ArticlePDF Available

Age and sex differences in reaction time in adulthood: Results from the United Kingdom Health and Lifestyle Survey


Abstract and Figures

Reaction times (RTs) slow and become more variable with age. Research samples are typically small, biased, and of restricted age range. Consequently, little is known about the precise pattern of change, whereas evidence for sex differences is equivocal. The authors reanalyzed data for 7,130 adult participants in the United Kingdom Health and Lifestyle Survey, originally reported by F. A. Huppert (1987). The authors modeled the age differences in simple and 4-choice reaction time means and variabilities and tested for sex differences. Simple RT shows little slowing until around 50, whereas choice RT slows throughout the adult age range. The aging of choice RT variability is a function of its mean and the error rate. There are significant sex differences, most notably for choice RT variability.
Content may be subject to copyright.
Age and Sex Differences in Reaction Time in Adulthood: Results From the
United Kingdom Health and Lifestyle Survey
Geoff Der
University of Glasgow
Ian J. Deary
University of Edinburgh
Reaction times (RTs) slow and become more variable with age. Research samples are typically small,
biased, and of restricted age range. Consequently, little is known about the precise pattern of change,
whereas evidence for sex differences is equivocal. The authors reanalyzed data for 7,130 adult partici-
pants in the United Kingdom Health and Lifestyle Survey, originally reported by F. A. Huppert (1987).
The authors modeled the age differences in simple and 4-choice reaction time means and variabilities and
tested for sex differences. Simple RT shows little slowing until around 50, whereas choice RT slows
throughout the adult age range. The aging of choice RT variability is a function of its mean and the error
rate. There are significant sex differences, most notably for choice RT variability.
Keywords: reaction time, cognitive aging, sex differences, adulthood, Health and Lifestyle Survey
The use of reaction times (RTs) as measures of cognitive func-
tioning has a long history, dating back at least to the 19th century
when Galton used them as part of the battery of tests included in
his “anthropometric laboratory” (Pearson, 1924). Over a century
later, their appeal remains undiminished, particularly in research
on aging. There are a number of reasons for this. One is their
relative simplicity. RT tasks are simpler to devise and administer
than most other cognitive measures or psychometric tests. None-
theless, they are commonly found to be correlated with other
cognitive measures and sometimes to be better predictors of im-
portant outcomes. For example, in a recent study (Deary & Der,
2005a), we found RTs to be better predictors of mortality than
scores on the Alice Heim 4 Test of General Intelligence (Heim,
Empirically, RTs are strongly associated with age. It is well
established that, during adulthood, RTs increase and become more
variable with age. Galton’s own data provided some of the earliest
evidence for slowing RTs (Johnson et al., 1985; Koga & Morant,
1923). Fozard, Vercruyssen, Reynolds, Hancock, and Quilter
(1994) and Deary and Der (2005b) summarized the support from
cross-sectional and longitudinal data, respectively. This decline
parallels age-related declines in other areas of cognitive function-
ing. In a meta-analysis of the relationship of age to a range of
cognitive measures, Verhaeghen and Salthouse (1997) found a
weighted average correlation between age and RT of .52; they also
found that “between 71% and 79% of the age-related variance in
the cognitive variables was shared with speed” (p. 246), which
included RT. From these and other results, they concluded that
“the speed variable may deserve special status in the context of
cognitive aging” (p. 246).
Madden (2001) described processing speed as “a fundamental
property of the central nervous system” (p. 288), and RTs are
measures of this processing speed. This viewpoint suggests a
simple and parsimonious explanation for age-related cognitive
decline, namely, that it is due to a general slowing of the system’s
processing speed (Salthouse, 1996). However, this processing
speed hypothesis of cognitive aging is not without its critics.
Sliwinski and Buschke (1999) showed that controlling for process-
ing speed reduces cross-sectional age effects much more that it
does longitudinal effects. They concluded that processing speed is
likely to reflect stable, non-age-related, individual differences.
Thus, despite the long research tradition and the wealth of
evidence linking RT and age, important aspects of the relationship
remain unclear. It has not been definitively established whether the
relationship is linear throughout the adult age range. If it is
nonlinear, is there evidence for a threshold at which cognitive
aging begins? Indeed, it is not even clear whether there is a single
pattern for the relationship of RT and age. If a slowing of general
processing speed is at the root of cognitive decline, then RT
measures would be expected to show a broadly similar age pattern,
even if other cognitive measures, such as measures of crystallized
intelligence, do not. Most of the evidence concerns mean RTs, and
much less is known about age differences in RT variability. Fi-
nally, it is not clear whether there are consistent sex differences in
RT, and if so, what part they would play in theories of cognitive
There are several reasons for these gaps in current knowledge.
Many of the samples studied consist entirely of older people. A
large part of the evidence for age changes is derived from studies
in which an older group was compared with a younger one, rather
Geoff Der, Medical Research Council Social and Public Health Sciences
Unit, University of Glasgow, Scotland; Ian J. Deary, Department of Psy-
chology, University of Edinburgh, Scotland.
Supplementary material is available online at
Geoff Der is funded by the Medical Research Council. Ian Deary is the
recipient of a Royal Society-Wolfson Research Merit Award.
Correspondence concerning this article should be addressed to Geoff
Der, Medical Research Council Social and Public Health Sciences Unit,
University of Glasgow, 4 Lilybank Gardens, Glasgow, United Kingdom
G12 8RZ. E-mail:
Psychology and Aging Copyright 2006 by the American Psychological Association
2006, Vol. 21, No. 1, 62–73 0882-7974/06/$12.00 DOI: 10.1037/0882-7974.21.1.62
than from data covering the whole age range. Hofer and Sliwinski
(2001) criticized this type of design as being prone to population
level confounds. One such source of confounding arises when the
older and younger groups are sampled from different populations,
for example, when the younger group comprises students (Botwin-
ick & Storandt, 1974; Harkins, Nowlin, Ramm, & Schroeder,
1974; Mathey, 1976).
Samples studied are frequently small, and hence lack power.
The meta-analysis of Verhaeghen and Salthouse (1997), although
restricted to studies reporting an age–speed correlation, illustrated
the range of typical sample sizes. Verhaeghen and Salthouse
amassed a total sample size of just over 11,000 from 28 studies.
The largest of these samples, that of Crook and West’s (1990)
study, had a sample size of 1,205. The samples of only three other
studies exceeded 400. Crook and West’s study was principally
concerned with age-related declines in name recall. RTs were only
used as predictors.
Of the large samples used to study RTs, few are representative
of the general population. Galton’s own subjects were a highly
selected group: They had to pay to be tested and as a result “a
sizable portion. . .consisted of professionals, semiprofessionals
and students” (Johnson et al., 1985, p. 876). The subjects in the
Baltimore Longitudinal Study of Aging (BLSA) were from the
“upper-middle socioeconomic level,” and “about 73% of the men
and 64% of the women had at least one college degree” (Fozard et
al., 1994, p. 180).
Attempts at combining data from different studies are under-
mined by the wide variety of different procedures and RT devices
used, some of them idiosyncratic. For example, the BLSA used a
nonstandard auditory task not originally designed for the collection
of RTs.
What is clearly needed is evidence from a large, representative
sample covering the whole adult age range and assessed using a
standard RT device. The aim of this study was to examine some of
the issues highlighted by reanalyzing data from just such a study,
the United Kingdom Health and Lifestyle Survey (HALS), which
we believe to be the largest population-based sample, covering the
whole adult age range, for which RT data are available. More
specifically, we aimed to delineate precisely and compare the
normal aging of four measures of RT: the mean and variability of
both simple and four-choice RTs. At the same time, we modeled
and tested for sex differences.
We also conducted additional analyses of RT variability con-
trolling for the mean. In comparison to mean RTs, the variability
of RT has received little attention until recent times. Anstey (1999)
found that RT mean and standard deviation were similarly related
to age, lung function, vibration sense, and grip strength and argued
that RT variability is worthy of more attention in cognitive aging
research. Hultsch and MacDonald (2004) reviewed the evidence
linking cognitive aging to three types of RT variability: interindi-
vidual variability, intraindividual variability across tasks, and in-
traindividual variability within tasks, which they referred to as
“diversity,” “dispersion,” and “inconsistency,” respectively. In-
consistency was our main focus, and inconsistency is known to be
elevated in a number of neurological conditions, including, for
example, mild dementia (Hultsch, MacDonald, Hunter, Levy-
Bencheton, & Strauss, 2000) and chronic fatigue syndrome
(Fuentes, Hunter, Strauss, & Hultsch, 2001).
Rabbitt, Osman, Moore, and Stollery (2001) suggested another
reason for studying RT variability. They claimed that people’s
fastest choice RTs are relatively unaffected by age but that “sub-
stantial individual differences in mean CRTs [choice RTs] often
reflect differences in the numbers of unnecessarily slow responses
that people make” (p. 982). The implication is that the increase in
mean RT with age is a result of increasing variability, and vari-
ability is, therefore, a more important component of cognitive
aging. They also found inconsistency to be associated with scores
on the Culture Fair Intelligence Test (Cattell, 1950). Hultsch,
MacDonald, and Dixon (2002) reported a similar finding: There
were no age-related increases in inconsistency for the fastest 20%
of response latencies, but marked increases for the slowest 20% of
The data for our analysis are drawn from the HALS. According
to the introduction, the main HALS report aimed to be “accessible
to as wide a readership as possible” and therefore presented “a
descriptive account of the information. . .principally in tabular
form” and “without the use of statistics” (Cox et al., 1987). In
keeping with this brief, the presentation of the RT data included
summary statistics (means and standard errors) tabulated by sex
and 10-year age bands. On the basis of the tabulated data, Huppert
(1987) concluded that age was associated with longer RTs and
more RT variability, both between and within subjects. She also
noted different patterns of aging for simple and choice RT and
“consistent but very small sex differences” (p. 44).
The data from HALS are available for secondary analysis from
the United Kingdom Economic and Social Research Council Data
Archive. We reanalyzed the data, modeling RTs as a continuous
function of age and explicitly testing for sex differences. Because
RT mean and variability are correlated, those analyzing the age
patterning of RT variability typically control for the mean in one
of two ways: either by examining the coefficient of variation (RT
SD/RT M) or by analyzing RT variability controlling for the mean.
The coefficient of variation provides a simple summary measure
that is useful for comparing the variability in different distribu-
tions. Because one of our aims was to compare the age patterns of
simple and four-choice RTs, it was an obvious candidate. In
contrast, controlling for the mean, although more complex, allows
greater flexibility in determining the relationship between the
variability and mean and thus more precise control for its influ-
ence. This is clear when one considers that modeling the coeffi-
cient of variation is equivalent to incorporating the reciprocal of
the RT mean as an offset in a model of the RT variability. It is
assumed to be the correct functional form without the assumption
being tested. Below, we describe both approaches and compare the
The HALS was set up with the aim of exploring the relationship of
lifestyles, behaviors, and circumstances to physical and mental health of a
large representative sample of the adult British population. Household
addresses were randomly selected from electoral registers using a three-
stage clustered design. From each household selected, one adult was
chosen using a standard sampling technique. The resulting sample com-
prised 9,003 British adults interviewed between autumn 1984 and summer
1985. The interviews were conducted in the respondents’ homes over the
course of two visits. An RT task was included in the second visit in which
7,414 respondents took part. These two samples were compared with the
national census on a range of characteristics. The single and divorced/
separated were found to be slightly underrepresented, and those with the
least education and lowest incomes were less likely to take part in the
second visit. Nonetheless Blaxter (1987) concluded that “these sources of
bias are small and the study appears to offer a good and representative
sample of the population” (p. 1).
RT was measured using a portable device designed especially for the
study. A diagram and full description are given in Deary, Der, and Ford
(2001). Briefly, the device has an LCD display screen beneath which are
five keys labeled 1, 2, 0, 3, and 4. The central 0 key is used for the simple
RT task. The index finger of the respondent’s preferred hand rests on this
key, and the respondent is told to press it as quickly as possible after 0
appears in the display. Eight practice trials were followed by 20 test trials.
The keys labeled 1, 2, 3, and 4 are used for the four-choice RT task. The
respondent rests the index and middle finger of each hand on the keys and
presses the corresponding key when one of the four digits appears in the
display. There were 8 practice trials and 40 test trials. In the test trials, the
digits 1 through 4 each appeared 10 times in a randomized order. For both
tasks, the time between the response and the display of the next digit varied
randomly between1sand3s.
The device does not store the results of individual trials but calculates
the mean and standard deviation of the test trials in milliseconds. For the
choice RT task, the mean and standard deviation of the correct and
incorrect responses were recorded separately, and the number of errors was
also recorded. The primary focus of this study was on four measures: the
mean and standard deviation of the simple RTs and those of the correct
responses to the choice RT task.
The analysis had two parts. First, we modeled the change in the RT
measures over the full age range, while also allowing for any gender
differences. Then, we examined in more detail the intraindividual variabil-
ity. All the analyses were performed using SAS software (Version 8.2, SAS
Institute, Cary, NC). For the principal analyses, the general linear models
procedure (PROC GLM) is used to fit polynomial regression models. For
these models, age was centered on the mean (44.9 years), and the polyno-
mial terms derived from the centered value. We also used the Box-Cox
transformation (Box & Cox, 1964) to normalize and stabilize variance
(PROC TRANSREG) and nonparametric regression (Hastie & Tibshirani,
1990) with cubic spline smoothers to check the functional form of some of
the relationships that we found (PROC GAM).
Of the 7,414 respondents who took part in the second interview,
7,216 completed the RT task. The reasons that the task was not
completed were problems with the use of hands (25), poor eyesight
(20), equipment failure (78), and miscellaneous other reasons (75).
In addition, we excluded 81 cases in which eight or more errors
were made on the choice RT task and 5 cases in which the
recorded standard deviation in one of the tasks was less than 10
ms. An error rate of 20% suggests problems in correctly carrying
out the task and is the same cutoff that was used in another study
using the same RT device (Deary & Der, 2005b). This left a
working sample of 7,130 respondents with an age range between
18 and 94 years. Although the whole working sample was included
in all statistical models, we excluded those who were 82 years or
older from graphs of the results because the small numbers and
consequent variability tended to unduly influence the scale of the
Figure 1 shows the means and standard errors of the four RT
measures by sex and age in 2-year bands. The corresponding
means and standard deviations are given in Table S1 which is
available at
male and female results have been offset slightly on the x-axis so
that the overlap of the standard error bars can be seen more clearly.
As expected, both simple and choice RTs slow as people age and
become more variable. However, the patterns appear to be differ-
ent. Simple RT shows little slowing until the 40s, and there is even
some suggestion that the intrasubject standard deviation (ISD),
which is a measure of intrasubject variability, decreases until the
mid 30s. Choice RT, in contrast, shows slowing throughout the age
range with corresponding increases in ISD. There is some evidence
of gender differences for each measure, with RTs in women being
slower and more variable. The largest and most consistent differ-
ence across the range of ages is the increased choice RT ISD of
Preliminary models were fitted to the data and the residuals exam-
ined. The results of the preliminary models and more details of their
residuals are given in Table S2 and Figure S1 in the online supple-
ment (available at
These revealed positive skewness throughout the age range, decreas-
ing slightly at older ages, together with increased variances at older
ages. The results would, therefore, have been biased and difficult to
interpret. One way to reduce the biases caused by skewness or
nonconstant variance is to transform the data. In practice, the choice
of transformation is often made by trying some of the common forms
(e.g., the log, square root, or reciprocal) and choosing the one with the
best results. Box and Cox (1964) suggested a more rigorous proce-
dure. They considered the family of transformations that can be
defined as y
0 and log(y) for
0 and
suggested fitting a range of values for
and choosing the one that
yields the maximum likelihood for the model in question, in this case
a polynomial regression of RT on age and sex. The resulting trans-
formation is optimal in two senses: It is chosen from a continuous
range of parameter values, as opposed to a few discrete values, and it
is optimized for the model to be fitted. The values of
obtained and
used to transform the data were RT mean, 1.65; RT SD, 0.36;
CRT mean,1.25; and CRT SD, 0.31.
The transformed data were then modeled as before. The sum-
mary statistics for these models of the transformed data are given
in Table 1 of this article and in Figure S2 in the online supplement
(available at
2 shows the predicted values and 95% confidence intervals for
these models transformed back to the original units. To aid com-
parison between this figure and Figure 1, we plotted each RT
measure on the same scale in both. The most notable difference is
that the curves for both simple RT measures are shallower in
Figure 2 with predicted means for the oldest subjects much lower
than the observed means in Figure 1. In comparison, the curves for
the choice RT measures are not markedly different from those in
Figure 1. The gender differences already noted remain, with the
biggest difference for choice RT variability and the smallest for
choice RT mean. Women’s mean RTs, both simple and choice,
appear to slow more rapidly at older ages.
Figure 1. Means and standard errors of reaction time measures (in milliseconds) by age and sex. Solid line men; dashed line women. The results
for men and women have been offset slightly on the x-axis so that the overlap of the standard error bars can be seen more clearly. SRT simple reaction
time; CRT choice reaction time; ISD intrasubject standard deviation.
Figure 3 shows the plotting of the coefficient of variation for
simple RT and choice RT by age and sex. The upper panels plot
the mean and standard errors in 2-year age groups. As in Figure 1,
the male and female plots are slightly offset. The lower panels
show the predicted values and confidence intervals for the fitted
models. To aid comparison, we drew all four plots to the same
scale. Parameter estimates for the models are given in Table 2.
Quite different patterns are evident: Simple RT has a curvilinear
relationship to age with no significant gender difference, whereas
choice RT shows only a slight, and mainly linear, increase with
age but a clear gender difference, reflecting the gender difference
in ISD. The decrease in simple RT ISD suggested by Figure 1 and
Figure 2 is more evident here.
Initial modeling of simple RT ISD controlling for the mean
resulted in a model with quadratic terms in simple RT mean and
age, with no significant effect of sex. To check the functional form
of the relationships with RT mean and age, we compared the
results with those from a semiparametric model that included
spline smoothers for each and used the generalized cross-
validation function to determine the degree of smoothing. This
suggested that a more complex polynomial was needed and led us
to fit a model with quintic terms in both age and simple RT mean.
Figure 4 shows the partial regression plots from the semiparamet-
ric models with the partial fit from the parametric model overlaid
as a dashed line. Because the semiparametric models are explor-
atory, interpretation of the results should be conservative. None-
theless, some of the main features can be discerned. The relation-
ship between the mean and the variability, controlling for age, is
linear throughout the main range. The departure from linearity is
confined to the top 1% of the distribution. The age pattern of
simple RT variability, controlling for the mean, broadly agrees
with results from the analysis of the coefficient of variation and
from the earlier models in which the mean was not controlled. The
semiparametric analysis and the higher degree polynomial suggest
a flatter trajectory between the initial decline observed in people in
their 20s until the beginning of the steep increase observed around
60 years of age. The estimates from the parametric model are given
in Table 3.
Table 4 shows the results of modeling choice RT ISD control-
ling for the mean and number of errors. When cubic effects of both
variables were included in the model, age was no longer signifi-
cant. Introducing linear and quadratic terms attenuated the effect of
age, but it remained significant until the cubic terms were entered.
The gender difference was attenuated but remained significant. As
with simple RT ISD, the results of this model were compared with
a semiparametric model using spline smoothers for the mean and
the number of errors. The right hand panels of Figure 4 show the
partial regression fits from these models, again with a dashed line
indicating the corresponding fit from the parametric model. There
is good agreement of the main range of the data. The relationship
between the mean and the variability is linear in the middle 98%
of the range. As with simple RT ISD, the relationship to the mean
appears different for the top 1% of the distribution. There is also
some suggestion of a departure from linearity in the bottom 1%.
Table 1
Summary Statistics for Models of Box–Cox Transformed Reaction Time (RT) Measures
Outcome Parameter Estimate SE F p F
SRT mean Intercept 0.6060111935 3.6921508E-7
Age 0.0000002515 2.523713E-8 1,007.76 .0001
0.0000000097 1.0046087E-9 205.10 .0001 .145
0.0000000001 3.556609E-11 9.91 .0017
Sex 0.0000029779 5.4592303E-7 100.64 .0001 .012
Age*sex 0.0000000044 2.3858424E-8 0.34 .5603
0.0000000031 1.3219769E-9 5.56 .0184 .001
SRT ISD Intercept 2.148890098 0.00238173
Age 0.001594583 0.00008856 594.16 .0001
0.000079802 0.00000488 264.06 .0001 .107
Sex 0.016784090 0.00288069 33.95 .0001 .004
CRT mean Intercept 0.7997475115 9.6802073E-7
Age 0.0000016995 6.616757E-8 4,006.14 .0001
0.0000000210 2.6339174E-9 101.84 .0001 .365
0.0000000001 9.32484E-11 1.81 .1791
Sex 0.0000026562 1.4313197E-6 31.32 .0001 .003
Age*sex 0.0000000519 6.2552832E-8 0.01 .9101
0.0000000104 3.4660044E-9 9.00 .0027 .001
CRT ISD Intercept 2.499384309 0.00144255
Age 0.002044210 0.00011700 1,940.54 .0001
0.000030828 0.00000419 86.38 .0001 .218
0.000000470 0.00000018 0.68 .4111
Sex 0.017092024 0.00217163 159.80 .0001 .017
Age*sex 0.000374711 0.00017699 0.08 .7744
0.000009681 0.00000613 0.10 .7575
0.000000654 0.00000028 5.62 .0178 .001
Note. F tests are based on Type I sums of squares, so the significance of each effect is conditional on all prior
effects in the model. The values of eta squared are for age, sex, and the Age Sex interaction. Sex is coded so
that women are the reference group. SRT simple RT; ISD intrasubject standard deviation; CRT choice
RT; age
age squared; age
age cubed.
Figure 2. Predicted values and confidence intervals for reaction time measures (in milliseconds) by age and sex. Solid line men; dashed line women.
SRT simple reaction time; CRT choice reaction time; ISD intrasubject standard deviation.
Figure 3. Simple and choice reaction time coefficient of variation: means and standard errors and predicted values and confidence intervals. Solid line
men; dashed line women. SRT simple reaction time; CRT choice reaction time; CV coefficient of variation.
Variability increases with the number of errors made. The biggest
difference (11 ms.) is between those who make no errors and
those who make one. Thereafter, the increase with each additional
error diminishes.
We will summarize and discuss our findings under three head-
ings: patterns of aging for the four RT measures, sex differences,
and the relationship between the mean and variability of the two
RT measures.
Patterns of Aging
The most notable result is that simple and four-choice RT age
differently. Both age nonlinearly, but aging for simple RT is more
markedly nonlinear. The simple RT mean barely increases until
people reach approximately 50 years of age, whereas the choice
Table 2
Results for Models of Simple and Choice Reaction Time Coefficient of Variation
Outcome Parameter Estimate SE F p F
SRT CV Intercept 22.94380512 0.22589922
Age 0.08742994 0.00992201 165.55 .0001
0.00587227 0.00054712 115.20 .0001
CRT CV Intercept 19.94761827 0.11426463
Age 0.07526313 0.00926781 165.86 .0001
0.00091284 0.00033216 5.05 .0247
0.00005861 0.00001443 9.03 .0027
Male 1.27156240 0.17201528 96.79 .0001
Age*male 0.03153821 0.01401967 0.03 .8518
0.00008442 0.00048576 3.31 .0689
0.00005238 0.00002188 5.73 .0167
Note. SRT simple reaction time; CV coefficient of variation; CRT choice reaction time; age
squared; age
age cubed.
Figure 4. Partial regression plots for models of simple and choice reaction time intrasu-
bject standard deviation (ISD) controlling for the mean. Solid line spline smoother;
shaded area confidence band; dashed line polynomial regression predicted values.
SRT simple reaction time; CRT choice reaction time. Reaction time variations are
given in milliseconds.
RT mean increases throughout the adult age range. The variabili-
ties show approximately the same pattern as their corresponding
means, except for two differences: Simple RT variability decreases
in early adulthood, whereas the mean does not; and for women,
choice RT mean increases more rapidly from age 70 than the
variability. These differences are mirrored in the different age
patterns of the coefficient of variation.
The difference in overall pattern is intriguing. It requires expla-
nation in the context of the processing speed hypothesis of cogni-
tive aging and for critics of the hypothesis, who maintain instead
that speed is a stable nonaging source of individual differences.
Part of such an explanation might lie in the different cognitive
loads that the two tasks involve. For this, there is relevant evidence
from the West of Scotland Twenty-07 study conducted by the
Social and Public Health Sciences Unit of the Medical Research
Council of the United Kingdom. The same RT procedure was used
in the study, but researchers also administered Part 1 of the Alice
Heim 4 Test of General Intelligence to 900 community-based
adults whose mean age was approximately 56. Choice RT was
more highly correlated with the intelligence test score than with
simple RT (.49 vs. .31; Deary et al., 2001). The relationship
underlying these correlations was approximately linear for choice
RT but complex and nonlinear for simple RT, suggesting that the
latter has little cognitive load at above-average ability levels (Der
& Deary, 2003). However, because the sample was approximately
56 years old, this population is close to the point at which age
differences in simple RT mean increase. Similar data are needed
from older and younger subjects.
Another, possibly complementary, explanation could be that the
two RT tasks require different practice periods to reach optimal
speed. Rabbitt (1993) showed that older and less able subjects need
more practice to reach their peak performance. In comparison with
the tasks on which Rabbitt’s results are based, the four-choice RT
task used in our study would be classified as very easy and,
therefore, one in which peak performance could be quickly at-
tained. Moreover, the need for more practice could equally be
regarded as part of normal aging and thus appropriately reflected
in the resulting RTs.
These two explanations could be regarded as components of a
more general phenomenon—the age–task complexity effect. How-
ever, Salthouse (1992) concluded that this effect was primarily due
to the demands on working memory, which are relatively low in
both of the RT tasks used here.
One further piece of evidence relevant to the question of
whether simple or choice RT better reflects cognitive aging is their
relative predictive validity. The 900 subjects of the Twenty-07
study mentioned above were followed up with respect to mortality
to age 70. Choice RT was found to be a better predictor of
mortality than simple RT and even better than the score on the
Alice Heim 4 Test of General Intelligence (Deary & Der, 2005a).
Table 3
Estimates From Model of Simple Reaction Time (SRT) ISD Controlling for the Mean
Parameter Estimate SE F p F
Intercept 129.3588217 30.31587857
SRT mean 1.0019975 0.26937844 6,879.72 .0001
SRT mean
0.0016852 0.00086651 379.81 .0001
SRT mean
0.0000024 0.00000125 1.44 .2305
SRT mean
1.7272E-9 8.2259E-10 2.93 .0869
SRT mean
4.4484E-13 1.978 1E-13 5.80 .0161
Age 0.1048165 0.10114728 16.95 .0001
0.0081782 0.00840201 70.51 .0001
0.0007556 0.00030182 3.25 .0716
0.0000341 0.00001158 1.13 .2885
0.0000008 0.00000029 7.80 .0052
Note. ISD intrasubject standard deviation; SRT mean
SRT mean squared; SRT mean
SRT mean
cubed; SRT mean
SRT mean quadrupled; SRT mean SRT mean quintupled; age
age squared; age
age cubed; age
age quadrupled; age
age quintupled.
Table 4
Estimates From Model of Choice Reaction Time (CRT) ISD Controlling for the Mean
Parameter Estimate SE F p F
Intercept 6.57045976 16.18467401
CRT mean 0.03733208 0.05705502 6696.96 .0001
CRT mean
0.00037165 0.00006386 271.35 .0001
CRT mean
0.00000019 0.00000002 93.88 .0001
Errors 12.92143070 1.29162500 242.61 .0001
2.90978654 0.61324601 55.02 .0001
0.22658497 0.07141164 8.55 .0035
Sex 7.94190010 0.80107272 98.29 .0001
Note. Sex is coded so that women are the reference group. ISD intrasubject standard deviation; CRT
CRT mean squared; CRT mean
CRT mean cubed; errors
errors squared; errors
errors tripled.
However, we should not forget that simple and choice RTs are
highly correlated. In this study, the means are correlated at
.67, implying that 45% of the variance in choice RT can be
explained by the simpler measure. Simple RT and the reasons
behind its increase with age remain questions worth studying.
The decrease in simple RT variability suggested in Figure 2 (and
more evident after controlling for the mean) has been noted pre-
viously. Pierson and Montoye (1958) studied simple RT in 400
males between the ages of 8 and 83. Using frequency of modal
response as their measure of consistency, Pierson and Montoye
concluded that “there is an increase in consistency of response
with age from eight years until about 30, after which a decline is
evident. Although an individual is capable of the fastest response
at about age 20, he is most consistent about 10 years later” (p.
419). In a more recent study of 291 individuals ranging in age from
6 to 89, Li et al. (2004) specifically focused on variability, which
they referred to as robustness. They concluded that “maximum
processing speed, processing robustness, and fluid intelligence
were achieved by individuals in their mid 20s” (p. 159). However,
their measure of robustness was a composite, and their numbers
were rather small for such a large age range. If, as Rabbitt and
others have suggested, RT variability plays an important role in
cognitive aging, it may also do so during the developmental phase
and thus warrant further study in that context.
Sex Differences
Sex differences were found for each of the four measures
examined, with the strongest and most robust being that for choice
RT variability. There was a suggestion of this effect in our analysis
of data from the Twenty-07 study (Deary & Der, 2005b), but
otherwise we are not aware of its having been previously reported.
It was still significant, although attenuated somewhat, in the anal-
ysis of variability controlling for the mean even when the effect of
age was no longer significant.
One possible explanation is that it is due to some sex difference
in the speed–accuracy tradeoff. We have examined this in two
ways. In the analysis of variability controlling for the mean and
number of errors, we tested for interactions between sex and error
rate, but none were significant. We also analyzed only those who
made no errors (results not shown), and the sex difference
A further possibility, suggested by Reimers and Maylor (in
press), is that the difference may reflect systematic sex difference
in performance across a block of trials. Their study included a
12-trial two-choice RT task in which participants pressed a key to
indicate whether a face presented briefly on screen was male or
female. They analyzed data from 5,137 individuals. They also
found RT in women to be more variable than that in men and the
effect remained in an analysis of the coefficient of variation. In
addition, they found that women were slower, but more accurate,
than men in the first two trials and slightly faster with the same
error rate in Trials 3 to 12. Excluding the first two trials eliminated
the sex difference in variability. They concluded that trial-to-trial
shifts in the speed–accuracy tradeoff could explain the sex differ-
ence in variability. Although their design may have exaggerated
trial-to-trial effects by not having a practice phase, their hypothesis
merits further research. Not having the data for individual trials,
we were not able to test it with the HALS data.
The sex differences for choice RT mean are the weakest and
most variable across the age range. Given that choice RT is more
widely studied than simple RT and RT mean more so than vari-
ability, it is easy to see how other studies with smaller samples and
incomplete age coverage could yield equivocal results.
Both simple and choice RT mean increases more rapidly for
women at older ages. This could partly reflect sex differences in
survival to these ages, whereby the men still alive and eligible for
sampling would represent a healthier subset. However, for choice
RT, the simultaneous narrowing of variability would not fit with
this explanation.
Reaction Time Variability
When the RT mean and variability are combined in the coeffi-
cient of variation, the age patterns are again different for simple
and choice RT. Choice RT is comparatively flat, with only a slight,
mainly linear, increase with age and the sex differences in choice
RT variability is still evident in the coefficient of variation. In
contrast, the simple RT coefficient of variation is markedly non-
linear and without significant sex differences.
The results are extended in the analyses of RT variability
controlling for RT mean. When choice RT variability is adjusted
for its mean and the number of errors, there is no longer a
significant age effect. Cubic terms in choice RT mean and number
of errors are needed to reduce the age effect to nonsignificance. A
positive relationship between the number of errors and choice RT
variability is expected even though the results are based only on
correct responses, because the responses immediately following an
error are known to be slower (Rabbitt, 1969).
Simple RT variability retains a significant relationship to age
after one controls for its mean. The decrease in simple RT vari-
ability in early adulthood, discussed above, was more evident in
the analyses in which the mean was controlled for, suggesting that
studies with smaller sample sizes, such as the sample size in the
study by Li et al. (2004), may be able to increase their power by
controlling for mean RT.
Strengths of the Study
This study has a number of strengths. Foremost among these is
the quality of the sample, which is very large, is representative of
the population, and covers the whole adult age range. The RT task
is of a standard format and uses a device designed and built
especially for the study. Both means and variabilities were col-
lected for simple and choice RTs. The Box-Cox procedure was
used to find a data transformation that was optimal for the analysis.
Limitations of the Study
The main limitation of the study is that the device used to record
RTs did not store the results of the individual trials. Their absence
precludes direct testing of some principal hypotheses. One is the
suggestion made by Rabbitt et al. (2001) that the slowing of RT
with age is mainly due to an increase in the number of atypically
slow responses. Another is the hypothesis of Reimers and Maylor
(in press) that was mentioned earlier.
Because the interviews were conducted in the respondents’
homes, the test conditions were not as rigorously controlled as test
conditions in a laboratory setting. The data are cross-sectional and,
therefore, inferences from the age differences that they exhibit to
the aging process are subject to potential confounding, notably by
period and cohort effects as well as by survivorship bias at the
older ages.
This study provides a detailed description of the aging patterns
of simple and four-choice RT means and variabilities. The patterns
differ markedly between simple and choice RTs even though the
two are highly correlated. Choice RT slows and becomes more
variable throughout adulthood, whereas simple RT barely changes
until people are approximately 50 years old. If speed, as measured
by RTs, deserves “special status in the context of cognitive aging”
(Verhaeghen & Salthouse, 1997, p. 246), then these results need to
be understood.
The relationship between RT mean and variability, whether
represented in the coefficient of variation or in analyses of vari-
ability controlling for the mean, also has different age patterning
for simple and choice RTs.
Sex differences are demonstrated for each of the measures, with
the most consistent and robust being that for choice RT variability.
This effect remained significant, even in analyses that accounted
entirely for age differences. This is a novel finding that awaits a
definitive explanation.
In 1987, Nettelbeck called for studies of large representative
samples in the field of processing speed measures. The original
report of the HALS study (Cox et al., 1987) was published in the
same year. By revisiting and reanalyzing the data, we have both
confirmed and elaborated on the original descriptive results in
exactly the way Nettelbeck envisaged.
Anstey, K. J. (1999). Sensorimotor variables and forced expiratory volume
as correlates of speed, accuracy, and variability in reaction time perfor-
mance in late adulthood. Aging, Neuropsychology, and Cognition, 6,
84 89.
Blaxter, M. (1987). Sample and data collection. In B. D. Cox (Ed.), The
Health and Lifestyle Survey (pp. 1–3). London: Health Promotion Re-
search Trust.
Botwinick, J., & Storandt, M. (1974). Cardiovascular status, depressive
affect, and other factors in reaction time. Journal of Gerontology, 29,
Box, G. E. P., & Cox, D. R. (1964). An analysis of transformations.
Journal of the Royal Statistical Society, Series B, 26, 211–252.
Cattell, R. B. (1950) Culture Fair Intelligence Test: A measure of “g.”
Savoy, IL: Institute for Personality and Ability Testing.
Cox, B. D., Blaxter, M., Buckle, A. L. J., Fenner, N. P., Golding, J. F.,
Gore, M., et al. (1987). The Health and Lifestyle Survey. London: Health
Promotion Research Trust.
Crook, T. H., & West, R. L. (1990). Name recall performance across the
adult life-span. British Journal of Psychology, 81, 335–349.
Deary, I. J., & Der, G. (2005a). Reaction time explains IQ’s association
with death. Psychological Science, 16, 64 69.
Deary, I. J., & Der, G. (2005b). Reaction time, age and cognitive ability:
Longitudinal findings from age 16 to 63 years in representative popu-
lation samples. Aging Neuroscience and Cognition, 12, 187–215.
Deary, I. J., Der, G., & Ford, G. (2001). Reaction times and intelligence
differences: A population-based cohort study. Intelligence, 29, 389 –399.
Der, G., & Deary, I. J. (2003). IQ, reaction time and the differentiation
hypothesis. Intelligence, 31, 491–503.
Fozard, J. L., Vercruyssen, M., Reynolds, S. L., Hancock, P. A., & Quilter,
R. E. (1994). Age differences and changes in reaction time: The Balti-
more Longitudinal Study of Aging. Journal of Gerontology: Psycholog-
ical Sciences, 49, 179 –189.
Fuentes, K., Hunter, M. A., Strauss, E., & Hultsch, D. F. (2001). Intrain-
dividual variability in cognitive performance in persons with chronic
fatigue syndrome. Clinical Neuropsychologist, 15, 210 –227.
Harkins, S. W., Nowlin, J. B., Ramm, D., & Schroeder, S. (1974). Effects
of age, sex, and time-on-watch on a brief continuous performance task.
In E. Palmore (Ed.), Normal aging II: Reports from the Duke longitu-
dinal studies, 1970 –1973. Durham, NC: Duke University Press.
Hastie, T. J., & Tibshirani, R. J. (1990). Generalized additive models.
London: Chapman and Hall.
Heim, A. W. (1970). Manual for the AH4 group test of general intelli-
gence. Windsor, England: National Foundation for Educational
Hofer, S. M., & Sliwinski, M. J. (2001). Understanding ageing: An eval-
uation of research designs for assessing the interdependence of ageing-
related changes. Gerontology, 47, 341–352.
Hultsch, D. F., & MacDonald, S. W. S. (2004). Intraindividual variability
in performance as a theoretical window onto cognitive aging. In R. A.
Dixon, L. Backman, & L.-G. Nilsson (Eds.), New frontiers in cognitive
aging. Oxford, England: Oxford University Press.
Hultsch, D. F., MacDonald, S. W. S., & Dixon, R. A. (2002). Variability
in reaction time performance of younger and older adults. Journal of
Gerontology: Psychological Sciences, 57B, 101–115.
Hultsch, D. F., MacDonald, S. W. S., Hunter, M. A., Levy-Bencheton, J.,
& Strauss, E. (2000). Intraindividual variability in cognitive perfor-
mance in older adults: Comparison of adults with mild dementia, adults
with arthritis, and healthy adults. Neuropsychology, 14, 588 –598.
Huppert, F. A. (1987). Cognitive Function. In Health and Lifestyle Survey.
London: Health Promotion Research Trust.
Johnson, R. C., McClearn, G. E., Yuen, S., Nagoshi, C. T., Ahern, F. M.,
& Cole, R. E. (1985). Gallon’s data a century later. American Psychol-
ogist, 40, 875– 892.
Koga, Y., & Morant, G. M. (1923). On the degree of association between
reaction times in the case of different senses. Biometrika, 15(3/4),
346 –372.
Li, S-C., Lindenberger, U., Hommel, B., Aschersleben, G., Prinz, W., &
Baltes, P. B. (2004). Transformations in the couplings among intellec-
tual abilities and constituent cognitive processes across the life span.
Psychological Science, 15, 155–163.
Madden, D. J. (2001). Speed and timing of behavioural processes. In J. E.
Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (5th
ed.). San Diego, CA: Academic Press.
Mathey, F. J. (1976). Psychomotor performance and reaction speed in old
age. In D. Kuhn (Series Ed.) & H. Thomae (Vol. Ed.), Contributions to
human development: Vol. 3. Patterns of aging: Findings from the Bonn
Longitudinal Study of Aging (pp. 36 –50). Basel, Switzerland: Karger.
Medical Research Council, Social and Public Health Sciences Unit. (2005).
West of Scotland Twenty-07 Study: Health in the community. Retrieved
February 21, 2006, from
Nettelbeck, T. (1987). Inspection time and intelligence. In P. A. Vernon
(Ed.), Speed of information processing and intelligence. Norwood, NJ:
Pearson, K. (1924). The life letters and labours of Francis Galton: Volume
II. Researches of middle life. Cambridge, England: Cambridge Univer-
sity Press.
Pierson, W. R., & Montoye, H. J. (1958). Movement time, reaction time,
and age. Journal of Gerontology, 13, 418 421.
Rabbitt, P. (1969). Psychology refractory delay and response-stimulus
interval duration in serial, choice-response tasks. In W. Koster (Ed.),
Attention and performance II (pp. 195–219). Amsterdam: North-
Rabbitt, P. (1993). Crystal quest: A search for the basis of maintenance of
practised skills into old age. In A. D. Baddeley & L. Weiskrantz (Eds.),
Attention: selection, awareness, and control. A tribute to Donald Broad-
bent. Oxford, England: Clarendon Press.
Rabbitt, P., Osman, P., Moore, B., & Stollery, B. (2001). There are stable
individual differences in performance variability, both from moment to
moment and from day to day. Quarterly Journal of Experimental Psy-
chology, 544, 981–1003.
Reimers, S., & Maylor, E. A. (in press). Gender effects on response time
variability and trial-to-trial performance: Reply to Deary and Der (in
press). Aging, Neuropsychology, and Cognition.
Salthouse, T. A. (1992). Why do adult age-differences increase with task
complexity? Developmental Psychology, 28, 905–918.
Salthouse, T. A. (1996). The processing-speed theory of adult age differ-
ences in cognition. Psychological Review, 103, 403– 428.
Sliwinski, M., & Buschke, H. (1999). Cross-sectional and longitudinal
relationships among age, cognition, and processing speed. Psychology
and Aging, 14, 18 –33.
Verhaeghen, P., & Salthouse, T. A. (1997). Meta-analyses of age-cognition
relations in adulthood: Estimates of linear and nonlinear age effects and
structural models. Psychological Bulletin, 122, 231–249.
Received September 16, 2004
Revision received August 8, 2005
Accepted September 9, 2005
... As well, stimulus complexity most influences signal processing time as confirmed in their study by Miller and Lov (2001). The RT is influenced by a several factors, and so far the following have been the most studied: stimulus type and intensity (Kemp et al., 1973), age (Rose et al., 2002;Der and Deary, 2006) and gender (Der and Deary, 2006), hand dominance (Boulinquez and Bartelemy, 2000), skill level of a particular activity, i.e the influence of training (Fontani et al., 2006), way of presentation of stimuli (Ando et al., 2002;Vignais et al., 2015) and other. ...
... As well, stimulus complexity most influences signal processing time as confirmed in their study by Miller and Lov (2001). The RT is influenced by a several factors, and so far the following have been the most studied: stimulus type and intensity (Kemp et al., 1973), age (Rose et al., 2002;Der and Deary, 2006) and gender (Der and Deary, 2006), hand dominance (Boulinquez and Bartelemy, 2000), skill level of a particular activity, i.e the influence of training (Fontani et al., 2006), way of presentation of stimuli (Ando et al., 2002;Vignais et al., 2015) and other. ...
... Kompleksnost stimulusa najviše utiče na vreme obrade signala što potvrđuju u svojoj studiji Miler i Lov (2001). Na VR utiče veliki broj faktora, a u dosadašnjem periodu najviše su ispitivani sledeći: tip i intenzitet stimulusa (Kemp et al., 1973), godine (Rose et al., 2002;Der and Deary, 2006) i pol (Der and Deary, 2006), dominantnost ruke (Boulinquez and Bartelemy, 2000), nivo veštine određene aktivnosti, odnosno uticaj treninga (Fontani et al., 2006), način prikazivanja stimulusa (Ando et al., 2002;Vignais et al., 2015) i drugi. ...
Conference Paper
Full-text available
Introduction In the last decade, in the world, but also in our country, there can be a significant increase in the participation of female recreational athletes in the half marathon races. Proper preparation is necessary for running a half marathon, in order to avoid harmful consequences. The impact of endurance training program on the functional abilities and body composition of women who run regularly, is investigated in this study. The aim of the study was to determine the effects of an eight-week training for the half marathon, on selected functional variables and body composition of women who are engaged in recreational running. Method The sample of subjects consisted of 19 female adults, 43.84±6.71 years old, from the Belgrade Racing Club (BRC). The sample of variables consisted of: maximum oxygen consumption (VO2max) - estimated on the basis of a 20-meter “Shuttle run” test; time at 5km - measured on maximum intensity race at a distance of 5km; anaerobic threshold running pace - estimated based on 5km running test; body weight, percentage of body fat and muscle mass - measured by bioelectrical impedance, model In body 720. All variables were measured before and after eight-week training program. The basis of the training program consisted of running three times a week. The results of the research were analyzed using descriptive and comparative statistics methods (t-test for dependent samples). Results and discussion The results of the t-test showed that there was a statistically significant improvement in the variables: VO2max (p<0.001), time at 5km race (p<0.001) and pace at the anaerobic threshold (p<0.001). The body mass of the subjects did not change statistically significantly (p=0.248), because the mass of the fat component decreased statistically significantly (p=0.022), while at the same time the muscle mass increased statistically significantly (p=0.020). Conclusion The eight-week experimental training program had a statistically significant effect on the improvement of all the variables, except the body weight.
... Also, other methodological specifications such as the setup of the current study and the simultaneous exposure of the subjects to both the visuospatial and auditory-verbal stimuli in this study might play a role. Response times are a function of factors such as sex, although this is controversial with many studies not finding a difference and one finding a difference merely in righthanded individuals [59,60,[62][63][64]; also, age [60,62,63], limb dominance [59,60], practice [65][66][67], and properties of stimulus such as duration or intensity [61] might predict the response time. Attention as well might affect the speed of responses, especially the complicated ones [61,68,69]. ...
... Also, other methodological specifications such as the setup of the current study and the simultaneous exposure of the subjects to both the visuospatial and auditory-verbal stimuli in this study might play a role. Response times are a function of factors such as sex, although this is controversial with many studies not finding a difference and one finding a difference merely in righthanded individuals [59,60,[62][63][64]; also, age [60,62,63], limb dominance [59,60], practice [65][66][67], and properties of stimulus such as duration or intensity [61] might predict the response time. Attention as well might affect the speed of responses, especially the complicated ones [61,68,69]. ...
... This variability of reaction time increased in older people and also increased with decreasing the sound intervention volume. The effects of age on response times have been documented earlier [60,62,63]. The effect of the intervention volume on reaction time variability might imply that these interventions could have played a positive role in decreasing the intrasubject variability, perhaps through masking and reducing potential auditory distractions existing in the lab environment. ...
Full-text available
Introduction: Binaural beats (BBs) are phantom sound illusions perceived when two sounds of slightly different frequencies are separately transmitted to the ears. It is suggested that some BB frequencies might entrain the brain and enhance certain cognitive functions such as working memory or attention. Nevertheless, studies in this regard are very scarce, quite controversial, and merely covering a very small portion of this vast field of research (e.g., testing only a few BB frequencies), not to mention adopting some limited methodologies (e.g., no assessment of the loudness of the BB sound, adopting only between-subject analyses, and testing only one perceptual modality). Hence, we aimed to assess the potential effects of alpha, beta, and gamma BBs on cognitive-behavioral parameters of working memory and attention examined simultaneously in two different modalities (visuospatial and auditory-verbal). Methods: This within-subject five-arm randomized placebo-controlled clinical trial included 155 trials in 31 healthy right-handed subjects (17 women, 14 men, 30.84 ± 6.16 years old). Each subject listened to 8-minute sessions of 10 Hz, 16 Hz, and 40 Hz binaural beats versus 240 Hz pure tone and silence (in random orders). In each 8-minute block, they played a dual 2-back task with feedback enabled. Their cognitive-behavioral parameters (working memory capacities, signal detection measures (hit rate, false alarm rate, sensitivity, and response bias), and reaction speed measures (response time and intrasubject response time variability)) were calculated. The effects of the sound interventions and short-term training on these working memory and attention measures were assessed statistically using mixed-model linear regressions, repeated-measures ANOVAs and ANCOVAs, Bonferroni post hoc tests, and one-sample t-tests (α = 0.05). Results: The following are some major statistically significant findings (P ≤ 0.05): In the visuospatial modality, the 10 Hz BB reduced the response time and intrasubject response time variability and reduced the extent of decline over time in the case of visuospatial working memory, sensitivity, and hit rate. In the auditory-verbal modality, the 10 Hz intervention reduced the hit rate, false alarm rate, and sensitivity. The 10 Hz intervention also caused the lowest intermodality discrepancies in hit rates and false alarm rates, the highest response time discrepancies, and negative discrepancies in working memories and sensitivities (indicating the superiority of the visuospatial modality). The response biases tended to be liberal-to-neutral in the verbal modality and rather conservative in the visuospatial modality. Reactions were faster in the visuospatial modality than the auditory-verbal one, while the intrasubject variability of reaction times was smaller in the auditory-verbal modality. Short-term training can increase the hit rate, working memory, and sensitivity and can decrease the false alarm rate and response time. Aging and reduced sound intervention volume may slow down responses and increase the intrasubject variability of response time. Faster reactions might be correlated with greater hit rates, working memories, and sensitivities and also with lower false alarm rates. Conclusions: The 8-minute alpha-band binaural beat entrainment may have a few, slight enhancing effects within the visuospatial modality, but not in both modalities combined. Short-term training can improve working memory and some cognitive parameters of attention. Some BB interventions can affect the intermodality discrepancies. There may be differences between the two modalities in terms of the response speeds and intrasubject response time variabilities. Aging can slow down the response, while increasing the volume of audio interventions may accelerate it.
... A community-based study (4) found that 30% of Chinese older persons aged over 80 years experienced cognitive impairment. Cognitive impairment in older adults could be attributed to the natural brain dysfunction over time (13,44,45) or the consequence sequela of long-term chronic physical diseases such as hypertension (46), cardiovascular disease (47), and diabetes (48). Moreover, all participants were recruited from nursing homes in this study. ...
Background Cognitive impairment is a major health concern in older adults. Few studies have examined the association between environmental factors and cognitive impairment, especially in high altitude areas. In this study, the prevalence of cognitive impairment in older adults living in high altitude was compared with those living in low altitude areas.Methods This was a comparative study conducted at Qinghai (high altitude group), and Guangzhou (low altitude group), China. Cognition, depressive symptoms and quality of life (QOL) were assessed using the Montreal Cognitive Assessment (MoCA), Patient Health Questionnaire (PHQ-9) and WHO Quality of Life brief version–WHOQOL-BREF, respectively.ResultsAltogether, 644 older adults (207 in Qinghai and 437 in Guangzhou) completed the assessment. The prevalence rate of cognitive impairment was 94.7% (95% CI: 91.6–97.7%) in older adults living in the high altitude area, while the corresponding figure was 89.2% (95% CI: 86.3–92.1%) in the low altitude area. After controlling for covariates, the high altitude group appeared more likely to have cognitive impairment (OR = 2.92, 95% CI: 1.23–6.91, P = 0.015) compared with the low altitude group. Within the high altitude group sample, multinomial logistic regression analysis revealed that older age (aged 74 and above) was significantly associated with higher risk of severe cognitive impairment (OR = 3.58, 95%CI: 1.44–8.93, P = 0.006), while higher education level (secondary school and above) was associated with decreased risk of moderate cognitive impairment (OR = 0.43, 95%CI: 0.22–0.85, P = 0.006). Within the high altitude group, QOL did not differ significantly between normal/mild, moderate and severe cognitive impairment subgroups across physical [F(1, 207) = 1.83, P = 0.163], psychological [F(1, 207) = 1.50, P = 0.225], social [F(1,207) = 2.22, P = 0.111] and environmental domains [F(1,207) = 0.49, P = 0.614].Conclusion This study found that cognitive impairment was more common among older adults living in the high altitude area. Regular screening and appropriate interventions should be provided to older adults in need.
... s 27,28 . Previous studies indicate that the simple reaction times for hearing and optic testing of 65-year-olds are approximately 10% and 16% longer, respectively, than those of 25-year-olds 34,35 . One study that used taste disks demonstrated that older adults showed age-associated deterioration in taste-detection thresholds, whereas the somatic sensations of the tongue were well retained 1 . ...
Full-text available
Abstract In super-aged societies, high salt intake substantially increases the risk of stroke and cardiovascular disease. Perceived low salty taste often prompts the addition of table salt to food. However, it remains unclear how older adults perceive the nature and intensity of salty taste in the mouth and brain. We compared the perceptions of salty taste intensities of older adults with those of young adults. Participants were 74 healthy adults: 31 older (age, 60–81 years [65.0 ± 5.5 SD]) and 43 young (age, 21–39 years [25.0 ± 3.6 SD]). Our research project comprises three sequential experiments. This article reports on the first two, which were (1) static and (2) dynamic sensory evaluations of taste perceptions in the mouth. Participants assessed the taste of 0.3 M and 0.5 M sodium chloride solutions in two types of sensory evaluations: (1) a cup tasting test, in which they sipped the solution from cups, spat it out, and rated static salty taste intensity, and (2) a time-intensity sensory evaluation, in which the solutions were delivered to participants’ tongues through a custom-made delivery system while they recorded dynamic taste intensities on a hand-held meter. Older adults perceived significantly lower taste intensities than young adults (p = 0.004 and p
Smartphones touchscreen interactions may help resolve if and how real-world behavioral dynamics are shaped by aging. Here, in a sample spanning the adult life span (16 to 86 years, N = 598, accumulating 355 million interactions), we clustered the smartphone interactions according to their next inter-touch interval dynamics. There were age-related behavioral losses at the clusters occupying short intervals (∼100 ms, R² ∼ 0.8) but gains at the long intervals (∼4 s, R² ∼ 0.4). Our approach revealed a sophisticated form of behavioral aging where individuals simultaneously demonstrated accelerated aging in one behavioral cluster versus a deceleration in another. Contrary to the common notion of a simple behavioral decline with age based on conventional cognitive tests, we show that the nature of aging systematically varies according to the underlying dynamics. Of all the imaginable factors determining smartphone interactions, age-sensitive cognitive and behavioral processes may dominatingly shape smartphone dynamics.
Implicit learning allows us to acquire complex motor skills through repeated exposure to sensory cues and repetition of motor behaviours, without awareness or effort. Implicit learning is also critical to the incremental fine-tuning of the perceptual-motor system. To understand how implicit learning and associated domain-general learning processes may contribute to motor learning differences in people who stutter, we investigated implicit finger-sequencing skills in adults who do (AWS) and do not stutter (ANS) on an Alternating Serial Reaction Time task. Our results demonstrated that, while all participants showed evidence of significant sequence-specific learning in their speed of performance, male AWS were slower and made fewer sequence-specific learning gains than their ANS counterparts. Although there were no learning gains evident in accuracy of performance, AWS performed the implicit learning task more accurately than ANS, overall. These findings may have implications for sex-based differences in the experience of developmental stuttering, for the successful acquisition of complex motor skills during development by individuals who stutter, and for the updating and automatization of speech motor plans during the therapeutic process.
Reaction times (RTs) are a measure of the time elapsed between the sensory stimulation and the occurrence of a response. It depends upon many factors. This paper focuses on one of the possible sources of variability introduced by the device used to gather RTs data and compares RTs performance between subjects who completed a Stroop task through desktop or mobile devices. The research hypothesizes that mobile devices users’ (a) have faster RTs in incongruent trials than the desktop ones, (b) have a lower error rate, and (c) perceive the task easier. The results showed significant differences in RTs, error rate, and perception of the difficulty of the task between devices but only the last two hypotheses are supported by data. Subjects under 40 years achieved a lower error rate than older people in the incongruent and neutral trials. Moreover, participants always perceived the task easier when they complete it through mobile devices. Some implications of the results are discussed, and a second ongoing within-subjects study is described.
The aim of the article was the intraindividual evaluation of reaction time at the Men’s World Athletics Championships from 1999 to 2019. We generated the rating of sprinters from the age point of view with comparison of two periods with different false start rules. In the result section, we analysed the sprinters that took part at WCH at least 3 times and appeared in the final. We assessed the reaction speed from the aging point of view, or more different false start judging conditions. The results shows that the stricter start judging rules in sprint disciplines did not have a significant influence on the reaction time. We also confirmed a research that the sprinters over 30 years old sprinters can achieve very low reaction time at the start. Reaction abilities can be improved by regular and systematic training, so it is necessary to pay attention to them in the training process and focus on their monitoring and subsequent improvement.
Full-text available
Reaction time variables are used widely in studies of human cognitive ageing and in research on the information processing foundations of psychometric intelligence. The research is largely based on biased population samples. In the present study, large (500+), representative samples of the population of the West of Scotland were tested at ages 16, 36 and 56 years on simple and choice reaction time. Participants were re-tested eight years later, at which time they also took the Paced Auditory Serial Addition Test (PASAT). We report simple and choice reaction time means and their variabilities, their stability across 8 years, and their correlations with the PASAT. Simple and choice reaction times become slower and more variable with age. Women from age 36 to 63 show more variability in choice reaction times than men, an effect which remains after controlling for mean reaction time. Reaction time differences largely account for age differences, but not sex differences, in PASAT scores.
In the analysis of data it is often assumed that observations y1, y2, …, yn are independently normally distributed with constant variance and with expectations specified by a model linear in a set of parameters θ. In this paper we make the less restrictive assumption that such a normal, homoscedastic, linear model is appropriate after some suitable transformation has been applied to the y's. Inferences about the transformation and about the parameters of the linear model are made by computing the likelihood function and the relevant posterior distribution. The contributions of normality, homoscedasticity and additivity to the transformation are separated. The relation of the present methods to earlier procedures for finding transformations is discussed. The methods are illustrated with examples.
This chapter makes a case for considering intraindividual differences in performance across tasks (dispersion) and intraindividual variability across occasions (inconsistency), in addition to mean level of performance, in characterizing groups and developing theoretical accounts of group differences and developmental trajectories. It reviews evidence that measures of intraindividual differences in variability tend to be stable over time and domains. © Roger A. Dixon, Lars Bäckman, and Lars Göran-Nilsson 2004. All rights reserved.
Background: Cross-sectional studies of samples varying widely in age have found moderate to high levels of shared age-related variance among measures of cognitive and physiological capabilities, leading researchers to posit common factors or common causal influences for diverse age-related phenomenon. Objective: The influence of population average changes with age on cross-sectional estimates of association has not been widely appreciated in developmental and ageing research. Covariances among age-related variables in cross-sectional studies are highly confounded in regards to inferences about associations among rates of change within individuals since covariances can result from a number of sources including average population age-related differences (fixed age effects) in addition to initial individual differences and individual differences in rates of ageing (random age effects). Analysis of narrow age-cohort samples may provide a superior analytical basis for testing hypotheses regarding associations between rates of change in cross-sectional studies. Conclusions: The use of age-heterogeneous cross-sectional designs for evaluating interdependence of ageing-related processes is discouraged since associations will not necessarily reflect individual-level correlated rates of change. Typical cross-sectional studies do not provide sufficient evidence for the interdependence of ageing-related changes and should not serve as the basis for theories and hypotheses of ageing. Reanalyzing existing cross-sectional studies using a sequential narrow-age cohort approach provides a useful alternative for evaluating associations between ageing-related changes. Longitudinal designs, however, provide a much stronger basis for inference regarding associations between rates of ageing within individuals.
The relationship between performance on measures of sensorimotor and physiological function and performance on measures of reaction time (RT) was investigated using an individual differences approach. Older participants showed greater intra-individual variability on the RT measures. They were also slower on measures of four choice visual RT but not on measures of three choice auditory RT. Sensorimotor and physiological variables explained nearly all the age differences in performance on the RT tasks and showed similar patterns of relationships with measures of speed, accuracy, and intra-individual variability. The results support a common factor interpretation of the relationships among these variables in old age.
A century ago, Francis Galton obtained data from thousands of individuals on a variety of sensory, psychomotor, and physical attributes. A substantial portion of these data has remained unanalyzed. In this article, we report on the reliability of the measures, developmental trends in mean scores, correlations of the measures with age, correlations among measures, occupational differences in scores, and sibling correlations. Test-retest correlations generally were very substantial. Growth continued for some individuals, especially those from lower economic strata, until they were in their mid-20s. Developmental trends during later childhood, adolescence, and early maturity were found to be similar to those described in contemporary developmental psychological literature, except that the tempo of development appears to have been slightly slower then than now. Persons from lower economic strata were smaller and weaker and showed less sensory efficiency. In addition, analyses of variance indicated that persons from lower economic strata continued their physical growth longer than persons from more advantaged environments. Sibling resemblances were substantial on most of the measures; opposite-sex siblings resembled one another less than did same-sex siblings.