyses). Not including the buffer trials, there were 16 trials per condition, with the exception of the different-shapes condition of the shape classification task which consisted of 32 trials (16 same-size, different-shape trials and 16 different-size, different-shape trials). The dominant hand was used to signal the following responses: word (vs. nonword), two words (vs. one word + one nonword), same category (vs. different categories), synonym (vs. antonym), same shapes (vs. different), target present (vs. absent). Table 1. Participant Characteristics by Decade. Decade n Sex M age and SD 18–21 30–39 40–49 50–59 60–69 70–79 80–89 24 18 18 18 18 22 13 14 males 19 males 10 males 17 males 18 males 11 males 16 males 19.71 (0.95) 34.83 (2.43) 45.22 (2.82) 53.44 (2.41) 65.94 (2.73) 73.86 (2.90) 85.15 (2.64) 132 BONNIE LAWRENCE ET AL. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 For the line-length discrimination task, participants pressed the key corresponding to the location of the longer line. For the abstract matching task, participants pressed the key corresponding to the location of the best match – upper left or upper right. Verbal battery In the single lexical decision task, participants decided whether or not a consonant-vowel-consonant letter string of high frequency (e.g., top), low frequency (e.g., rib), or a pronounceable nonword (e.g., ged), is a word in English. In the double lexical decision task, participants decided whether or not two strings of letters (presented one above the other) were both words in English. This task consisted of four conditions: related word pairs (e.g., high and low), unrelated word pairs (e.g., bread and queen), nonword-word pairs (e.g., jarton and ink), and word-nonword pairs (e.g., flesh and biper). In the category judgment task, participants decided whether or not two words were from the same semantic category. This task consisted of four conditions: high-typicality word pairs from the same category (e.g., ruby and emerald), lowtypicality word pairs from the same category (e.g., camel and buffalo), high-typicality word pairs from different categories (e.g., dog and carrot), and low-typicality word pairs from different categories (e.g., turnip and amethyst). In the antonymsynonym judgment task, participants decided whether the meanings of two related words were the same or opposite. The four conditions for this task were as follows: synonym pairs made up of high frequency words (e.g., loud and noisy), synonym pairs made up of lower frequency words (e.g., grieve and mourn), antonym pairs made up of high frequency words (e.g., hard and soft), and antonym pairs made up of lower frequency words (e.g., domestic and foreign). Visuospatial battery In the line-length discrimination task, participants decided which of two lines was longer. This task included two conditions: 10% and 20% difference in line length. In the shape classification task, participants decided whether or not two objects were the same shape, regardless of any difference in size. This task included three conditions: same shape and same size, same shape and different sizes, and different shapes. In the visual search task, participants decided whether a red square target was present in an array of red circles and green squares. This task included six conditions: target present and target absent conditions with three set sizes of 9, 15, and 25 items. In the abstract matching task, participants selected one of two arrays that was the best match to a third array which was positioned below the two upper arrays. Each array varied on four different dimensions (number, shape, and color of the elements, and orientation of the array) that could take on one of three values. Specifically, each array consisted of either two, three, or four shapes (circles, triangles, or squares) that were colored either red, yellow, or blue, and arranged either horizontally, vertically, or diagonally. On each trial, one and only one of the four dimensions was relevant to determining the best match. This task included four conditions: all three irrelevant dimensions held constant, two irrelevant dimensions held constant, one irrelevant dimension held constant, and no irrelevant dimensions held constant. RESULTS The visuospatial tasks used in the present study spanned a greater range of difficulty (as indexed by mean RTs for the young adult group) than the verbal tasks. For the young adult group, verbal RTs ranged from 0.639 s on the easiest condition of the single lexical decision task to 1.279 s on the most difficult condition of the synonymantonym judgment task (Table 2), whereas visuospatial RTs ranged from 0.473 s on the easier condition of the line-length discrimination task to 1.693 s on the most difficult condition of the abstract matching task (Table 3). In order to compare verbal and visuospatial processing speed on tasks of equivalent difficulty (Fisk & Fisher, 1994; Hale & Myerson, 1996), the following tasks and conditions were removed from the visuospatial analyses: line-length discrimination (all conditions), shape classification (2 easiest conditions), and abstract matching (2 hardest conditions). As this left a much smaller number of visuospatial than verbal conditions, nonword conditions also were removed from single and double lexical decision tasks to make the number of visuospatial and verbal conditions 133 VERBAL AND VISUOSPATIAL PROCESSING SPEED Table 2. Verbal Response Times (RTs), SDs, and Error Rates Across the Adult Life Span. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 Age group Lexical decision High frequency word RT SD % Error Low frequency word RT SD % Error Nonword* RT SD % Error Double lexical decision Related words RT SD % Error Word-nonword* RT SD % Error Unrelated words RT SD % Error Nonword-word* RT SD % Error Category membership High typicality – same RT SD % Error High typicality – different RT SD % Error Low typicality – same RT SD % Error Low typicality – different RT SD % Error Young 30s 40s 50s 60s 70s 80s 0.639 0.092 3.6 0.691 0.126 1.0 0.720 0.104 1.7 0.751 0.127 0.3 0.840 0.181 0.3 0.883 0.200 0.3 1.004 0.205 1.4 0.719 0.120 12.8 0.822 0.191 6.3 0.870 0.115 7.6 0.910 0.204 7.3 1.032 0.304 4.9 1.098 0.343 8.0 1.210 0.389 8.2 0.774 0.151 2.7 0.900 0.206 4.5 0.949 0.164 3.5 0.983 0.213 1.9 1.198 0.277 6.1 1.342 0.355 3.7 1.221 0.232 7.0 0.771 0.129 1.3 0.898 0.223 2.1 0.825 0.111 0.7 0.928 0.174 0.3 0.943 0.168 0.3 1.020 0.197 0.3 1.129 0.211 1.0 1.078 0.184 7.0 1.237 0.257 5.6 1.150 0.136 4.9 1.220 0.267 4.9 1.230 0.203 3.8 1.448 0.481 3.7 1.455 0.307 1.9 0.892 0.147 1.8 1.031 0.244 1.0 0.982 0.108 1.0 1.069 0.210 0.3 1.059 0.180 0.0 1.243 0.285 0.6 1.306 0.271 1.4 0.897 0.182 3.1 1.061 0.251 0.3 0.950 0.173 3.1 1.053 0.286 3.1 1.038 0.225 1.7 1.234 0.463 4.0 1.281 0.244 3.8 0.949 0.179 5.7 1.134 0.266 2.8 1.087 0.143 4.2 1.203 0.249 4.2 1.357 0.360 5.2 1.456 0.297 2.0 1.677 0.378 7.7 1.016 0.184 1.3 1.249 0.272 0.7 1.225 0.150 0.7 1.255 0.226 0.7 1.290 0.217 0.7 1.475 0.353 1.1 1.608 0.323 1.0 1.122 0.230 4.4 1.409 0.504 2.4 1.272 0.205 3.8 1.358 0.309 2.4 1.442 0.339 3.5 1.610 0.396 3.4 1.750 0.451 7.7 1.235 0.281 3.1 1.520 0.393 1.7 1.434 0.180 2.4 1.481 0.302 1.4 1.520 0.300 0.7 1.704 0.465 0.6 1.834 0.447 1.0 Table continues. 134 BONNIE LAWRENCE ET AL. Table 2. (continued). Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 Age group Synonym-antonym High-frequency synonyms RT SD % Error High-frequency antonyms RT SD % Error Low-frequency synonyms RT SD % Error Low-frequency antonyms RT SD % Error Young 30s 40s 50s 60s 70s 80s 1.076 0.207 5.2 1.270 0.308 2.4 1.275 0.214 4.2 1.258 0.253 3.8 1.380 0.266 4.2 1.603 0.425 4.3 1.773 0.347 6.3 1.057 0.198 4.2 1.308 0.291 1.4 1.288 0.245 2.4 1.311 0.350 2.1 1.455 0.294 1.7 1.540 0.394 2.3 1.667 0.390 3.4 1.237 0.261 8.3 1.551 0.500 5.9 1.507 0.340 9.4 1.503 0.387 3.1 1.537 0.393 2.4 1.767 0.471 7.4 1.824 0.405 5.3 1.279 0.269 4.4 1.501 0.307 3.1 1.485 0.273 5.2 1.570 0.384 1.7 1.622 0.345 2.8 1.823 0.580 1.4 1.925 0.475 1.4 Note. Tasks and conditions that have been omitted from the regression analyses are marked with an asterisk. more comparable.1 Thus, the data included in the regression analyses consisted of 12 verbal conditions (RTs ranging from 0.639 to 1.279 in the young adult group) and 9 visuospatial conditions (RTs ranging from 0.621 to 1.298 in the young adult group). To assess whether there were age differences in speed-accuracy trade-off in either domain, mean verbal and visuospatial error rates were calculated for each participant by collapsing across conditions within a domain. When these error rates were regressed on age, only the error rate for the visuospatial tasks increased significantly, t(129) = 1.988, p < .05; the error rate for the verbal tasks actually declined, t(129) = 1 The nonword conditions were selected for removal because it has been suggested that RTs in such conditions reflect qualitatively different processes from those in word conditions (e.g., Coltheart, Davelaar, Jonasson, & Besner, 1977; Madden, Pierce, & Allen, 1993). That is, nonword RTs may represent the time at which participants give up waiting to retrieve a word (i.e., a strategic decision) as opposed to the time required for a specific cognitive process (i.e., lexical access) to occur. –2.725, p < .01. However, in both cases the changes were extremely small: 0.16% per decade for the visuospatial and – 0.23% per decade for the verbal. The increase in visuospatial errors was entirely attributable to lower accuracy on the part of participants in their 80s. With this group deleted from the analysis, the change in visuospatial errors decreased with age (– 0.03% per decade, p > .1). Verbal and visuospatial error rates for each condition are presented in Tables 2 and 3. To test for the presence of differential slowing at the group level, the participants were divided into a young adult group (ages 18–21) and six older groups, subdivided by decade (e.g., 30–39 years, 40–49 years, etc.). For each decade, the mean RT was calculated for each experimental condition. The verbal and visuospatial RTs for each decade were then regressed on the mean RTs of the young adult group from the corresponding tasks and conditions (Fig. 1). For all groups, linear functions fit to the verbal and visuospatial RTs separately provided excellent descriptions of the data; all r2s > .92. Tests for separate regressions revealed that the slopes 135 VERBAL AND VISUOSPATIAL PROCESSING SPEED Table 3. Visuospatial Response Times (RTs), SDs, and Error Rates Across the Adult Life Span. Age group Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 Line-length discrimination* 10% difference* RT SD % Error 20% difference* RT SD % Error Shape classification Same shape – same size* RT SD % Error Same shape – different size* RT SD % Error Different shapes RT SD % Error Visual search 9 items – target present RT SD % Error 9 items – target absent RT SD % Error 15 items – target present RT SD % Error 15 items – target absent RT SD % Error 25 items – target present RT SD % Error 25 items – target absent RT SD % Error Young 30s 40s 50s 60s 70s 80s 0.473 0.147 0.0 0.606 0.190 0.0 0.592 0.136 0.0 0.608 0.127 0.0 0.684 0.126 0.7 0.693 0.161 0.0 0.886 0.298 1.4 0.458 0.127 0.3 0.577 0.161 0.3 0.572 0.123 0.0 0.590 0.113 1.0 0.676 0.115 0.7 0.725 0.266 0.0 0.863 0.301 1.4 0.549 0.112 1.6 0.644 0.137 0.3 0.717 0.151 1.4 0.759 0.129 0.7 0.864 0.143 0.3 0.965 0.294 1.1 1.033 0.236 1.0 0.583 0.108 3.6 0.715 0.186 1.7 0.810 0.137 3.8 0.862 0.166 2.8 0.912 0.161 0.3 0.996 0.263 2.6 1.070 0.209 3.8 0.621 0.142 0.5 0.777 0.247 1.0 0.823 0.132 0.7 0.849 0.174 1.4 0.881 0.129 0.7 1.020 0.262 0.9 1.072 0.236 0.0 0.638 0.120 4.4 0.732 0.132 1.4 0.771 0.123 3.1 0.802 0.146 2.4 0.893 0.193 1.7 0.887 0.166 2.3 1.138 0.250 5.3 0.698 0.148 1.0 0.879 0.188 1.0 0.945 0.156 1.0 0.975 0.140 1.0 1.116 0.247 0.7 1.190 0.244 0.9 1.422 0.261 0.5 0.671 0.136 2.9 0.809 0.150 1.7 0.890 0.139 3.8 0.874 0.173 3.5 0.992 0.196 3.5 1.072 0.212 2.0 1.297 0.291 4.3 0.865 0.209 2.3 1.160 0.263 1.0 1.226 0.221 0.7 1.251 0.220 1.4 1.391 0.377 0.3 1.607 0.420 0.0 1.845 0.400 1.9 0.750 0.171 3.9 0.960 0.204 2.4 1.016 0.168 5.9 0.986 0.171 6.6 1.158 0.300 7.6 1.351 0.368 8.2 1.469 0.257 14.4 1.019 0.230 0.8 1.434 0.401 0.7 1.617 0.345 0.3 1.537 0.356 1.0 1.698 0.531 0.3 2.037 0.576 0.6 2.352 0.543 3.4 Table continues. 136 BONNIE LAWRENCE ET AL. Table 3. (continued). Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 Age group Abstract matching Level 1 RT SD % Error Level 2 RT SD % Error Level 3* RT SD % Error Level 4* RT SD % Error Young 30s 40s 50s 60s 70s 80s 0.920 0.175 1.7 1.182 0.223 2.9 1.193 0.176 1.6 1.400 0.308 1.3 1.712 0.474 1.6 2.013 0.606 1.9 2.801 0.860 8.1 1.298 0.334 7.8 1.774 0.518 6.3 1.847 0.333 6.3 2.280 0.741 5.2 2.888 1.028 4.1 3.225 1.051 7.6 4.875 1.867 9.2 1.471 0.389 5.2 2.055 0.500 3.1 2.081 0.472 2.1 2.430 0.840 8.3 3.239 1.342 6.9 3.662 1.188 7.7 5.643 1.460 17.3 1.693 0.477 3.6 2.411 0.736 2.4 2.421 0.620 3.5 2.717 0.696 8.3 4.085 1.534 5.2 4.443 1.321 8.2 6.458 1.918 15.4 Note. Tasks and conditions that have been omitted from the regression analyses are marked with an asterisk. for the verbal and visuospatial regressions were significantly different within each decade; all ts(17) > 2.63, all ps <.05. It should be noted that although the visuospatial slope coefficients for each decade increased systematically with increasing age, the increase in verbal slope coefficients was less systematic. That is, although the two oldest groups (age 70 and above) showed the greatest slowing on verbal tasks, the 30–39 group was slower than either the 40–49, 50–59 or the 60–69 groups. Nevertheless, as may be seen in Figure 2, the difference between the visuospatial and verbal slope coefficients did increase systematically with each decade. To test for the presence of differential slowing of verbal and visuospatial processing at the individual level, the mean visuospatial and verbal RTs for each individual were regressed separately on the RTs of the young adult group from the corresponding tasks and conditions. The median percentage of variance explained by the regression of individual RTs on those of the young adult group was 91% in the visuospatial domain and 84% in the verbal domain (the semi- interquartile ranges were 5.8% and 7.0%, respectively). The resulting visuospatial and verbal slopes may be termed processing-time coefficients (Hale & Jansen, 1994) and provide a general processing speed index for each domain. Consistent with the tests for separate regressions on the group means, comparisons of individuals’ verbal and visuospatial processing-time coefficients revealed a systematic increase in the percentage of individuals with greater visuospatial than verbal slowing (see Fig. 3); 52% (22/42) of the participants from 19–39 years of age showed greater visuospatial than verbal slowing as compared with 89% (79/89) of the participants from 40–90 years of age. The age-related trend in processing-time coefficients was examined by plotting the verbal and visuospatial processing-time coefficients for each individual as a function of age. Polynomial regressions were conducted on the verbal and visuospatial coefficients separately to test for the presence of a nonlinear age-related trend. A significant quadratic trend was revealed for the visuospatial processing-time coefficients (incremental F = 19.12, p < .001). The quadratic trend 137 Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 VERBAL AND VISUOSPATIAL PROCESSING SPEED Fig. 1. The mean verbal (Vrb) and visuospatial (Vsp) response times (RTs) for each decade plotted as a function of the mean RTs of the young adult group from the corresponding tasks and conditions. Filled circles represent the visuospatial data and unfilled circles represent the verbal data. Solid lines represent the verbal and visuospatial regression lines. In each panel, if performance of the older and young adult groups were equal, the points would fall along the dashed line. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 138 Fig. 2. BONNIE LAWRENCE ET AL. The difference between visuospatial and verbal slope coefficients as a function of the mean age of each decade group. The solid line represents the best-fitting polynomial function. was not significant for the verbal processingtime coefficients (incremental F = < 1.0, ns), but there was a significant linear trend (incremental F = 6.51, p < .05). The quadratic equation that best fit the visuospatial data was y = 1.668 – 0.047 x + 0.001x2. The linear equation that best fit the verbal data was y = 0.992 + 0.005x. Although the quadratic trend was not significant for the verbal processing-time coefficients, both verbal and visuospatial processing coefficients were fit with an exponential function in order to compare the rates of age-related change as measured by the same mathematical model. The exponential model was slope = exp [d * (age – 20)] (1) in which d is the parameter governing the rate of increase in processing time. This equation is modified from Equation 2 of Cerella and Hale (1994) so that slowing begins at age 20, the mean age of the young adult group. (This modification was motivated by the following logic. Because an individual’s RTs are regressed on those of the 20-year-old group, the slope for an average 20-year-old is 1.0 by definition. In Equation 1, therefore, 20 is subtracted from a participant’s age so that, because exp (0) = 1.0, the predicted slope for an individual of age 20 is 1.0.) The rate parameter for the visuospatial function (d = 0.024) differed significantly from the rate parameter for the verbal function (d = 0.006), t(260) = 11.77, p < .0001. Based on these parameter estimates, the predicted increase in verbal processing time is nearly 50% from 20 Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 VERBAL AND VISUOSPATIAL PROCESSING SPEED Fig. 3. 139 The percentage of individuals with visuospatial slope coefficients greater than their verbal slope coefficients slowing for each age group. to 90 years. In contrast, the predicted increase in visuospatial processing time is more than 500% over the same period (Fig. 4). DISCUSSION In the present study, 131 adults from 18 to 90 years of age were tested on both verbal and visuospatial information-processing tasks, and age differences in processing speed were compared at both the group and individual levels. At both levels of analysis, the data within a specific domain (i.e., verbal or visuospatial) were well described by a single regression line. At the group level, for each decade in each domain, the regression of the older group’s mean RTs on those of the young adults accounted for at least 92% of the variance. At the individual level, the median percentage of variance explained by the regression of individual RTs on the mean RTs of the young adult group was 91% in the visuospatial domain and 84% in the verbal domain. Because the regression line typically accounted for most of the variance in RTs, the slope of this line was used as an index of processing speed in both the visuospatial and verbal domains and at both the group and individual levels (Hale & Jansen, 1994). Consistent results were obtained at both levels of analysis and revealed that although both visuospatial and verbal processing time increased with age, the increase was far greater in the visuospatial domain than in the verbal domain. At the group level, tests for separate regressions indicated that beginning with the group of participants in their 40s, and for each group thereafter, the slope coefficients for the Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 140 Fig. 4. BONNIE LAWRENCE ET AL. Individual verbal and visuospatial slope coefficients plotted as a function of age. The solid lines represent the best-fitting exponential functions. visuospatial regressions were significantly greater than those for the verbal regressions. At the individual level within each group, the percentage of participants whose visuospatial slopes were greater than their verbal slopes increased as a function of age, beginning at 50% for the young adult group and increasing to more than 98% for all participants 60 years of age and older. Although the visuospatial slope coefficients for each decade increased systematically with increasing age, the increase in verbal slope coefficients was less systematic. That is, although the two oldest groups (age 70 and above) showed the greatest slowing on verbal tasks, the next slowest group consisted of participants in their 30s. It is possible that this pattern in the verbal data reflects sampling differences between age groups, as those in their 30s tended to be nearly as slow as those in their 40s on the visuospatial tasks. However, because age differences in RTs on verbal tasks appear to be relatively small compared with individual differences, verbal speed measures may be particularly sensitive to sampling differences. One solution to the problem posed by such sampling differences is to examine how the difference between the visuospatial and verbal slope coefficients changes as a function of age, so that each group’s visuospatial speed is assessed relative to its own verbal speed. The results of such an analysis revealed a systematic, nonlinear in- Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 VERBAL AND VISUOSPATIAL PROCESSING SPEED crease in the difference between visuospatial and verbal slope coefficients across the life span. When individual visuospatial and verbal slope coefficients were regressed on age, a similar pattern was observed. The verbal slopes increased in an approximately linear fashion with age whereas the visuospatial slopes increased in a positively accelerated fashion, so that the difference between the two increased nonlinearly. In order to compare the age-related increases in verbal and visuospatial processing time, the individual verbal and visuospatial slope coefficients were fit with an exponential model (Cerella & Hale, 1994), and the results of nonlinear regression analyses revealed that visuospatial processing time increased at a faster rate than verbal processing time. More specifically, the exponential rate parameter for visuospatial processing time was 0.024 whereas the exponential rate parameter for verbal processing time was 0.006. Although the difference between the two values may seem small, the process of exponential increase is analogous to compound interest, and this apparently small rate difference results in large differences in the degree of age-related slowing by the latter portion of the life span. For the current data set, the predicted increase in visuospatial processing time over the period from 18 to 90 years of age is nearly 500% whereas the predicted increase in verbal processing time over the same period is not quite 50%. Because the relation between the visuospatial RTs of older and young adults is positively accelerated (e.g., Hale, Myerson, Faust, & Fristoe, 1995; Hale, Myerson, & Wagstaff, 1987), the RTs from the most difficult task (i.e., abstract matching) strongly influenced this estimate of the degree of visuospatial slowing. However, even when the abstract matching data were not included in the analysis, the oldest group (i.e., those in their 80s) was still more than 300% slower than the young adults in the visuospatial domain. The present results are consistent with the experimental and meta-analytic findings of Hale and colleagues suggesting that age-related declines in visuospatial processing speed are 141 greater than age-related declines in verbal processing speed (Hale & Myerson, 1996; Lima et al., 1991). Importantly, the present results extend these previous findings with older adults aged 65-75 years by demonstrating that the pattern of differential decline is evident perhaps as early as the fourth decade and that the difference continues to increase through the ninth decade. The present results are also consistent with previous studies suggesting that the decline in visuospatial processing speed follows an exponential time course (Cerella & Hale, 1994) whereas the decline in verbal processing speed is more nearly linear (Madden, 1992). However, these studies did not compare changes in verbal and visuospatial processing, and the present study is the first to examine the time course of age-related slowing of verbal and visuospatial processing in the same participants. Although the function describing the decline in visuospatial processing speed is clearly exponential in form (Cerella & Hale, 1994; the present study), the form of the function describing life span changes in verbal processing speed is not as easily characterized. Although a test for the presence of a quadratic trend in the verbal processing speed data from the present study was not significant, to conclude that the decline is linear would be to accept the null hypothesis. Such conclusions are always unwarranted and would be especially so in the present case for two reasons: first, because of the obvious variability in verbal processing speed coefficients; and second, because of the difficulty distinguishing linear and exponential trends when the rate of change is low, as it is in the verbal domain. For purposes of describing or predicting changes in verbal processing speed, the present data suggest that a linear model is as good as an exponential one. For purposes of comparing changes in the verbal and visuospatial domains, however, an exponential model may be preferable as it appears to be appropriate in both domains whereas a linear model is only appropriate in the verbal domain. Ultimately, determination of the mechanism(s) underlying the differential decline in verbal and visuospatial process- 142 BONNIE LAWRENCE ET AL. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 ing speed may be required in order to decide between alternative mathematical models of the time course of age-related slowing. Possible Mechanisms The results of the present experiment appear to parallel the well-documented differential decline in scores on the more verbal and more visuospatial subtests of the WAIS-R (Wechsler, 1981), and thus the possible mechanisms that have been proposed to explain age-related changes in such test scores should also be considered as potential explanations for the present findings. For example, it has been suggested that from the standpoint of cognitive aging, the Performance-Verbal distinction may actually be a speeded-nonspeeded distinction because many of the Performance measures have a speed component whereas the Verbal measures are primarily nonspeeded (e.g., Mittenberg, Seidenberg, O’Leary, & DiGiulio, 1989). However, the present results argue against such an interpretation because there was clear evidence of greater declines in the visuospatial domain despite the fact that both verbal and visuospatial processing were assessed using speeded tasks. It has also been suggested that the differential decline on the Verbal and Performance subscales of the WAIS-R may reflect age-related differences in the efficiency of processing familiar and unfamiliar information (Mittenberg et al., 1989). Evidence against this interpretation of differential decline is provided by Halpern (1984), who studied age differences in the speed of processing information regarding familiar traffic signs. Halpern found that whereas young adults took approximately the same amount of time to process familiar verbal and visuospatial information from such signs, older adults were much slower to process familiar information when it was nonverbal than when it was verbal. The results of the Halpern (1984) study suggest that it is the nature of the stimuli (i.e., verbal vs.visuospatial), and not whether people are processing familiar or unfamiliar information, that is responsible for the greater age differences observed on visuospatial tasks. Consistent with this interpretation, greater age-related differences have been reported on spatial working memory tasks than on verbal working memory tasks (Myerson et al., 1998; Tubi & Calev, 1989; Wechsler, 1997). Because working memory tasks necessarily involve either novel information or novel combinations of information, the data from such tasks reinforce the suggestion that there is a differential decline in the efficiency of processing verbal and visuospatial information distinct from any differential effect of age on tasks involving familiar versus unfamiliar information or on speeded versus unspeeded tasks. The present findings and those just discussed (Halpern, 1984; Myerson et al., 1998; Tubi & Calev, 1989; Wechsler, 1997) also argue against interpreting the differential decline of verbal and visuospatial abilities in terms of the fluid/ crystallized distinction (Horn & Catell, 1967). Horn (1985, 1987) has clearly distinguished speed and working memory abilities from both crystallized intelligence (reflecting acculturation and knowledge) and fluid intelligence (reflecting broad reasoning abilities). Thus, the finding of differential age sensitivity of verbal and visuospatial speed and working memory cannot be adequately explained in terms of the fluid/ crystallized distinction. In addition to the preceding psychological hypotheses, two types of possible neurological mechanisms for the pattern of differential decline must also be considered. First, it is possible that age-related brain changes are specific, resulting in greater structural changes in visuospatial regions than in verbal regions (e.g., greater loss of neurons or connections in visuospatial cortical areas than in verbal areas). Alternatively, it is possible that age-related brain changes are global but that they have different functional consequences depending on the nature of the neural architecture (e.g., equivalent loss of neurons or of connections in visuospatial and verbal regions might affect the former to a greater extent than the latter because of intrinsic differences in connectivity). With respect to the first of these hypotheses (i.e., that age-related neurobiological changes are specific to certain brain regions) as applied to visuospatial and verbal processing, it has been suggested based on behavioral evidence that the Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 VERBAL AND VISUOSPATIAL PROCESSING SPEED right cerebral hemisphere may be subject to greater age-related deterioration than the left hemisphere (e.g., Ellis & Oscar-Berman, 1989; Klisz, 1978; Schaie & Schaie, 1977). However, neurobiological studies have failed to reveal significant differences in the age-related deterioration of the hemispheres or of the verbal and visuospatial cortical areas (e.g., Raz et al., 1997; Terry, DeTeresa, & Hansen, 1987). Although the frontal lobes and hippocampal regions appear to be especially age-sensitive (Raz et al.), there is at present no reason to suspect that changes in these structures would differentially affect verbal and visuospatial processing. With respect to the second hypothesis (i.e., that equivalent changes differentially affect specific brain regions), it is possible that preexisting aspects of the neural architecture determine the magnitude of the effects of biological aging on specific aspects of cognition and behavior. For example, Cerella and Hale (1994) used a simplified version of a neural network model to illustrate how connectivity differences might determine the age sensitivity of a neural structure. In their example, age-related breaks in a network could be bypassed through either one intermediate node (in the case of what was termed a unimedial network) or two intermediate nodes (in the case of what was termed a bimedial network). The unimedial network exemplifies a structure with high original connectivity density (i.e., prior to aging) and the bimedial network exemplifies a structure of lower connectivity density. It should be noted that the hypothesized differences in connectivity involve differences in the pattern of connections between neurons, rather than differences in the number of neurons. As a result, these differences might be difficult to ascertain anatomically, despite their potentially profound behavioral consequences (Cerella & Hale, 1994). For the simple cases considered by Cerella and Hale (1994) (i.e., unimedial and bimedial networks), it is possible to calculate the effects of breaks in the connectivity on the amount of time such networks would take to process information. Cerella and Hale demonstrated that the lower connectivity density (bimedial) network is particularly sensitive to age-related breaks. The 143 differential sensitivity of lower and higher connectivity networks is depicted in Figure 5 (adapted from Figure 18, Cerella & Hale, 1994), which plots the relative processing time in a unimedial and a bimedial network as a function of the probability of a connection being cut. As the probability of a cut increases, the expected processing time for the bimedial network increases to a far greater extent than that for the unimedial network. It should be noted that although the x axis in Figure 5 is labeled probability (cut) it might equally correctly be labeled age because the proportion of cuts (and hence the probability of a specific connection being cut) is assumed to increase with advancing age. Thus, the present findings raise the possibility that the architecture of visuospatial cortical areas may be of lower connectivity density compared with that of verbal cortical areas. However, this account is purely speculative at the present time, and other interpretations are worth pursuing. For example, differences in the cognitive architecture of verbal and visuospatial tasks (Fisher & Glaser, 1996) could also make performance in one domain decline more than the other even though the biological substrates for the two task domains were equally affected by age. Discriminating between these alternatives will require both the development of more refined neurobiological techniques and more explicit formulations of the kinds of differences in cognitive architecture that might be responsible for the greater declines in visuospatial abilities observed in the present study. Conclusions Regardless of the mechanisms underlying the differential decline in information-processing speed, the differential decline of verbal and visuospatial information-processing speed is likely to have profound consequences for other aspects of cognition. For example, age-related changes in processing speed have been strongly implicated in age-related changes in working memory (e.g., Fry & Hale, 1996; Kail & Salthouse, 1994; Salthouse, 1996). If age-related changes in working memory are the result of age-related changes in processing speed, then visuospatial working memory should decline to Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 144 Fig. 5. BONNIE LAWRENCE ET AL. The differential sensitivity to damage of lower (unimedial) and higher (bimedial) connectivity networks. As the probability of a cut increases, the expected processing time for the bimedial network increases to a far greater extent than that for the unimedial network. Adapted, with permission of the authors, from Figure 18, Cerella & Hale, 1994. a greater extent than verbal working memory. Consistent with this hypothesis, greater age-related declines are observed in spatial working memory than in verbal working memory (Myerson et al., 1998; Tubi & Calev, 1989; Wechsler, 1997). Age-related changes in processing speed have also been implicated in age-related changes in higher cognitive functions, where these latter changes appear to be mediated through processing speed’s impact on working memory (e.g., Fry & Hale, 1996; Kail & Salthouse, 1994; Salthouse, 1996). Thus, the differential decline in verbal and visuospatial processing speed observed in the present experiment may be part of a much broader pattern of differential decline in working memory and higher cognitive abilities across the adult life span. REFERENCES Birren, J. E. (1965). Age changes in speeded behavior: Its central nature and physiological correlates. In A. T. Welford & J. E. Birren (Eds.), Behavior, aging and the nervous system (pp. 191-216). Springfield, IL: Charles C. Thomas. Brinley, J. F. (1965). Cognitive sets, speed and accuracy of performance in the elderly. In A. T. Welford & J. E. Birren (Eds.), Behavior, aging and the nervous system (pp. 114-149). Springfield, IL: Charles C. Thomas. Cerella, J. (1985). Information processing rates in the elderly. Psychological Bulletin, 98, 67-83. Cerella, J. (1990). Aging and information-processing rate. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (3rd ed., pp. 201211). New York: Academic Press. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 VERBAL AND VISUOSPATIAL PROCESSING SPEED Cerella, J., & Hale, S. (1994). The rise and fall of information processing rates across the life span. Acta Psychologica, 86, 109-197. Cerella, J., Poon, L. W., & Williams, D. M. (1980). Age and the complexity hypothesis. In L. W. Poon (Ed.), Aging in the 1980’s: Psychological issues (pp. 332-340). Washington, D. C.: American Psychological Association. Coltheart, M., Davelaar, E., Jonasson, J. T., & Besner, D. (1977). Access to the internal lexicon. In S. Dornic (Ed.), Attention and performance VI (pp. 535-556). Hillsdale, NJ: Lawrence Erlbaum. Ellis, R. J., & Oscar-Berman, M. (1989). Alcoholism, aging and functional cerebral asymmetries. Psychological Bulletin, 106, 128-147. Fisher, D. L., & Glaser, R. A. (1996). Molar and latent models of cognitive slowing: Implications for aging, dementia, depression, development, and intelligence. Psychonomic Bulletin & Review, 3, 458-480. Fisk, A. D., & Fisher, D. L. (1994). Brinley plots and theories of aging: The explicit, muddled, and implicit debates. Journal of Gerontology: Psychological Sciences, 2, P81-P89. Fry, A. F., & Hale, S. (1996). Processing speed, working memory, and fluid intelligence: Evidence for a developmental cascade. Psychological Science, 7, 237-241. Hale, S., & Jansen, J. (1994). Global processing-time coefficients characterize individual and group differences in cognitive speed. Psychological Science, 5, 384-389. Hale, S., & Myerson, J. (1996). Experimental evidence for differential slowing in the lexical and nonlexical domains. Aging, Neuropsychology, and Cognition, 3, 154-165. Hale, S., Myerson, J., Faust, M., & Fristoe, N. (1995). Converging evidence for domain-specific slowing from multiple nonlexical tasks and multiple analytic methods. Journal of Gerontology: Psychological Sciences, 50, P202-P211. Hale, S., Myerson, J., & Wagstaff, D. (1987). General slowing of nonverbal information processing: Evidence for a power law. Journal of Gerontology, 42, 131-136. Halpern, D. F. (1984). Age differences in response time to verbal and symbolic traffic signs. Experimental Aging Research, 10, 201-204. Horn, J. L. (1985). Remodeling old models of intelligence. In B. Wolman (Ed.), Handbook of intelligence (pp. 267-300). New York: Wiley. Horn, J. L. (1987). A context for understanding information processing studies of human abilities. In P. A. Vernon (Ed.), Speed of information-processing and intelligence (pp. 201-238). New Jersey: Ablex. Horn, J. L., & Catell, R. B. (1967). Age differences in fluid and crystallized intelligence. Acta Psychologica, 26, 107-129. 145 Kail, R., & Salthouse, T. A. (1994). Processing speed as a mental capacity. Acta Psychologica, 86, 199225. Klisz, D. (1978). Neuropsychological evaluation in older persons. In M. Storandt, I. C. Stiegler, & M. F. Elias (Eds.), The clinical psychology of aging (pp. 71-98). New York: Plenum Press. Lima, S. D., Hale, S., & Myerson, J. (1991). How general is general slowing? Evidence from the lexical domain. Psychology and Aging, 6, 416-425. Madden, D. J. (1992). Four to ten milliseconds per year: Age-related slowing of visual word identification. Journal of Gerontology: Psychological Sciences, 47, P59-P68. Madden, D. J., Pierce, T. W., & Allen, P. A. (1993). Age-related slowing and the time course of semantic priming in visual word identification. Psychology and Aging, 8, 490-507. Mayr, U., & Kliegl, R. (1993). Sequential and coordinative complexity: Age-based processing limitations in figural transformations. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1297-1320. Mittenberg, W., Seidenberg, M., O’Leary, D. S., & DiGiulio, D. V. (1989). Changes in cerebral functioning associated with normal aging. Journal of Clinical and Experimental Neuropsychology, 11, 918-932. Myerson, J., Hale, S., Rhee, S., & Jenkins, L. (1998). Selective interference with verbal and spatial working memory in young and older adults. Manuscript submitted for publication. Raz, N., Gunning, F. M., Head, D., Dupuis, J. H., McQuain, J., Briggs, S. D., Loken, W. J., Thornton, A. E., & Acker, J. D. (1997). Selective aging of the human cerebral cortex observed in vivo: Differential vulnerability of the prefrontal gray matter. Cerebral Cortex, 7, 268-282. Salthouse, T. A. (1985). Speed of behavior and its implications for cognition. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging, (2nd ed., pp. 400-426). New York: Van Nostrand Reinhold. Salthouse, T. A. (1996). The processing-speed theory of adult age differences in cognition. Psychological Review, 103, 403-428. Schaie, K. W., & Schaie, J. P. (1977). Clinical assessment and aging. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (pp. 692-723). New York: Van Nostrand Reinhold. Sliwinski, M. J., & Hall, C. B. (1998). Constraints on general slowing: A meta-analysis using hierarchical linear models with random coefficients. Psychology and Aging, 13, 164-175. Swearer, J. M., & Kane, K. J. (1996). Behavioral slowing with age: Boundary conditions of the generalized slowing model. Journal of Gerontology: Psychological Sciences, 51B, P189-P200. 146 BONNIE LAWRENCE ET AL. Downloaded By: [Washington University in St Louis] At: 23:16 11 May 2010 Terry, R. D., DeTeresa, R., & Hansen, L. A. (1987). Neocortical cell counts in normal human adult aging. Annals of Neurology, 21, 530-539. Tubi, N., & Calev, A. (1989). Verbal and visuospatial recall by younger and older subjects: Use of matched tasks. Psychology and Aging, 4, 493-495. Wechsler, D. (1981). WAIS-R manual: Wechsler Adult Intelligence Scale – Revised. San Antonio, TX: Psychological Corporation. Wechsler, D. (1997). Wechsler Adult Intelligence Scale – III: Administration and scoring manual. San Antonio, TX: Psychological Corporation.
© Copyright 2026 Paperzz