Decreased Reaction Time Variability Is Associated With Greater Cardiovascular Responses To Acute Stress
Decreased Reaction Time Variability Is Associated With Greater Cardiovascular Responses To Acute Stress
Decreased Reaction Time Variability Is Associated With Greater Cardiovascular Responses To Acute Stress
C 2016 The Authors. Psychophysiology published by Wiley Periodicals, Inc. on behalf of Society for Psychophysiological Research
V
DOI: 10.1111/psyp.12617
Abstract
Cardiovascular (CV) responses to mental stress are prospectively associated with poor CV outcomes. The association
between CV responses to mental stress and reaction times (RTs) in aging individuals may be important but warrants
further investigation. The present study assessed RTs to examine associations with CV responses to mental stress in
healthy, older individuals using robust regression techniques. Participants were 262 men and women (mean
age 5 63.3 6 5.5 years) from the Whitehall II cohort who completed a RT task (Stroop) and underwent acute mental
stress (mirror tracing) to elicit CV responses. Blood pressure, heart rate, and heart rate variability were measured at
baseline, during acute stress, and through a 75-min recovery. RT measures were generated from an ex-Gaussian
distribution that yielded three predictors: mu-RT, sigma-RT, and tau-RT, the mean, standard deviation, and mean of
the exponential component of the normal distribution, respectively. Decreased intraindividual RT variability was
marginally associated with greater systolic (B 5 2.009, SE 5 .005, p 5 .09) and diastolic (B 5 2.004, SE 5 .002,
p 5 .08) blood pressure reactivity. Decreased intraindividual RT variability was associated with impaired systolic
blood pressure recovery (B 5 2.007, SE 5 .003, p 5 .03) and impaired vagal tone (B 5 2.0047, SE 5 .0024, p 5 .045).
Study findings offer tentative support for an association between RTs and CV responses. Despite small effect sizes and
associations not consistent across predictors, these data may point to a link between intrinsic neuronal plasticity and
CV responses.
Descriptors: Reaction times, Cardiovascular, Acute stress, Reactivity and recovery
Much research has implicated increased or sustained cardiovascular metabolic disorders. Additionally, other physiological processes
(CV) responses to acute mental stress as detrimental to CV including dysfunctional proinflammatory responses and sustained
outcomes (Chida & Steptoe, 2010; Panaite, Salomon, Jin, & hypothalamic-pituitary-adrenal (HPA) axis reactivity have been
Rottenberg, 2015; Schwartz et al., 2003; Treiber et al., 2003). For associated with future development of CV disease risk factors
example, greater blood pressure reactivity to, and recovery from, (Brydon & Steptoe, 2005; Hamer, Endrighi, Venuraju, Lahiri, &
acute mental stress has been prospectively associated with the pro- Steptoe, 2012; Steptoe & Marmot, 2006).
gression of atherosclerotic plaques, greater intima media thickness, In the psychophysiology field, reaction time (RT) measures are
and hypertension (Chida & Steptoe, 2010; Panaite et al., 2015). often obtained to index cognitive ability, but a potentially novel
Mechanistically, individuals characterized by exaggerated CV use of RTs may be as a proxy measure of neural efficiency (NE).
responses to acute mental stress in the laboratory setting may This refers to the effectiveness with which the neural apparatus
respond in a similar fashion to natural stressors in everyday situa- communicate and process information (Jensen, 2006). On a basic
tions. This sustained hemodynamic activity may, over time, level, greater NE is represented by a smaller variability in RTs or a
increase tonic blood pressure and lead to hypertension and related lower intraindividual mean (Neubauer & Fink, 2009) suggesting
that a simple RT computation may not sufficiently capture the cog-
nitive process involved. Regardless of the way one chooses to
This research was supported by the British Heart Foundation RG/05/006 assess RT, research on RTs and CV outcomes is sparse. Recent epi-
and RG/10/005/28296.
Address correspondence to: Andrew J. Wawrzyniak, Ph.D, Psychia-
demiological work (Hagger-Johnson, Deary, Davies, Weiss, &
try and Behavioral Sciences, University of Miami Miller School of Batty, 2014) observed that slower and more variable RT was asso-
Medicine, 1120 NW 14th Street, Suite 1434, Miami, FL 33136-2107, ciated with an increased risk of all-cause and CV disease mortality
USA. E-mail: [email protected] in 5,134 adults from the NHANES III (National Health and Nutri-
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in tion Examination Survey) study adjusting for age, sex, and ethnic
any medium, provided the original work is properly cited. minority status. Limited work has also examined the relationship
739
740 A.J. Wawrzyniak et al.
between RTs or cognitive ability and a wide range of CV responses or a current diagnosis or treatment for a mental health disorder.
to acute mental stress that are implicated in cardiac risk. Ginty and Up-to-date medical records were used to verify participants’ health
colleagues (Ginty, Phillips, Der, Deary, & Carroll, 2011a) meas- characteristics to meet the inclusion criteria.
ured cognitive ability and simple RT at baseline in a large commu- Participants gave informed consent to participate in the study,
nity sample of individuals 55 years and over, and blood pressure and ethical approval was obtained from the Joint University
and heart rate reactivity to acute stress at 7-year follow-up. Results College London/University College London Hospital Research
showed that low cognitive ability and slow RT were significant Ethics Committee. Participants were prohibited from using any
predictors of blunted heart rate stress reactivity after accounting for antihistamine or anti-inflammatory medications for 7 days prior to
covariates including sociodemographics and medication use. In fur- psychophysiological testing, and were rescheduled if they pre-
ther analyses using a different age cohort (Ginty et al., 2011b), sented with colds or other sign of infection on their research
lower heart rate responses to acute stress were associated with appointment day. In addition, they were instructed to not consume
slower RT at 5- and 12-year follow-up independent of covariates caffeinated beverages or tea for at least 2 h prior to their visit, and
including baseline heart rate, socioeconomic position, and cohort to not partake in vigorous physical activity nor consume alcoholic
type. Blood pressure responses to acute stress were not associated beverages in the day prior to their appointment.
with RT in either study. In the Dutch Famine Birth Cohort Study The current analytic sample consists of 262 participants with
(Ginty, Phillips, Roseboom, Carroll, & Derooij, 2012), it was complete RT measures. The original study sample included 543
observed that impaired cognitive ability was associated with a individuals with psychophysiologic stress data, but the RT study
blunted CV response to acute stress. Therefore, slow RT and low component was introduced part way through the study. Participants
cognitive ability seems to be associated with lower heart rate with RT data did not differ from the rest of the sample in regard to
responses to acute stress. Crucially, according to the allostatic load average age and body mass index (BMI), baseline blood pressure,
framework (McEwen & Seeman, 1999), exaggerated autonomic heart rate and heart rate variability, distribution of gender, or
responses to stress is regarded as maladaptive. However, some employment grade (p value range 5 .16–.86). In addition, there
recent work is also exploring the notion that blunted or diminished were no between-groups differences in blood pressure, heart rate,
reactivity may be a marker of heightened chronic stress resulting in and heart rate variability responses to mental stress (p value
disturbances of biological systems. For example, blunted CV and range 5 .42–.86).
inflammatory stress responses were observed in otherwise healthy
individuals with Type II diabetes (Steptoe et al., 2014). Further- Materials
more, others have reported that blunted stress reactivity was associ- Participants’ height and weight were measured by a research nurse
ated with adverse health-related outcomes including substance according to a standardized protocol to determine participants’
addiction, eating disorders, and depressive disorders (Phillips, BMI (kg/m2). The latest grade of employment was used as an index
Ginty, & Hughes, 2013). of socioeconomic status (SES; Steptoe & Marmot, 2002). Detailed
To date, no study has examined the relationship between differ- RT data were obtained from a modified Stroop color-naming task
ent RT measures and a wide range of CV responses to acute mental (Stroop, 1935), whereas CV responses were elicited using a mirror-
stress in a healthy, aging sample. In addition, CV recovery from tracing task. Order of task presentation was counterbalanced
acute stress and heart rate variability, which are important predic- whereby half of the participants completed the Stroop task fol-
tors of CV risk (Panaite et al., 2015; Villareal, Li, & Massumi, lowed by mirror-tracing while the other half completed the tasks in
2002) have not been examined. Few studies have examined RTs the reverse order.
and CV reactivity and recovery as separate phenomena whereby
the cognitive measures are independent of the acute stressor RT data assessment. RTs were obtained from participants’ per-
designed to elicit CV changes (Ginty et al., 2012). Such a study formance on the Stroop color-naming task (Heathcote, Popiel, &
may help understand whether RTs predict a wide range of CV Mewhort, 1991), which was administered for 5 min. The task
responses to stress independently of known covariates for reactivity involved successive presentations of target color words at the top
and recovery. of a computer screen for 500 ms. The target words were printed in
Therefore, the present study examined the association between a discordant color (e.g., the word green printed in blue ink), and
several RT measures modeled using an ex-Gaussian distribution participants had to press a computer key that corresponded to the
(Vaurio, Simmonds, & Mostofsky, 2009) and CV responses to name of the target color word among a choice of four colors printed
mental stress in a healthy sample of participants drawn from the at the bottom of the screen (yellow, blue, red, and green). They
Whitehall II epidemiological cohort (Marmot et al., 1991). We were required to respond as fast and accurately as possible. The
hypothesized that RTs would be associated with hemodynamic and computer recorded each participant’s RT between presentation of
cardiac reactivity to, and recovery from, acute mental stress inde- target words and the pressing of the key as well as whether or not
pendently of a wide range of covariates. the correct key was pressed. Since RT task duration was standar-
dized across participants (5 min), different numbers of trials were
Method presented. These raw RT data were used to compute three RT vari-
Participants and Design ables (see below). Participants were provided with standard written
instructions and were allowed to practice for 1 min prior to task ini-
A subsample of participants from the Whitehall II epidemiological tiation. They were also instructed not to talk during the duration of
cohort was recruited between 2006 and 2009 for a psychophysio- the task.
logic study of acute stress responses and future CV disease risk fac-
tors. Exclusion criteria included a history or objective signs of Acute mental stress. Acute stress was elicited using a mirror-
coronary heart disease, a diagnosis or current treatment for hyper- tracing task (Campden Instruments Ltd.). Participants were
tension, inflammatory diseases, cancer treatment in the past 5 years, instructed to trace around the marked contour of a star with an
Reaction times and stress responses 741
electronic stylus while looking at the star’s reflection in a mirror. tracing task so that higher scores reflect greater reactivity. Recov-
The apparatus beeped and recorded an error every time the partici- ery from stress was computed by subtracting baseline values from
pant deviated from the marked contour. Performance was deter- the values obtained during each of the three time points into the
mined by the number of times the drawing of the star was recovery period so that greater scores indicate impaired recovery
completed, as well as the number of errors made during the draw- (slower return to baseline). HRV data were log-transformed prior
ing. Participants were told that an average person completes the to analyses other than for the robust regressions. Distributions of
drawing five times in the 5 min allowed with a minimum number change scores were visually screened for outliers, but no observa-
of mistakes. Standard written instructions were provided, and par- tion needed correction. The effect of acute stress on these CV out-
ticipants were allowed to practice the tasks for 1 min. comes was examined using repeated measures analysis of variance
(ANOVA) with Greenhouse-Geisser correction for degrees of free-
Stress-induced cardiovascular reactivity and recovery dom where appropriate.
assessment. Cardiovascular outcomes as indices of sympathetic
activation to the heart (Kapuku et al., 1999) included systolic blood RT data reduction. Only RTs for correct trials were used in the
pressure (SBP), diastolic blood pressure (DBP), and heart rate analyses (Silvia, Jones, Kelly, & Zibaie, 2011). Due to increased
(HR). Heart rate variability (HRV) expressed as the root mean cognitive demands when errors are processed in the brain (Coles,
square of successive N-N differences (r-MSSD, ms) was used as a Scheffers, & Holroyd, 2001; Falkenstein, Hoormann, Christ, &
measure of parasympathetic influence on the heart. SBP and DBP Hohnsbein, 2000; Koehn, Dickinson, & Goodman, 2008), RT
were monitored continuously from the finger during the entire test- slows during the second of two sequential incorrect trials (Hajcak
ing session using an appropriately calibrated Finometer (TNO Bio- & Simons, 2008). Therefore, incorrect trial times were excluded
medical Instrumentation, Amsterdam, The Netherlands), which from the analysis. All RTs faster than 200 ms were considered as
employs the vascular unloading technique (Imholz, Wieling, van “anticipatory” errors and where thus removed. Additionally, RTs
Montfrans, & Wesseling, 1998). Beatscope software was used for greater than four standard deviations above the individual mean
data reduction and to compute SBP and DBP variables. were considered as outliers (Schmiedek, Oberauer, Wilhelm, Suss,
HR and HRV were measured continuously using an ActiHeart & Wittman, 2007). Only eight participants required removal of one
monitoring device (Cambridge Neurotechnology, UK) attached to RT that was greater than four standard deviations, and a single case
the participant’s chest with electrocardiogram (ECG) electrodes. required the removal of two RTs. After removing these cases, all
The ActiHeart records both HR and movement; validity has been RTs were within four standard deviations of each participant’s
reported during running and resting conditions (Brage, Brage, mean.
Franks, Ekelund, & Wareham, 2005). The raw data were reduced
and analyzed using the HRV Analysis Software (Biomedical Signal RT data modeling. Ex-Gaussian distributions have been previ-
Analysis Group, University of Kuopio, Finland). ously used to characterize RT data (Hervey et al. 2006; Leth-
Steensen, Elbaz, & Douglas, 2000; Vaurio et al., 2009). The ex-
Procedure Gaussian distribution of RTs can be described as the normal, or
Gaussian, distribution, plus an independent, exponentially distrib-
Psychophysiological testing was carried out at either 9:30 am or uted variable. The mean and variance of the normal distribution
1:00 pm. A research nurse ensured that participants were not suffer- along with the exponential component together form the ex-
ing from a cold or a viral infection and that the pretesting instruc- Gaussian distribution (Heathcote et al., 1991; Leth-Steensen et al.,
tions had been followed. Anthropometric measures were obtained 2000). This distribution is comprised of mu (mu-RT—a central
according to standard protocol, and participants were escorted to a tendency measure similar to the normal distribution’s mean), sigma
stress laboratory where they sat in a padded recliner for the entire (sigma-RT—a value for the normal distribution’s variation), and
duration of the testing protocol. After instrumentation, participants tau (tau-RT—the mean of the exponential component of the distri-
rested quietly for 30 min. CV assessment of the last 5 min of this bution) (Hervey et al., 2006).
rest period were averaged to provide baseline values. In the ex-Gaussian distribution, mu-RT and sigma-RT represent
A research assistant trained in psychophysiology subsequently the mean and standard deviation, respectively, of the response
administered the RT task and the mirror-tracing task while CV data times. Tau-RT reflects the intraindividual variability whereby
were continuously assessed. Self-report ratings of task difficulty greater values indicate longer but infrequent response times. Exam-
and task involvement were obtained immediately after each task. ining RT data using the ex-Gaussian method avoids the possibility
Participants were then required to rest quietly for 75 min during a of more variable RTs being considered as outliers or noise, and pre-
stress recovery period while CV data were continuously assessed. vents the need to trim or log-transform the data in attempts to fit a
Three time periods that were equally spaced between each other normal distribution (Hervey et al., 2006). Ex-Gaussian distribution
(15–20, 40–45, and 70–75 min poststress) were assessed to provide was calculated for all RTs using a maximum likelihood fitting sys-
CV stress recovery values as per previous studies in our lab tem to generate mu-RT, sigma-RT, and tau-RT variables for each
(Steptoe & Marmot, 2006, Steptoe et al., 2014), which adequately participant. These RTs were then used as predictor variables of
captures both short- and long-term changes in recovery across mul- stress reactivity and stress recovery.
tiple autonomic measures. During recovery, participants remained
seated on the reclining armchair and were allowed to read nature Covariates. Participants’ sociodemographics and physiological
magazines or watch nature DVDs (Figure 1). characteristics including age, sex, BMI, and grade of employment
(Steptoe, Willemsen, Kunz-Ebrecht, & Owen, 2003) known to
Statistical Analysis have a direct impact on CV stress responses were used as covari-
ates. In addition, number of correct Stroop trials (a marker of cog-
Cardiovascular data. Reactivity to stress was computed by sub- nitive ability), stress task perceived difficulty (Sherwood, Davis,
tracting baseline values from the value obtained during the mirror- Dolan, & Light, 1992), and involvement (Silvia et al., 2011) were
742 A.J. Wawrzyniak et al.
treated as covariates because it may influence stress responses. Par- high breakdown point and 95% asymptotic efficiency for normal
ticipants rated task difficulty and involvement on a 7-point Likert- errors (Koller & Stahel, 2011) was used. This method has the
type scale but, due to the bimodal distribution of responses, we advantage of assigning less weight to observations with large
dichotomized it into a high/low binomial variable using a median regression residuals, allowing the inclusion of all available data
split. Finally, the baseline value of the appropriate CV outcome points. Notably, standardized betas are not provided in robust
was included in the models. regressions in that the standardization assumes that all data points
are equally weighted. However, R squared (R2) was calculated
Robust regression. To examine the association between the ex- using the unweighted and unscaled values of the predictors and out-
Gaussian distribution of RTs for each participant and their CV comes (Street, Carroll, & Ruppert, 1988).
responses to acute stress, we used robust regression. The highly Descriptive statistics are presented as means (6SD) or percent-
robust and efficient SMDM regression estimator that provides a age as appropriate. Analyses of associations between RT variables
Reaction times and stress responses 743
Table 1. Subject Characteristics p < .001. This measure significantly decreased in response to acute
stress and then increased during the three recovery points.
Variable Mean 6 SD or n (%)
Sex Stroop Task RT Performance
Male 102 (38.9)
Female 160 (61.1) Mean percent (6 SD) of correct trials was 63.64% (6 .22). The
Age (yrs) 63.3 6 5.5 mean (6 SD) RT for the correct trials was 2,344.21 (6 39.86) ms.
BMI (kg/m2) 25.70 6 4.0
Marital status
Married 156 (59.5) RT and Cardiovascular Stress Reactivity
Not married 105 (40.1)
Current smoker 13 (5.0) A summary of the regression models for the RT predictors (mu-
Ex-smoker 72 (27.5) RT, sigma-RT, tau-RT) and stress-induced CV response outcomes
Employment grade is provided in Table 3.
Higher 75 (28.6)
Mu-RT (ex-Gaussian mean RT component) was not associated
Intermediate 109 (41.6)
Lower 78 (29.8) with SBP (B 5 2.0026, SE 5 .002, p 5 .24) or DBP (B 5 2.00013,
Baseline systolic blood pressure (mmHg) 126.9 6 16.1 SE 5 .001, p 5 .88) stress reactivity after controlling for study
Baseline diastolic blood pressure (mmHg) 74.6 6 10.3 covariates. However, there was an association with HR (B 5 .0018,
Baseline heart rate (bpm) 67.2 6 8.8 SE 5 .001, p 5 .03; model R2 5 .13, p 5 .029) indicating that
Baseline heart rate variability (ms) 23.7 6 13.5
slower RT was an independent predictor of greater stress-induced
n 5 262. HR increases; baseline HR and task difficulty rating were also sig-
Note. Data are shown as mean 6 SD. BMI 5 body mass index. nificant covariates (p < .01). Additionally, changes in HRV were
associated with slower mu-RT (B 5 2.0018, SE 5 .001, p 5 .05,
model R2 5 .72, p 5 .05) with baseline HRV, employment grade,
and CV stress responses are presented as B (6SE) and R2 along
and stress task appraisal as significant covariates (p < .05).
with p values. Results were considered to be statistically significant
Sigma-RT (ex-Gaussian variance RT component) was not sig-
at the .05 level, and were considered a trend at the < .1 level.
nificantly associated with SBP (B 5 2.0075, SE 5 .006, p 5 .21),
Analyses were performed using R version 3.2.2 (R Core Team,
DBP (B 5 2.0001, SE 5 .002, p 5 0.97), HR (B 5 .0026,
2015) and the R package ‘robustbase’ (Rousseeuw et al., 2013).
SE 5 .0022, p 5 .23), or HRV (B 5 2.0032, SE 5 .0024, p 5 .19)
stress reactivity after adjustment for covariates.
Results Tau-RT (ex-Gaussian intraindividual variability RT component)
Participant Characteristics was only marginally, inversely associated with SBP reactivity
(B 5 2.009, SE 5 .005, p 5 .09; model R2 5 .05, p 5 .07; Figure
Participant sociodemographic characteristics and baseline CV val- 2a), tentatively suggesting that lower intraindividual variability in
ues are presented in Table 1. Data were obtained from 262 partici- RT predicts greater stress-induced increases in SBP, with BMI
pants (mean age 5 63.3 6 5.5 years; 61.1% female). Average BMI being a significant factor (p < .05). A similar pattern of marginally
indicated that the sample was only slightly overweight significant association was observed with DBP reactivity
(mean 5 25.7 6 4.0 kg/m2). Overall, participants were normoten- (B 5 2.004, SE 5 .002, p 5 .08; model R2 5 .10, p 5 .07; Figure
sive but only slightly above the optimal blood pressure cutoffs 2b). HR reactivity was not significantly associated with tau-RT
(mean SBP 5 126.9 6 16.1 mmHg; mean DBP 5 74.6 6 10.3 (B 5 2.003, SE 5 .002, p 5 0.15) and neither was HRV reactivity
mmHg). (B 5 2.0038, SE 5 .002, p 5 .10; model R2 5 .71, p 5 .09). How-
ever, baseline HRV, employment grade, and stress task involve-
Acute Mental Stress Effect ment were significant covariates in the final model (p < .05).
Table 2. Summary of Average Cardiovascular Values at Baseline, During Stress, and During Recovery Periods
Variable Baseline Acute stress Recovery 1 (120 min) Recovery 2 (145 min) Recovery 3 (175 min)
SBP (mmHg) 126.88 6 16.1 158.72 6 23.5 137.87 6 18.8 136.69 6 18.7 138.00 6 18.5
DBP (mmHg) 74.61 6 10.3 89.57 6 13.1 80.93 6 11.4 80.38 6 11.4 81.33 6 11.2
HR (bpm) 67.15 6 8.8 76.21 6 10.3 66.43 6 8.4 65.34 6 8.6 65.69 6 8.4
HRV (ms) 23.65 6 13.5 18.22 6 9.9 25.62 6 13.0 26.43 6 14.8 26.17 6 14.4
n 5 262.
Note. Data are shown as means 6 SD. SBP 5 systolic blood pressure; DBP 5 diastolic blood pressure; HR 5 heart rate; HRV 5 heart rate variability.
744 A.J. Wawrzyniak et al.
Table 3. Summary of Robust Regressions Analyses Between Reaction Time (RT) Predictors and Cardiovascular Response Outcomes
n 5 262.
Note. Data are shown as mean 6 SE and R2 derived from robust regression. Mu-RT is a measure of central tendency; sigma-RT is a measure of varia-
tion of the normal distribution; tau-RT is a measure of the mean of exponential component of distribution. Stress reactivity is a difference score
between stress task and baseline values; stress recoveries are the difference scores between each poststress recovery value and baseline. Regression
coefficients are adjusted for the baseline (prestress) value of the cardiovascular outcome, age, sex, BMI, employment grade, correct Stroop trials (cog-
nitive ability), and self-report stress task difficulty and involvement rating. SBP 5 systolic blood pressure; DBP 5 diastolic blood pressure; HR 5 heart
rate; HRV 5 heart rate variability.
*p .05. **p .09.
p 5 .63), 45 (B 5 2.0009, SE 5 .0007, p 5 .20), or 75 min post- between mu-RT and HRV recovery at 15 min (B 5 2.003,
stress (B 5 2.001, SE 5 .0007, p 5 .19). SE 5 .001, p 5 .69), a significant association emerged at 45 min
The association between mu-RT and HR stress recovery only poststress (B 5 2.0021, SE 5 .00093, p 5 .02; model R2 5 .125),
approached significance level at 15 min poststress (B 5 .0007, which then trended toward nonsignificance by the end of the stress
SE 5 .001, p 5 .09; model R2 5 .15, p 5 .09), but it became not sig- protocol (B 5 2.0015, SE 5 .001, p 5 .088; model R2 5 .15,
nificant at 45 min (B 5 .0004, SE 5 .0004, p 5 .35). However, at p 5 .07). Baseline HRV was also a significant factor (p < .001).
75 min poststress, this association was significant (B 5 .0009, There were no significant associations between sigma-RT and
SE 5 .0005, p 5 .05; model R2 5 .20, p 5 .05) suggesting that SBP, DBP, HR, or HRV stress recovery at any time point into the
slower RT predicted impaired HR recovery at the end of the stress recovery period (p range 5 .126–.878).
period adjusting for study covariates (baseline HR and age were Tau-RT was associated with impaired SBP stress recovery at 15
also significant factors, p < .01). Although there was no association min poststress (B 5 2.007, SE 5 .003, p 5 .03; model R2 5 .06,
a b
Figure 2. a: Scatter plot of the association between tau-RT and systolic BP reactivity to acute mental stress (n 5 262). b: Scatter plot of the associa-
tion between tau-RT and diastolic BP reactivity to acute mental stress (n 5 262). Reaction time tau is expressed in milliseconds and represents the
intraindividual variability in reaction times. Change in systolic and diastolic BP is the difference between baseline and acute stress values so that
greater scores reflect higher stress-induced reactivity. Individuals with lower intraindividual variability in reaction time tended to show marginally
greater systolic BP (p 5 .09) and diastolic BP (p 5 .08) stress reactivity. The association is fully adjusted for age, sex, BMI, employment grade (SES),
stress task perceived difficulty and involvement, correct Stroop RT trials (cognitive ability), and baseline BP.
Reaction times and stress responses 745
a b
Figure 3. a: Scatter plot of the association between tau-RT and systolic BP stress recovery at 20 min after acute stress (n 5 262). b: Scatter plot of
the association between tau-RT and systolic BP stress recovery at 75 min after acute stress (n 5 262). c: Scatter plot of the association between tau-
RT and heart rate variability stress recovery at 45 min after acute stress (n 5 262). Reaction time tau is expressed in milliseconds and represents the
intraindividual variability in reaction times. Change in SBP is the difference between the stress recovery time point and the baseline value so that
greater scores indicate impaired recovery (delayed return toward baseline) from acute stress. Change in heart rate variability is the difference
between the baseline and the 45-min poststress value so that lower values reflect impaired recovery from acute stress. Individuals with lower intrain-
dividual variability in reaction times showed impaired vascular and vagal recovery from acute stress. These associations are fully adjusted for age,
sex, BMI, employment grade (SES), stress task perceived difficulty and involvement, correct Stroop RT trials (cognitive ability), and baseline BP
and HRV.
p 5 .03), indicating that lower intraindividual variability in RT pre- R2 5 .12), indicating that lower intraindividual variability
dicted impaired SBP stress recovery independent of study covari- independently predicted impaired HRV stress recovery (Figure 3c).
ates (Figure 3a). There was no significant association between tau- Baseline HRV was again a significant factor (p < .001).
RT and SBP stress recovery at 45 min (B 5 2.006, SE 5 .004,
p 5 .13), but there was a marginally significant association at 75
min poststress (B 5 2.006, SE 5 .003, p 5 .087; model R2 5 .06, Discussion
p 5 .08) (Figure 3b). Baseline SBP and correct Stroop trial were
also significant factors (p < .04). The present study examined associations between three RT meas-
Tau-RT was neither associated with DBP stress recovery at any ures assessed with the Stroop task and modeled through an ex-
time point into the recovery period (p range 5 .12–.38) nor with Gaussian distribution, and CV responses to acute mental stress in a
HR stress recovery (p range 5 .60–.99). healthy, older sample. We hypothesized that the three RT variables
Finally, tau-RT was associated with HRV stress recovery at 45 (mu-RT, sigma-RT, tau-RT) would be associated with CV
min after stress (B 5 2.0047, SE 5 .0024, p 5 .045, model responses to acute stress. However, given the inconsistent findings
746 A.J. Wawrzyniak et al.
in the literature, we did not predict the direction of these heart rate variability), and then with a better return to baseline
associations. during recovery (systolic BP). Intrinsic plasticity may account
The results revealed a complex pattern of associations between for previous studies’ findings on RT such as in work associating
the RT measures and stress-induced responses that was not always lower cognitive ability and slower RT with blunted heart rate
consistent across predictors and was mostly modest in magnitude. stress reactivity (Ginty et al., 2011a, 2011b) and associations
The most consistent finding observed in these analyses was that between poorer cognitive ability and lower CV reactions to
sigma-RT, a measure of variation in RT in a normal distribution, acute stress (Ginty et al., 2011a, 2011b). Further support for
was not significantly associated with any of the CV reactivity and intrinsic neuronal plasticity as a factor in CV reactivity has been
recovery outcomes at any time point (ps .12). In contrast, mu- reported in fMRI studies that have observed associations
RT, a measure of average response time similar to the mean of a between blunted reactivity and neural hypoactivation (Ginty,
normal distribution, was associated with heart rate and heart rate Gianaros, Derbyshire, Phillips, & Carroll, 2013).
variability stress reactivity, which is mostly consistent with previ-
ous work (Ginty et al., 2011a, 2011b, 2012). This indicated that Strengths and Limitations
individuals with slower RT had greater heart rate and (lower) heart
rate variability responses to acute stress after adjustment for several The present study has both limitations and strengths. This study’s
covariates including age, sex, BMI, number of correct trials as a findings are bolstered by the fact that participants were carefully
proxy of cognitive ability, subjective ratings of task engagement selected on the basis of being free of any objective sign of chronic
and difficulty, and employment grade. Since we also adjusted for diseases. As such, participants in the study may represent an unusu-
the baseline (prestress) value of these cardiac markers, this result is ally healthy sample of the population that may potentially limit the
not attributable to differences in resting cardiac autonomic activity generalizability of the findings to a wider population. Although we
in individuals with slower RT. controlled for a number of covariates and accurately implemented
Furthermore, mu-RT was a significant, independent predictor of a reliable stress testing procedure, the cross-sectional nature of the
impaired heart rate and heart rate variability recovery from stress at study cannot rule out the possibility that other unmeasured factors
some, but not all, time points. This suggests that two important car- might have contributed to the findings. In addition, reverse causal-
diac parameters failed to return to baseline values by the end of the ity is a possibility in that heightened reactivity could influence RT.
stress protocol in individuals with slower RT responses. This find- Although this is plausible, further work would require an experi-
ing was underscored by the marginally significant association mental manipulation of reaction times in order to determine the
between mu-RT and impaired systolic BP recovery at 75 min causal relationship between RTs and CV responses to mental stress.
(p 5 .06), although this association was not replicated with diastolic Importantly, the effect sizes observed in the study were modest and
BP. This finding is novel and adds to the literature on RTs and not always consistent across predictors, and it may be difficult to
stress responses by showing that individuals with slower RT had interpret them before further replication is achieved. Another limi-
sustained cardiac and vascular activation (impaired recovery) fol- tation may be that our analyses were based on a subsample of par-
lowing mental stress. ticipants with available RT data, and therefore statistical power to
Tau-RT, which can be characterized as the values describing detect small associations might have been an issue. However, there
both the mean and the standard deviation of the exponential com- were no significant differences between participants with and with-
ponent (greater values reflect longer but infrequent response time), out RT data on several factors including CV responses to stress.
was only weakly associated with both systolic and diastolic BP Furthermore, we used time rather than frequency domain measures
reactivity (Figure 2a,b). In contrast, no association was observed of heart rate variability, and we may have had different results if
with heart rate and heart rate variability. However, this RT measure spectral analysis had been performed. Additionally, we did not
showed some interesting and novel associations with CV recovery. measure breathing patterns; respiratory rate can impact heart rate
Firstly, there was an inverse association with impaired recovery in variability, although parasympathetic modulation of heart rate not
blood pressure at 15 and, partially, at 75 min poststress (Figure related to respiratory rate have been used to index stress responses
3a,b), but not at 45 min. In addition, tau-RT was associated with (Houtveen, Rietveld, & de Geus, 2002).
impaired vagal control at 45 min (Figure 3c). These findings sug- Strengths of the study include the use of a mental stress protocol
gest that individuals characterized by impaired or slower intraindi- that included an adequate stress recovery period with continuous
vidual variability in RTs show sustained vagal and vascular CV assessment. This protocol allowed us to investigate associa-
activation following mild mental stress. Again, our results add to tions with stress recovery as well as reactivity that has seldom been
the literature by showing associations with stress recovery parame- examined in previous work (Ginty et al., 2011, 2012), but that is
ters, although these relationships appeared to be weak and not increasingly recognized as an important risk factor for CV morbid-
always consistent across time points. ity and mortality (Panaite et al., 2015).
As suggested in the introduction, the concept of neuronal In summary, we have offered some preliminary evidence that
efficiency may be used as a framework to interpret our results. RT measures are associated with CV responses to stress. Specifi-
Neuronal efficiency is an element of intrinsic plasticity, or non- cally, not only do average measures of RT relate to CV responses,
synaptic factors, that directly impacts the probability that a neu- but the variability in RTs, which can also be thought of as a proxy
ron will fire an action potential and bears implication for health measure for neuronal efficiency, may be predictive of CV reactiv-
outcomes including addiction (Kourrich, Calu, & Bonci, 2015; ity to, and recovery from, an acute stressor. Future studies, poten-
Zhang & Linden, 2003). In the field of psychophysiology, the tially using large, younger samples, are needed that implement
role of neuronal efficiency has been overlooked as a potential noninvasive measures to record RTs and examine their variability
mechanistic facet of the central nervous system response to in order to support the hypothesis that neuronal efficiency is related
mental stress. Low variability in neuronal firing (indexed by a to CV stress responses.
smaller tau-RT) can be seen phenotypically as being associated The clinical implication would be that, if this hypothesis is sup-
with higher reactivity to a stressor (seen with blood pressure and ported, it may potentially provide insight into alternative means of
Reaction times and stress responses 747
measuring CV dysfunction. Since persistent and exaggerated CV ical CV measurements to assess this dysfunction are typically cost
responses to acute stressors are detrimental to long-term health and prohibitive. Therefore, measuring neuronal efficiency through RTs
can foster disease progression, determining reactivity and recovery may serve as an early, low-effort, indirect measurement of CV
to an acute stressor can be clinically informative. However, preclin- reactivity.
References
Brage, S., Brage, N., Franks, P. W. Ekelund, U., & Wareham, N. J. (2005). stress, and in the field: Predictors of cardiac structure and function
Reliability and validity of the combined heart rate and movement sen- 2 years later in youth. Hypertension, 34, 1026–1031. doi: 10.1161/
sor Actiheart. European Journal of Clinical Nutrition, 59, 561–570. 01.HYP.34.5.1026
doi: 10.1038/sj.ejcn.1602118 Koehn, J. D., Dickinson, J., & Goodman, D. (2008). Cognitive demands of
Brydon, L., & Steptoe, A. (2005). Stress-induced increases in interleukin-6 error processing. Psychology Reports, 102, 532–538. doi: 10.2466/
and fibrinogen predict ambulatory blood pressure at 3-year follow-up. pr0.102.2.5322538
Journal of Hypertension, 23, 1001–1007. doi: 10.1097/01.hjh.00001 Koller, M., & Stahel, W. A. (2011). Sharpening Wald-type inference
66841.57474.d0 in robust regression for small samples. Computational Statistics
Chida, Y., & Steptoe, A. (2010). Greater cardiovascular responses to labo- and Data Analysis, 55, 2504–2515. doi: 10.1016/j.csda.2011.
ratory mental stress are associated with poor subsequent cardiovascular 02.014
risk status: A meta-analysis of prospective evidence. Hypertension, 55, Kourrich, S., Calu, D. J., & Bonci, A. (2015). Intrinsic plasticity: An emerg-
1026–1032. doi: 10.1161/HYPERTENSIONAHA.109.146621 ing player in addiction. Nature Reviews Neuroscience, 16, 173–184. doi:
Coles, M. G., Scheffers, M. K., & Holroyd, C. B. (2001). Why is there an 10.1038/nrn3877
ERN/Ne on correct trials? Response representations, stimulus-related Leth-Steensen, C., Elbaz, Z. K., & Douglas, V. I. (2000). Mean response
components, and the theory of error-processing. Biological Psychology, times, variability, and skew in the responding of ADHD children: A
56, 173–189. doi: 10.1016/S0301-0511(01)00076-X response time distributional approach. Acta Psychologica, 104,
Falkenstein, M., Hoormann, J., Christ, S., & Hohnsbein, J. (2000). ERP 167–190. doi: 10.1016/S0001-6918(00)00019-6
components on reaction errors and their functional significance: A tuto- Marmot, M. G., Smith, G. D., Stansfeld, S., Patel, C., North, F., Head, J.,
rial. Biological Psychology, 51, 87–107. doi: 10.1016/S0301-0511(99) . . . Feeney, A. (1991). Health inequalities among British civil servants:
00031-9 The Whitehall II study. Lancet, 337, 1387–1393. doi: 10.1016/0140-
Ginty, A. T., Gianaros, P. J., Derbyshire, S. W., Phillips, A. C., & Carroll, 6736(91)93068-K
D. (2013). Blunted cardiac stress reactivity relates to neural hypoactiva- McEwen, B. S., & Seeman, T. (1999). Protective and damaging effects of
tion. Psychophysiology, 50, 219–229. doi: 10.1111/psyp.12017 mediators of stress. Elaborating and testing the concepts of allostasis
Ginty, A. T., Phillips, A. C., Der, G., Deary, I. J., & Carroll, D. (2011a). and allostatic load. Annals of the New York Academy of Sciences, 896,
Cognitive ability and simple reaction time predict cardiac reactivity in 30–47. doi: 10.1111/j.1749-6632.1999.tb08103.x
the West of Scotland Twenty-07 Study. Psychophysiology, 48, 1022– Neubauer, A. C., & Fink, A. (2009). Intelligence and neural efficiency.
1027. doi: 10.1111/j.1469-8986.2010.01164.x Neuroscience and Biobehavioral Review, 33, 1004–1023. doi: 10.1016/
Ginty, A. T., Phillips, A. C., Der, G., Deary, I. J., & Carroll, D. (2011b). j.neubiorev.2009.04.001
Heart rate reactivity is associated with future cognitive ability and cog- Panaite, V., Salomon, K., Jin, A., & Rottenberg, J. (2015). Cardiovascular
nitive change in a large community sample. International Journal of recovery from psychological and physiological challenge and risk
Psychophysiology, 82, 167–174. doi: 10.1016/j.ijpsycho.2011.08.004 for adverse cardiovascular outcomes and all-cause mortality.
Ginty, A. T., Phillips, A. C., Roseboom, T. J., Carroll, D., & Derooij, S. R. Psychosomatic Medicine, 77, 215–226. doi: 10.1097/PSY.000000000
(2012). Cardiovascular and cortisol reactions to acute psychological 0000171
stress and cognitive ability in the Dutch Famine Birth Cohort Phillips, A. C., Ginty, A. T., & Hughes, B. M. (2013). The other side of the
Study. Psychophysiology, 49, 391–400. doi: 10.1111/j.1469-8986.2011. coin: Blunted cardiovascular and cortisol reactivity are associated with
01316.x negative health outcomes. International Journal of Psychophysiology,
Hagger-Johnson, G., Deary, I. J., Davies, C. A., Weiss, A., & Batty, G. D. 90, 1–7. doi: 10.1016/j.ijpsycho.2013.02.002
(2014). Reaction time and mortality from the major causes of death: R Core Team. (2015). R: A language and environment for statistical com-
The NHANES-III study. PLoS ONE, 9, e82959. doi: 10.1371/journal. puting. Vienna, Austria: R Foundation for Statistical Computing.
pone.0082959 Retrieved from https://www.R-project.org/
Hajcak, G., & Simons, R. F. (2008). Oops! . I did it again: An ERP and Rousseeuw, P. J., Croux, C., Todorov, V., Ruckstuhl, A., Salibian-Barrera,
behavioral study of double-errors. Brain Cognition, 68, 15–21. doi: M., Verbeke, T., . . . Maechler, M. (2013). Robustbase: Basic robust
10.1016/j.bandc.2008.02.118 statistics. R package version 0.9-10. Retrieved from http://CRAN.R-
Hamer, M., Endrighi, R., Venuraju, S. M., Lahiri, A., & Steptoe, A. project.org/package5robustbase
(2012). Cortisol responses to mental stress and the progression of coro- Schmiedek, F., Oberauer, K., Wilhelm, O., Suss, H. M., & Wittmann, W.
nary artery calcification in healthy men and women. PLoS ONE, 7, W. (2007). Individual differences in components of reaction time distri-
e31356. doi: 10.1371/journal.pone.0031356 butions and their relations to working memory and intelligence. Journal
Heathcote, A., Popiel, S. J., & Mewhort, D. J. (1991). Analysis of response of Experimental Psychology General, 136, 414–429. doi: 10.1037/
time distributions: An example using the Stroop task. Psychological 0096-3445.136.3.414
Bulletin, 109, 340–347. doi: 10.1037/0033-2909.109.2.340 Schwartz, A. R., Gerin, W., Davidson, K. W., Pickering, T. G., Brosschot,
Hervey, A. S., Epstein, J. N., Curry, J. F., Tonev, S., Arnold, L. E., J. F., Thayer, J. F. . . . Linden, W. (2003). Toward a causal model of
Conners, C. K., . . . Hechtman, L. (2006). Reaction time distribution cardiovascular responses to stress and the development of cardiovascu-
analysis of neuropsychological performance in an ADHD sample. Child lar disease. Psychosomatic Medicine, 65, 22–35. doi: 10.1097/
Neuropsychology, 12, 125–140. doi: 10.1080/09297040500499081 01.PSY.0000046075.79922.61
Houtveen, J. H., Rietveld, S., & de Geus, E. J. (2002). Contribution of tonic Sherwood, A., Davis, M. R., Dolan, C. A., & Light, K. C. (1992). Cardio-
vagal modulation of heart rate, central respiratory drive, respiratory vascular reactivity assessment: Effects of choice of difficulty on labora-
depth, and respiratory frequency to respiratory sinus arrhythmia during tory task responses. International Journal of Psychophysiology, 12,
mental stress and physical exercise. Psychophysiology, 39, 427–436. 87–94. doi: 10.1016/0167-8760(92)90046-E
doi: 10.1111/1469-8986.3940427 Silvia, P. J., Jones, H. C., Kelly, C. S., & Zibaie, A. (2011). Trait self-
Imholz, B. P., Wieling, W., van Montfrans, G. A., & Wesseling, K. H. focused attention, task difficulty, and effort-related cardiovascular reac-
(1998). Fifteen years experience with finger arterial pressure monitor- tivity. International Journal of Psychophysiology, 79, 335–340. doi:
ing: Assessment of the technology. Cardiovasc Research, 38, 605–616. 10.1016/j.ijpsycho.2010.11.009
doi: 10.1016/S0008-6363(98)00067-4 Steptoe, A., Hackett, R. A., Lazzarino, A. I., Bostock, S., La Marca, R.,
Jensen, A. R. (2006). Clocking the mind: Mental chronometry and individ- Carvalho, L. A., & Hamer, M. (2014). Disruption of multisystem
ual differences. Amsterdam, The Netherlands: Elsevier. responses to stress in Type 2 diabetes: Investigating the dynamics of
Kapuku, G. K., Treiber, F. A., Davis, H. C., Harshfield, G. A., Cook, B. B., allostatic load. Proceedings of the National Academy of Sciences
& Mensah, G. A. (1999). Hemodynamic function at rest, during acute U S A, 111, 15693–15698. doi: 10.1073/pnas.1410401111
748 A.J. Wawrzyniak et al.
Steptoe, A., & Marmot, M. (2002). The role of psychobiological pathways Treiber, F. A., Kamarck, T., Schneiderman, N., Sheffield, D., Kapuku, G.,
in socio-economic inequalities in cardiovascular disease risk. European & Taylor, T. (2003). Cardiovascular reactivity and development of
Heart Journal, 23, 13–25. doi: 10.1053/euhj.2001.2611 preclinical and clinical disease states. Psychosomatic Medicine, 65, 46–
Steptoe, A., & Marmot, M. (2006). Psychosocial, hemostatic, and inflam- 62. doi: 10.1097/00006842-200301000-00007
matory correlates of delayed poststress blood pressure recovery. Psy- Vaurio, R. G., Simmonds, D. J., & Mostofsky, S. H. (2009). Increased
chosomatic Medicine, 68, 531–537. doi: 10.1097/01.psy.0000227751. intra-individual reaction time variability in attention-deficit/hyperactiv-
82103.65 ity disorder across response inhibition tasks with different cognitive
Steptoe, A., Willemsen, G., Kunz-Ebrecht, S. R., & Owen, N. (2003). Soci- demands. Neuropsychologia, 47, 2389–2396. doi: 10.1016/j.neuropsy
oeconomic status and hemodynamic recovery from mental stress. Psy- chologia.2009.01.022
chophysiology, 40, 184–191. doi: 10.1111/1469-8986.00020 Villareal, R. P., Liu, B. C., & Massumi, A. (2002). Heart rate variability
Street, J. O., Carroll, R. J., & Ruppert, D. A. (1988). Note on computing and cardiovascular mortality. Current Atherosclerosis Reports, 4, 120–
robust regression estimates via iteratively reweighted least squares. 127. doi: 10.1007/s11883-002-0035-1
American Statistician, 42, 152–154. doi: 10.1080/00031305.1988. Zhang, W., & Linden, D. J. (2003). The other side of the engram:
10475548 Experience-driven changes in neuronal intrinsic excitability. Nature
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Reviews Neuroscience, 4, 885–900. doi: 10.1038/nrn1248
Journal of Experimental Psychology, 18, 643–662. doi: 10.1037/
h0054651 REVISED November 7, 2015; ACCEPTED November 29, 2015)