| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Center for Health Sciences, SRI International (formerly Stanford Research Institute), Menlo Park, California
Gary E. Swan, Center for Health Sciences, SRI International, 333 Ravenswood Avenue, Menlo Park, CA 94025 E-mail: gary.swan{at}sri.com.
Decision Editor: Margie E. Lachman, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
2(58, N = 316) = 69.7, p = .14. The shared executive control factor had a heritability of 79% and accounted for 10%–56% of the genetic variance in performance on each of the four tests. Of the 4 tests examined in this analysis, Digit Symbol Substitution appeared to be the marker of executive control with the largest genetic component, whereas verbal fluency stood out as displaying a pattern of genetic and environmental influences distinct from the other 3 measures.
EXECUTIVE control in cognition refers to the development of action plans, sequencing, initiation, monitoring of outcomes, and the inhibition of distracting or competing influences over behavior (Royall and Mahurin 1995
). Executive control is hypothesized to consist of at least three components: (a) inhibition of a response tendency or a mental process, (b) selectively attending to a mental process that requires access to working memory and switching attention between different processes, and (c) coding and checking representations in working memory for an aspect other than their intrinsic content (Smith and Jonides 1998; Smith and Jonides 1999; West 1996).
Deficits in executive control lead to disorganization and ineffective planning, making even relatively simple activities of daily living such as cooking, cleaning, and other goal-directed behaviors difficult and less likely to conclude in a successful outcome (Carlson et al. 1999; Royall and Mahurin 1995). Decline in executive control has been observed to occur as a consequence of aging in both human (Panek, Rush, and Slade 1984; Parkin and Java 1999; Spieler, Balota, and Faust 1996; Uttl and Graf 1997; West and Bell 1997) and nonhuman primates (Smith, Roberts, Gage, and Tuszynski 1999) and may be a result of decline in dopaminergic activity in frontal cortical areas (Volkow et al. 1998). The apparent widespread decline in executive control with age has led some to speculate that such decrements, although largely undocumented, have significant effects on the public's health and associated costs (Fogel 1994).
At the extreme end of the health–disease continuum, loss of executive control occurs as a consequence of frontal stroke (Leskala et al. 1999), dementing illness such as Alzheimer's disease (Perry and Hodges 1999) and vascular dementia (Roman 1999), frontotemporal dementia (Yamaoka et al. 1996), other neurological conditions including Huntington's (Jason et al. 1997) and Parkinson's disease (Dujardin et al. 1999), and frank injury to the frontal region of the brain (Luria 1969). Less serious decrements in executive control have also been noted for individuals with mild cognitive impairment (Hanninen et al. 1997), ventricular enlargement (Breteler et al. 1994; Longstreth et al. 2000), and white matter hyperintense regions (Breteler et al. 1994; Boone et al. 1992) in the absence of any known pathological process.
Evidence for a genetic substrate to executive control is provided indirectly through studies of relatives of individuals with neurological or psychiatric conditions characterized by a loss of executive control. For example, performance on tasks involving executive control is worse in relatives of patients with schizophrenia than in controls (Byrne, Hodges, Grant, Owens, and Johnstone 1999; Faraone et al. 1999). Previous studies in aging twins also provide evidence for genetic influence on at least one index of executive control, the Digit Symbol Substitution Test (Finkel, Pedersen, McGue, and McClearn 1995; McClearn et al. 1997; Swan et al. 1990; Swan, LaRue, Carmelli, Reed, and Fabsitz 1992). Other, more direct, evidence for genetic involvement in executive control comes from studies showing lower levels of performance on measures of executive control in individuals with at least one apolipoprotein E 
4 allele (Berr et al. 1996; Haan, Shemanski, Jagust, Manolio, and Kuller 1999). Linkage to chromosome 17q21 has been reported in several families with clinical and neuropathological characteristics of frontotemporal dementia, a condition marked, in part, by disinhibition and decline in executive control (Lynch et al. 1994; Wilhelmsen, Lynch, Pavlou, Higgins, and Nygaard 1994; Yamaoka et al. 1996).
Several measures of executive control are commonly used in studies of both clinical and nonclinical aging samples (Mitrushina, Boone, and D'Elia 1999). Color–word interference (as measured by the Stroop test, Stroop 1935) involves the ability to inhibit an overlearned response in favor of an unusual one (Mitrushina et al. 1999, p. 74). Verbal fluency (as measured by Controlled Oral Word Association, Benton, Hamsher, Varney, and Spreen 1983) requires "efficient organization of verbal retrieval and recall and involves short-term memory, ability to initiate and maintain word production set and cognitive flexibility... as well as response inhibition capacity" (Mitrushina et al. 1999, p. 132). Digit Symbol Substitution (Wechsler 1981) is considered a measure of information processing speed that also places demands on working memory (Mitrushina et al. 1999). Performance on Trail Making B (Reitan 1958) assesses attention, visual scanning, speed of eye–hand coordination, and information processing, as well as the ability to alternate between sets of stimuli (Mitrushina et al. 1999). Psychometric analyses of these scales suggest commonality (Mitrushina et al. 1999). Statistical evidence of commonality is supported by clinical studies showing decreased performance on all four measures in individuals with disease or injury to frontal systems (Mitrushina et al. 1999).
Despite the presumed commonality among the four measures, there is some heterogeneity among them at the functional level. For example, Verbal Fluency involves, in addition to speed of information processing, flexibility, and inhibition of incorrect verbal responses, the use of verbal memory, a feature that separates it from functions used in the other tests. It is unknown to what extent functional heterogeneity among measures of executive control is determined by genetic influences in common with the tasks or by independent genetic and environmental factors. Moreover, if genetic influences shared by the tasks do play an important role in determining consistency in performance across all four measures, to what extent do these factors exert their influence independently on each measure or through an intermediate latent construct (e.g., "executive control") that reflects differential genetic effects on each of the measures? The answer to this question has important implications for the precise measurement of the executive-control phenotype in future measured genetic investigations seeking to identify association or linkage with specific genomic targets.
In this study, we used behavioral genetic methods to investigate the genetic architecture of several measures of executive control administered to a sample of community-dwelling elderly male twins. The application of multivariate genetic model fitting provides the following information: (a) whether the covariation among test performance is best described by a latent executive control factor, and, if so, (b) the extent to which genetic and environmental influences contribute to individual differences in the underlying factor, and (c) the extent to which individual differences in performance on the measures treated separately are accounted for by genetic and/or environmental influences.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Swan et al. 1990). Originally, this study was designed as a longitudinal investigation of cardiovascular disease (CVD) and associated CVD risk factors in 514 pairs of male twins, 254 monozygotic (MZ) and 260 dizygotic (DZ). All participating twins in this study were White men, American veterans of World War II and/or the Korean Conflict, were born during 1917–1927, and were 42 to 56 years old when first examined in 1969–1972. Details of the recruitment process, response rate, determination of zygosity, and initial examination protocol for this study are available elsewhere (Feinleib et al. 1977). Three follow-up examinations, after 10, 16, and 25 years, assessed CVD status and collected repeat measurements of physiological, biochemical, and psychosocial risk factors. By the time of the 1995–1997 examination, of the 622 twins seen at the 1986–1987 examination, 14% (n = 84) had died and 19% (n = 116) were alive but were not examined because of either illness, refusal, loss to follow-up, or participation by questionnaire only. All surviving twins were invited to participate in the 1995–1997 examination regardless of the participation status of their brothers. Of the 416 individuals examined in 1995–1997, 100 were singletons.
In the most recent follow-up (1995–1997) of the NHLBI Twin Study, participants received brain magnetic resonance imaging and a series of tests to assess executive control, verbal memory, and general cognitive function. Results from genetic analyses of brain structure (Pfefferbaum, Sullivan, Swan, and Carmelli 2000; Sullivan, Pfefferbaum, Swan, and Carmelli 2001) and of verbal memory (Swan et al. 1999) have been published elsewhere. The determination of cerebrovascular disease (also referred to as cerebrovascular accident, or CVA) relied on physician review of medical records including documentation of one or more of the following events: cerebral thrombus/embolism infarct, intracerebral hemorrhage, stroke, subarachnoid hemorrhage, or transient ischemic attack. Thirty-five participants (8.4% of the total sample) with medically documented CVA were excluded from the present analysis. The association between CVA and zygosity was not significant, 2(1, N = 416) = 0.28. Analyses in the present article are limited to intact twin pairs (80 MZ and 78 DZ pairs) who participated in the latest examination cycle of this cohort. All participants were community-dwelling individuals and were able to travel to one of four study sites (Boston University, Boston, MA; Harbor/University of California, Los Angeles, Medical Center, Torrance, CA; Indiana University Medical Center, Indianapolis, IN; or SRI International [formerly Stanford Research Institute], Menlo Park, CA). There were no participants with medically confirmed dementia in the sample used in this analysis.
The research reported in this article was reviewed and approved by the Institutional Review Boards of all four participating centers and is in compliance with the ethical rules for human experimentation as stated in the Declaration of Helsinki.
Measures
The following measures of executive control were included in this analysis: (a) Color–word interference (number of correct colors identified in 45 s during the interference trial; participants reporting difficulty with color perception were not given the test; Stroop 1935), (b) verbal fluency (total number of correct words generated in three 1-min trials; Benton et al. 1983), (c) Digit Symbol Substitution (number of correct digit–symbol pairings in 90 s; Wechsler 1981), and (d) Trail Making B (seconds to completion; Reitan 1958). As described earlier, each of these measures is considered to be an index of executive control by investigators in the field (Mitrushina et al. 1999; Parkin and Java 1999; Salthouse and Fristoe 1995; Salthouse, Fristoe, and Rhee 1996) and requires the combining of several cognitive and perceptual–motor functions, including sustained attention, visual perception, and short-term memory.
The Mini-Mental State Exam (MMSE; Folstein, Folstein, and McHugh 1975) was administered to characterize the overall level of functioning of the study sample.
Statistical Methods
Quantitative genetic modeling (Neale and Cardon 1992) involves the decomposition of the observed variation of a quantitative phenotype such as executive-control performance scores used here into genetic and environmental components of variance. The genetic variance may be due to additive (A) or dominant (D) genetic influences, whereas the environmental variance may be due to shared (C) environmental factors (e.g., early family environment shared by twins reared in the same household) and to individual nonshared environmental factors and/or measurement error (E). The above influences on the phenotype are assigned parameters a, d, c, and e, which are equivalent to the standardized regression coefficients of the phenotype observed on the latent factors A, D, C, and E, respectively. The amount of variance due to each source is calculated as the square of these parameters. In the twin paradigm, information in the MZ and DZ variance–covariance matrices is used to estimate the parameters a2, d2, c2, and e2 corresponding to the proportion of total phenotypic variance accounted for by genetic factors (i.e., heritability), shared environmental factors, and nonshared environmental factors, respectively. Examination of correlations within twin pairs in which the DZ intraclass correlations were much smaller than the MZ intraclass correlations suggested possible dominance effects on two of the measures (color–word interference and Digit Symbol Substitution). Model fitting with a term for dominance effects did not provide a significantly better fit over that in which additive effects were specified. Therefore, because the remaining two measures (verbal fluency and Trail Making B) showed evidence for additive genetic effects, we based the multivariate analyses on the ACE two factor model.
The objective of the present study was to use all the information in the observed variances and cross-test covariances of the four tests of executive control to investigate the nature of their genetic and environmental etiology (e.g., through shared or independent pathways). Specifically, we tested two models that represent different ways in which genes and the environment may account for the observed phenotypic correlations among the four tests. In the first, the independent pathway model (Fig. 1a), genes and the environment are hypothesized to have different and independent effects on twin covariances. In Fig. 1a, each of three Level I shared influences (A, C, and E) has its own path to performance on each of the four measures used here and, therefore, is identified as the independent pathway model. At Level II of this model, the residual variance in performance on each measure is assessed for genetic and environmental influences.
Under the second model, common pathway (Fig. 1b), both genes and the environment are hypothesized to contribute to an unmeasured intermediate (latent) variable which, in turn, is responsible for the observed phenotypic structure (Kendler, Heath, Martin, and Eaves 1987). There are three levels of interest in the common pathway model. First, general genetic and environmental influences are assessed at the level of the latent executive control factor labeled I in Fig. 1b. On the level labeled II in Fig. 1b, path coefficients representing the factor loadings for each test on the latent executive control factor are assessed. At the level labeled III in Fig. 1b, residual genetic and environmental influences are assessed for each cognitive task treated separately.
Another objective of genetic model fitting is to explain the pattern of observed variances and covariances by using as few parameters as possible. To achieve this goal we first compared the goodness of fit of the full ACE independent pathway model with the full ACE common pathway model. Once the decision was made on the best fitting general multivariate model, specific parameters of the multivariate model were individually tested to arrive at the most parsimonious model to fit the data.
The computer program Mx (Neale 1997) was used to fit the genetic models to the observed MZ and DZ variance–covariance matrices. The overall goodness of fit of each model was assessed by the maximum likelihood-ratio chi-square goodness-of-fit statistic (Neale and Cardon 1992) and Akaike's information criteria (AIC; Akaike 1987). A large chi-square indicates a poor fit, whereas a small chi-square indicates that the data structure is consistent with the model. The significance of specific parameters was tested by likelihood ratio tests by which the full genetic model was compared with submodels. When a submodel did not exhibit a significantly worse fit than the full model, the parameter tested was dropped from the full model and the reduced submodel was considered the best model to fit the data. In the final step of our multivariate genetic analysis, we estimated from the best fitting model the proportion of variance due to shared and nonshared genetic and environmental sources for each measure of executive control.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2(1, N = 619) = 3.12, p < .08. Overall, as with most longitudinal studies of older adults, there does appear to be a selection effect such that each returning/surviving portion of the cohort is successively healthier than the previous one (Tell, Fried, Hermanson, Manolio, Newman, and Borhani 1993). Comparison of the singletons (n = 100) with the analysis sample (n = 316) consisting of complete pairs seen at the 1995–1997 examination revealed few differences. Although the singletons were significantly older than members of intact pairs, 73.0 versus 71.7 years, t(408) = 4.07, p < .001, the two groups were not significantly different from each other with respect to years of education, global cognitive functioning (MMSE), or any of the measures of executive control.
Mean values for each zygosity on relevant characteristics are presented in Table 1 . There were no significant mean differences between MZ and DZ twins on age, years of education, or performance on color–word interference and Digit Symbol Substitution. MZ twins did, however, score higher, on average, than did DZ twins on verbal fluency (M = 36.0 vs. 32.8 correct words), t(315) = 2.29, p < .023, and they completed Trail Making B more quickly (M = 98.1 vs. 110.9 s), t(286) = -2.72, p < .007. MZ twins also demonstrated significantly less variability than DZ twins on Trail Making B, F'(149,165) = 1.54, p < .008.
|
23) or high (>23) MMSE scores and zygosity was not significant, 2(1, N = 316) = 2.69, p < .11. Because the MMSE was used only for descriptive purposes, MMSE score was not used as a basis for exclusion in this analysis. Table 2 presents correlations between the various measures, age, and education. Performance on color–word interference and verbal fluency was not associated significantly with age, whereas performance on Digit Symbol Substitution and Trail Making B was, to a small but significant degree. Performance on all four tests was associated to a moderate and significant degree with years of education. Given these associations, performance on each of the four measures was adjusted for age and education. Adjusted performance scores were then used in all subsequent genetic analyses.
|
|
|
|
|
Twenty-one percent of covariation in the executive control factor was due to nonshared environmental variance. The contribution of the executive control factor to the nonshared environmental components of variance in performance on color–word interference, verbal fluency, Digit Symbol Substitution, and Trail Making B were relatively small: 9%, 3%, 15%, and 9%, respectively (Table 5 ). Nonshared environmental effects on each specific test accounted for the remaining proportions of the phenotypic variance.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
These findings would suggest that verbal fluency is influenced largely by a separate set of genes unrelated to executive control. This conclusion is consistent with findings from previous research suggesting that individual differences in verbal fluency are determined by factors different from those influencing other measures of executive control. For example, Parkin and Java 1999 found reliable age differences on all frontal measures except for verbal fluency, a finding consistent with several previous studies showing word fluency to be relatively age insensitive (Parkin, Walter, and Hunkin 1995; Salthouse 1993, Salthouse 1996a, Salthouse 1996b). At least one author has speculated that verbal fluency may be more a measure of verbal IQ than of frontal lobe function (Miller 1984). Volkow and colleagues 1998 found performance on color–word interference and Symbol Digit Substitution (a variation of the Digit Symbol Substitution used in the present study) tests but not verbal fluency to be correlated with dopaminergic receptor activity in the frontal region after adjustment for age. It is possible that this particular test shares genetic and environmental influences with measures that more directly assess verbal abilities, including verbal memory. Future analyses of these data will determine the viability of this hypothesis.
The hypothesized polygenic substrate to executive control is given further credence through examination of residual genetic influences specific to performance on each measure. With the exception of Digit Symbol Substitution, residual genetic influences specific to each measure of executive control accounted for the majority of the observed genetic variance (ranging from 90% of total heritability in verbal fluency to 64% in Trail Making B). The best fitting model of the full twin variance–covariance matrix included a genetic component (e.g., the executive control factor) that is highly heritable. From this finding, one could conclude that executive control, though a polygenic trait, is influenced to a large degree by genetic factors. Moreover, the influence of the hypothesized executive control factor is largely through shared genetic as opposed to shared environmental pathways as reflected in the squared path coefficients presented in Fig. 2. Another conclusion from these results is that Digit Symbol Substitution appears to be the indicator of individual differences in executive control with the largest genetic variance and, given a choice of the four tests used here, would be the test of choice for future genetic studies of executive control.
The present analysis also reveals that although genetic influences play a role in individual differences on individual measures of executive control, the magnitude of these effects vary across the four different indexes examined here. Consistent with previously reported work (Finkel et al. 1995; McClearn et al. 1997; Swan et al. 1990; Swan et al. 1992), performance on Digit Symbol Substitution has a substantial heritable component (68%). Performance on color–word interference as well as on Trail Making B was similar with respect to the amount of variance attributable to genetic factors (both being 50%). Verbal fluency is notable for its comparatively smaller overall genetic component, 34%, about half the heritability of Digit Symbol Substitution. To our knowledge, this is the first time individual heritabilities have been reported for performance on the last three measures included in our analysis.
Despite the evidence in favor of a genetic hypothesis underlying individual differences in executive control, it is important to keep in mind that nonshared environmental factors play an undeniable role in up to 66% of the covariance on the different measures used here. Although they are not measured explicitly in the present study, such influences, no doubt, can be attributed to postsecondary schooling, occupational and avocational (e.g., reading habits, solving crossword puzzles) experiences as an adult, as well as life-long lifestyle and/or medical conditions such as tobacco smoking (Cerhan et al. 1998), elevated blood pressure (Elias, D'Agostino, Elias, and Wolf 1995; Launer, Masaki, Petrovich, Foley, and Havlik 1995; Swan, Carmelli, and LaRue 1996; Swan et al. 1998), or peripheral vascular disease (Haan et al. 1999).
Although work in humans and in nonhuman primates with lesions or head injury leaves little doubt that the prefrontal cortical region serves executive control functions, more recent work involving functional neuroimaging has shed light on the complexity of cortical and subcortical regions that are involved during tasks requiring executive control. These areas include the dorsolateral prefrontal, lateral orbital, anterior cingulate, motor, and oculomotor circuits (Bench et al. 1993; Golden 1976; Pardo, Pardo, Janer, and Raichle 1990; Perret 1974; Royall and Mahurin 1995). The ultimate complexity of the circuitry involved is, undoubtedly, far greater than currently recognized because of the differential involvement of a variety of regions underlying the processing of verbal and spatial information (Masterman and Cummings 1997; Smith, Jonides, Marshuetz, and Koeppe 1998).
The known neuroanatomic and metabolic substrate of frontal brain aging would further suggest higher order phenotypic complexity in the overall architecture of executive control. With aging, frontal brain regions experience numerous changes including reduction in volume, reduced neuronal size, loss of dendritic extensions, reduced number of synapses and receptor density, decreased concentration of neurotransmitters, decreased use of oxygen in the frontal cortex, and reduced blood flow (West 1996). That these changes may underlie, in part, reduction in executive control is given circumstantial support from the previously observed presence of dopamine modulation of frontal activity during performance of the color–word interference (Dolan et al. 1995), that frontal metabolism declines with age (Gur, Gur, Obrist, Skolnick, and Reivich 1987; Moeller et al. 1996), and that measures of dopamine D2 receptor activity are correlated positively with frontal metabolism (Volkow et al. 1993). It is important to point out that the issues of shared and/or specific genetic and environmental influences at the level of brain architecture (e.g., structural, volumetric) and component processes (e.g., regional blood flow, metabolism) are almost entirely unexplored.
Results from studies in twins that use refined phenotypic measures are beginning to appear in the literature. For example, differing degrees of heritability in elderly twins have been noted for components of intelligence (Finkel and Pedersen 2000; Finkel et al. 1995; McClearn et al. 1997) as well as for components of verbal learning and memory (Swan et al. 1999). The value of these studies along with the present one is the identification of multivariate candidate phenotypes that show promise (i.e., evidence for genetic variance) for inclusion in future investigations of association and/or linkage with specific gene variants. Moreover, the identification of markers of executive control that have minimal or no genetic variance is important because these processes may well be more amenable to environmental intervention in the aging adult.
A limitation of the present analysis is the unknown extent to which genetic and environmental factors interact to affect executive control. For example, it has been hypothesized that the presence of apolipoprotein E 4, a genetic variant involved in lipid metabolism and a known risk factor for excessive decline in several of the measures used here, results in increased cerebral susceptibility to the effects of environmental insults (Friedman et al. 1999; Mayeux et al. 1995). The effect of gene–environment interactions in executive control will be examined in the future through the use of matched twin-pair analyses of the extent to which differences in executive control between members of identical-twin pairs are associated with differences in various environmental exposures.
The nature of the sample used in the present analysis may limit the generalizability of the findings. The overall level of functioning of the sample as reflected in the mean overall MMSE score compared with normative data is about average for a sample of this age (Spreen and Strauss 1998). Moreover, compared with those who were not seen for a second neuropsychological examination, the characteristics of the sample available for this analysis suggest the presence of a "healthy survivor" effect. It is unknown from these data to what extent the presence of significant cerebral disease in the sample would have altered the conclusions favoring a common pathway model. Similarly, although these results can not be generalized beyond older men at present, it would be of great interest to repeat this analysis in a sample of older female twins.
The differences in means and/or variances as a function of zygosity noted for verbal fluency and Trail Making B represent violations of a basic assumption underlying twin analyses. The net effect of these differences would be an overestimation of heritability for each test. Because the latent executive control factor results from the computation of standardized scores, we have more confidence that the reported heritability estimate for it is unbiased. In fact, tests for differences in zygosity means and variances on the composite executive control score were not significant, providing support for this assertion.
Another caveat concerning the present analysis involves the bias inherent in the analytic approach toward overestimation of genetic effects to the detriment of detection of environmental influences shared by twins. An assumption of the underlying twin model is that any excess in MZ correlation as compared with DZ correlation is due to genetic influences rather than to an increase in correlated environmental effects among MZ pairs. Although our study did not find evidence for shared environmental influences on the executive control factor, this does not necessarily mean they do not exist. Large numbers of twin pairs (e.g., >1000 pairs of each zygosity) are typically required to detect these effects, and the present analysis (as are many other published investigations of twins) was almost certainly underpowered in this respect (Hopper 2000). It is significant, however, that of the four measures in this analysis, the one measure in the present study to suggest shared environmental influence, verbal fluency, could also most plausibly be influenced through twin–twin social interaction, one of the defining features of a shared environment. Other shared environmental factors could include family and educational experiences through high school in childhood.
In summary, although the best fitting model of covariation among measures of executive control included an executive control factor with substantial heritability, it is also evident that other sets of genes underlie performance on all four of the tests selected here. Notably, verbal fluency seems to be only loosely associated with the hypothesized polygenic executive control factor. The extent to which different genes are involved in the structure and function of specific neuroanatomic regions differentially involved in executive control remains to be determined.
|
|
Acknowledgments |
|---|
Site investigators for this study included Dorit Carmelli, PhD (Overall Principal Investigator, SRI International), Terry E. Reed, PhD (Indiana University Medical Center), Philip Wolf, MD (Boston University Medical Center), and Bruce Miller, MD (Harbor/UCLA Medical Center, Torrance, CA). We thank Lisa Jack, MS, Mary McElroy, MPH, Ruth Krasnow, Lois Abel, MA, Danene Clements, Sandy Kirkwood, PhD, Carol Miller, Ann Von Essen, Sharyn Moore, MA, and Kim McNulty for assistance with data collection and analysis.
Received for publication October 23, 2000. Accepted for publication July 23, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  W. S. Kremen, H. Xian, K. C. Jacobson, L. J. Eaves, C. E. Franz, M. S. Panizzon, S. A. Eisen, A. Crider, and M. J. Lyons Storage and Executive Components of Working Memory: Integrating Cognitive Psychology and Behavior Genetics in the Study of Aging J. Gerontol. B. Psychol. Sci. Soc. Sci., March 1, 2008; 63(2): P84 - P91. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Dickinson, M. E. Ramsey, and J. M. Gold Overlooking the Obvious: A Meta-analytic Comparison of Digit Symbol Coding Tasks and Other Cognitive Measures in Schizophrenia Arch Gen Psychiatry, May 1, 2007; 64(5): 532 - 542. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. H. Heflin, B. E. Meyerowitz, P. Hall, P. Lichtenstein, B. Johansson, N. L. Pedersen, and M. Gatz Cancer as a Risk Factor for Long-Term Cognitive Deficits and Dementia J Natl Cancer Inst, June 1, 2005; 97(11): 854 - 856. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Pajala, P. Era, M. Koskenvuo, J. Kaprio, A. Tolvanen, E. Heikkinen, K. Tiainen, and T. Rantanen Contribution of genetic and environmental effects to postural balance in older female twins J Appl Physiol, January 1, 2004; 96(1): 308 - 315. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
