Increasing Variance as a Function of Aging

1
Increasing Variance as a Function of Aging

• Harvey Babkoff
• Elisheva Ben-Artzi
• Leah Fostick
• Miriam Geal-Dor

2
Changes in Variance Due to Aging

• Changes in Mean and in Variance due to Aging
• Inter-Individual Variance and Aging
• Intra-Individual Variance and Aging

3

• Inter-Individual Variance and Aging

4
Lovden et al. (2004) and Schie (2000)

• Differences in physiological, socio-demographic
  and educational factors become greater as
  individuals age and therefore begin to have
  greater weighting than the genetic factors in
  determining cognitive performance.

5
Quantitative Genetic Analysis of Latent Growth
Curve Models ofCognitive Abilities in Adulthood
Chandra A. ReynoldsUniversity of California,
Riverside (2005)

• Increased variability in cognitive performance
  with age has been
• reported primarily in, but has not been limited
  to, cross-sectional
• analyses (e.g., Christensen, Mackinnon, Korten,
  Jorm, Henderson,
• Jacomb, 1999 Christensen, Mackinnon, Korten,
  Jorm, Henderson, Jacomb, Rodgers, 1999 Morse,
  1993). In the present case, such a pattern was
  seen with respect to systematic variances, that
  is, those explained by the latent growth model,
  for most measures. Variance increases have been
  ascribed to non-shared environmental or
  non-genetic stochastic processes (e.g., Finch
  Kirkwood, 2000). This interpretation was
  supported by our findings. The increasing
  environmental variation seen for nearly all
  cognitive traits in the present study could
  reflect stochastic processes that may have their
  seeds in early development and that are magnified
  in late life (Finch Kirkwood, 2000).

6
Quantitative Genetic Analysis of Latent Growth
Curve Models ofCognitive Abilities in
AdulthoodChandra A. ReynoldsUniversity of
California, Riverside (2005)

• Though many cognitive abilities exhibit marked
  decline over the adult years, individual
  differences in rates of change have been
  observed. In the current study, biometrical
  latent growth models were used to examine sources
  of variability for ability level (intercept) and
  change (linear and quadratic effects) for verbal,
  fluid, memory, and perceptual speed abilities in
  the Swedish Adoption/Twin Study of Aging. Genetic
  influences were more important for ability level
  at age 65 and quadratic change than for linear
  slope at age 65. Expected variance components
  indicated decreasing genetic and increasing
  non-shared environmental variation over age.
  Exceptions included one verbal and two memory
  measures that showed increasing genetic and
  non-shared environmental variance. The present
  findings provide support for theories of the
  increasing influence of the environment with age
  on cognitive abilities.

7
Examples of psychophysical data showing no
changes in either mean performance or variance as
individuals age
8
An example of stable performance across age in
accuracy of discrimination of auditory target
(tone) from among non-target tones. There is no
significant change in performance either in the
mean or in the variance of the distributions as a
function of aging
9
An example of stable performance across age in RT
to auditory target (tone) from among non-target
tones. There is no significant change in
performance either in the mean or in the variance
of the distributions as a function of aging
10
Another example of stable performance across age
in RT to auditory target (phonological stimulus)
from among non-target stimuli. There is no
significant change in performance either in the
mean or in the variance of the distributions as a
function of aging
11
(No Transcript)
12
The following data are an example of increased
inter-individual variance in the elderly together
with an increased mean but with a scaling
problem and change in distribution
13
Note what appears to be an increase in variance
from the young to the elderly subjects. This
seems, however, to reflect the significant
decrease in mean percent correct for the group of
elderly subjects relative to the younger subjects
and a resultant change in the distribution from
non-Gaussian to Gaussian.
14
Note the changes in the accuracy distributions of
the young versus the elderly subjects. While the
distribution of the young is not Gaussian, that
of the elderly appears to approach a Gaussian
distribution.
15
An example of change in performance as subjects
age in the accuracy of discrimination of a
semantic target (word) from among non-target
words. There is a significant change in mean
performance but not in the variance of the
distributions as a function of aging.
16
There was a significant increase in the variance
of the RT distributions to the semantic targets
from the young to the elderly (F 2.8278
plt.016). Both distributions are Gaussian.
17
The variance in the group aged 60 is
significantly larger than the variance in the
group aged 20-39. (F(1, 57) 5.66 plt.001). Both
distributions are Gaussian (See following slide).

18
(No Transcript)
19
(No Transcript)
20
Some cross-sectional studies have reported
results that were interpreted to mean that
inter-individual variability among the elderly is
related to variability in hearing loss (Helfer
and Wilber, 1990 Humes et al., 1994). However,
the results of other studies have found
inter-individual variability to be relatively
large even among elderly subjects with normal
audiograms for their age (Brasz et al., 2002
Scneider Pichora-Fuller, 2001 Versfeld and
Dreschler, 2002).
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Both the distributions of the hemispheric index
for the young and for the elderly are Gaussian.
There is a tendency for the variance of the
distribution of the hemispheric index of the
elderly to be larger than for the younger
subjects (F 1.985) plt.07
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28

• Intra-Individual Variance and Aging
• Speech Discrimination in Speech Noise

29
INTRA-INDIVIDUAL VARIANCE and AGINGIn recent
years there has been growing interest in
within-person variability as a potentially
informative individual difference parameter. At
least five factors may contribute to this
interest. (Correlates of within-person (across
occasion) variability in RT, Salthouse and
Berish,Neuropsychology, 2005)

• First, high levels of within-person variability
  may signify health-related problems.
• Second, unusual levels of within-person
  variability might function as an early indicator
  of impending cognitive change. That is, large
  fluctuations in ones momentary level of
  cognitive performance may be a precursor to
  certain types of cognitive pathology, and
  information about variability might provide a
  more useful baseline against which to evaluate
  the severity of extreme behaviors.
• Third, large within-person variability (i.e., low
  across-occasion consistency) could distort the
  evaluation of an individuals level of cognitive
  functioning.
• Fourth, within-person variability could
  contribute to inconsistency in research results
  across studies because in a single-occasion
  study, it is impossible to distinguish relatively
  stable trait variance from fluctuating state
  variance. For example, if there are age
  differences in within-person variability, it
  could lead to spurious conclusions of age-related
  increases in between-persons variability
  (Nesselroade, 2001). The possibility of large
  within-person variability in the performance of
  neuropsychological tests would also raise
  questions about the basis of the correspondence
  typically assumed between test performance and
  brain function.
• Fifth, identification of correlates of
  within-person variability may be informative
  about possible causes of individual differences,
  and particularly age-related individual
  differences, in cognitive functioning.

30
(No Transcript)
31
(Correlates of within-person (across occasion)
variability in RT. Salthouse and
Berish,Neuropsychology, 2005)

• There are two major results of these studies.
  The first is the demonstration of very large
  within-person variability in measures of RT. In
  two studies involving a total of 420 individuals,
  the median RT from one occasion to the next was
  found to vary as much as the mean (across
  occasion) RT varied among people who ranged from
  18 to 91 years of age. Because the reliability
  estimates indicate that for most participants the
  within-occasion RTs were more similar to one
  another than were the between-occasions RTs, this
  across-occasion variability cannot merely be
  attributed to
• random fluctuation.
• The second major result of the present studies is
  the consistent finding that measures of
  within-person variability appear to be secondary
  to measures of central tendency with respect to
  relations with age and with a variety of
  cognitive variables. For many variables, the mean
  and the SD are highly related because the
  variability around the mean is frequently greater
  when the mean is larger, but all of the analyses
  indicated that statistical control of the mean
  has a greater attenuating effect on correlations
  involving the SD than vice versa.

32
(No Transcript)
33
(No Transcript)
34
(No Transcript)