Results
Principal characteristics of the two groups of participants were
compared to check whether they were homogenous or not (see Table 1).
ANOVA with Group (Experimental vs. Control) as the between-subjects
factor was performed on age, body mass index (BMI), trait self-control,
maximal voluntary contraction (MVC), and average time to exhaustion
(TTE) on the handgrip task performed at the beginning of the session
(TTE-1). Table 1 shows that the two groups significantly differed in BMI
and TTE-1 only. The control group had significantly lower BMI and TTE-1
than the experimental group. In addition, BMI and TTE1 significantly
correlated (r = .330; p = .033). Consequently, BMI was used as a
covariate in ANOVA leading to a significant effect and including the
factor Group as a between-subjects factor. In the same way, the
difference between TTE-1 and TTE-2 was used as the main outcome to test
the ego-depletion effect, instead of TTE-2 only. The chi-square
statistic on the percentage of men and women showed no significant
difference between the groups.
———————————–
Insert Table 1 about here
———————————–
Maximal voluntary
contraction
ANOVA with Moment (before the mental task vs. after the mental task) and
Repetition (HG1 vs. HG2) as within-subjects factors and Group
(Experimental vs. Control) as a between-subjects factor was performed on
MVC performance. The results of this test showed that the interaction
Moment X Repetition X Group did not reach significance: F (1, 30)
= .004, p = .946, η ²p = .000. In the same
way, the interactions Moment X Repetition, Moment X Group and Repetition
X Group were not significant: F (1, 30) = 1.791, p = .191,η ²p = .056; F (1, 30) = .037, p =
.849, η ²p = .001; F (1, 30) = 2.925,p = .098, η ²p = .089, respectively. Simple
main effect analysis showed that the effect of Moment was significant:F (1, 30) = 83.254, p < .001,η ²p = .735. The mean value of MVC was
significantly lower after the time-to-exhaustion handgrip task (M= 13.3 kg, SE = .738) than before it (M = 17.5 kg, SE =
.862). In addition, we observed a significant effect of Repetition:F (1, 30) = 4.948, p = .034, η ²p =
.142. The mean value of MVC was significantly higher during the first
handgrip task (M = 15.7 kg, SE = .826) compared to the
second handgrip task (M = 15.0 kg, SE = .745). On average,
the participants’ MVC was reduced by 23.33% after the
time-to-exhaustion handgrip task. This decrease in MVC reflects a lower
capacity to produce a muscular force. Finally, we did not observe a
significant effect of Group: F (1, 30) = .321, p = .575,η ²p = .011.
Ego depletion effect
Since the initial performance of
the two groups in the handgrip task was not homogeneous, the
time-to-exhaustion (TTE) was calculated separately for each subject as
follows: TTE Delta (min) = time-to-exhaustion in the second handgrip
task minus time-to-exhaustion in the first handgrip task. Finally, to
highlight the ego-depletion effect, the TTE delta of the two groups was
compared with the two-sample t test.
The results indicated that the TTE delta was significantly lower for the
experimental group (M = -3.45 min, SD = 3.17) than for the
control group (M = -1.47 min, SD = 1.31): t (30) =
2.31, p = .028, d = 0.818. However, the results of the
Shapiro-Wilk test for the TTE delta also showed a violation of the
assumption of normality: W = .933, p = .046. Therefore, we
used the nonparametric Mann-Whitney U test to avoid any violation of
normality, and the results showed a significant difference as well:U (NVideo = 16, NStroop = 16) =
71, p = .032, r = .455; (see Figure 3). The decrease in
performance in the experimental group (M = 33.5%, SD =
30.2%) was larger than the decrease in performance observed in the
control group (M = 18.2%, SD = 25.2%).
Because the replication of the ego depletion effect is the main
objective of this paper, we also conducted ANCOVA with Group as a
between-subjects factor and BMI as a covariate. In this case, the effect
of Group no longer reached the significance level: F (1, 29) =
3.939, p = .057, η²p = .120. We also
performed the analyses while excluding participants who had a decrease
or an increase more than 2 standard deviations from the mean. One
participant was removed (point 26 on Figure 4). The results of the t
test without the outlier showed a significantly lower TTE delta for the
experimental group (M = -2.92 min, SD = 2.44) compared to
the control group (M = -1.47 min, SD = 1.31): t(29) = 2.08, p = .046, d = 0.749. While conducting ANCOVA
with BMI as a covariate and removing the outlier, the effect of Group
reached significance: F (1, 28) = 4.464, p = .044,η²p = .137.
———————————–
Insert Figure 3 about here
———————————–
Perceptions of effort and muscle pain during the handgrip
endurance
task
We conducted ANOVA with Group (Experimental vs. Control) as the
between-subjects factor and Time of measurement (before the mental task
vs. after the mental task) and Individual isotime (0%, 33%, 66% and
100%) as within-subjects factors on the subjective effort perception
during the two time-to-exhaustion handgrip tasks. When testing an effect
involving individual isotime, we applied a Greenhouse-Geisser correction
to consider any violation of the sphericity assumption. The interaction
Group X Time of measurement X Individual isotime did not reach
significance: F (2.11, 63.30) = 1.364, p = .263,η 2p = .043. In the same way,
the interactions Group X Time of measurement and Group X Individual
isotime did not reach significance: F (1, 30) = 2.081, p =
.160, η 2p = .065) and F(2.21, 66.41) = .887; p = .426,η 2p = .029, respectively. In
contrast, the interaction between Time of measurement and Individual
isotime reached significance: F (2.11, 63.30) = 11.825, p< .001, η 2p = .283
(see Figure 4). The perception of effort increased more sharply during
the second time-to-exhaustion task (TTE-2) than during the first
time-to-exhaustion task (TTE-1). The effect of Group was not
significant: F (1, 30) = .321, p = .575,η 2p = .011.
The same ANOVA was conducted on the subjective perception of muscle pain
during the two time-to-exhaustion handgrip tasks. The second-order
interaction between among Group, Time of measurement and Individual
isotime did not reach significance: F (1.696, 50.869) = .793,p = .439, η 2p = .026. In
the same way, the interactions Group X Time of measurement and Group X
Individual isotime did not reach significance: F (1, 30) = 1.195,p = .283, η 2p = .038 andF (2.053, 61.601) = .303, p = .745,η 2p = .010, respectively. In
contrast, the interaction between Time of measurement and Individual
isotime reached significance: F (1.696, 50.8869) = 6.302,p = .005, η 2p = .174
(see Figure 4). The perception of pain increased more sharply during the
second time-to-exhaustion task. Finally, the effect of Group was not
significant: F (1, 30) = 1.54, p = .224,η 2p = .049.
———————————–
Insert Figure 4 about here
———————————–
Performance in the mental
tasks
To assess the performance in the Stroop task, two separate ANOVA
with Time on task (TOT; T1, T2,
T3 & T4) and Type of trial (reading vs. naming ink color) as
within-subjects factors were conducted on the mean reaction time (RT)
and the error rate. A Greenhouse-Geisser correction was applied to the
degrees of freedom to avoid any problem of sphericity. The interaction
Time on task X Type of trial reached significance for mean RT (see
Figure 5): F (2.282, 34.227) = 4.881; p < .05;η 2p = .245. This interaction is
explained by the difference between ‘reading’ trials and ‘naming ink
color’ trials that increased with time on task. The interaction Time on
task X Type of trial and the simple effect of Time on task did not reach
significance for error rate: F (2.344, 35.154) = 0.125; p = .910;η 2p = .000 and F (1.024,
15.367) = 1.102; p = .312; η 2p= .053, respectively.
In addition, the simple effect of Type of trial is significant for mean
RT and error rate: F (1, 15) = 155.904; p < .0001;η 2p = .912 and F (1, 15) =
156.565; p < .0001;η 2p = .913, respectively. As
expected, reading trials leaded to shorter mean RT (629.44 ms vs. 741.19
ms) and smaller error rate
(6.66% vs. 15.42%) than ‘naming
ink color’ trials.
———————————–
Insert Figure 5 about here
———————————–
To verify whether the participants paid attention to the content of the
documentary, the percentage of their correct responses was compared to
the level of chance (i.e., 50%) via the one-sample t test. According to
the results, the participants answered significantly better than chance
(M = 78.1%, SD = 9.11%): t (16) = 12.4, p< .001, d = 3.09.
Subjective measurements using visual analog
scales
Subjective feeling of
fatigue
ANOVA with Group (Experimental vs. Control) as the between-subjects
factor and Time of measurement
(T1 = Baseline, T2 = Pre-Mental
Task, T3 = Post-Mental Task, T4 = Post-Handgrip Task) as the
within-subjects factor was performed on the subjective feeling of
fatigue. We applied a Greenhouse-Geisser correction to consider any
violation of the sphericity assumption. The results showed that the
interaction between Group and Time of measurement did not reach
significance: F (2.774, 83.212) = 2.24, p = .095,η 2p = .069. In the same way,
the effect of Group was not significant: F (1, 30) = 2.629,p = .115, η 2p = .081. In
contrast, the effect of Time of measurement reached significance:F (2.774, 83.212) = 22.299, p < .001,η 2p = .426. According to a
planned comparison conducted to explain the effect of time of
measurement, we observed a significant, linear increase in fatigue level
as time progressed, regardless of the first Mental task: t (90) =
7.637, p < .001.
Motivation
Repeated-measures ANOVA with Group (Experimental vs. Control) as a
between-subjects factor and Time of measurement (T1 = Pre-Mental Task,
T2 = Post-Mental Task) as a within-subjects factor was conducted on the
motivation to perform the time-to exhaustion handgrip task. According to
the results, the interaction between Group and Time of measurement did
not reach significance: F (1, 30) = 0.030, p = .862,η 2p = .001. The effects of Time
of measurement and Group did not reach significance either: F(2.57, 77.12) = 3.216, p = .083,η 2p = .097 and F (1, 30)
= 2.14, p = .154, η 2p =
.067, respectively.
Task difficulty
The results of the two-sample t test demonstrated that the participants
in the experimental group perceived the Stroop task (M = 74.9,SD = 19.7) to be more difficult than the control group did for
the documentary viewing task (M = 20.1, SD = 22.2):t (30) = − 7.39, p < .001, d = -2.61.
Boredom
The two-sample t test was performed to compare perceived boredom during
the depleting versus control tasks. The effect of Group (Experimental
vs. Control) did not reach significance: t (39) = − 1.11,p = .274, d = -0.394. Unexpectedly, the Stroop task
performed by the experimental group was not perceived as boring more
than the documentary viewing task performed by the control group.
EEG indices
Power spectral analysis
We used the unpaired t test with FDR correction for multiple comparisons
to compare the theta wave band power spectral density during the Stroop
task compared to the documentary viewing task. A statistically
significant higher theta power spectral density was observed mainly in
the frontal, prefrontal and central areas during the Stroop task
compared to the documentary viewing task, more specifically in
the AF7, AF3, FC5, FC3, AF4, AF8,
F8, F6, FC6, C6, C4, C2 and P4 electrodes. However, we also detected
significantly higher theta power during the documentary viewing task,
mainly in occipital areas, more precisely the
POz, O1, Iz, Oz, PO7 and PO8
electrodes (see Figure 6).
———————————–
Insert Figure 6 about here
———————————–
Later, to assess the effect of time on task on the stimulus-locked theta
power density recorded during the modified Stroop task, the EEG
recording was divided into 4 equal time periods of 7 min and 30 s. Then,
we performed a repeated-measures ANOVA with Time on task (T1, T2, T3 &
T4) and Type of trial (reading vs naming ink color) as within-subjects
factors on the 64 electrodes. The results showed that the effect of Time
on task on the stimulus-locked theta power density reached significance
only for 4 electrodes: FC1, P2, CP2, P4. The effect is identical for the
4 electrodes: stimulus-locked theta power decreased with time on task.
We reported hereafter the results of the ANOVA only for FC1: F(2.684, 40.258) = 11.008, p < .001,η 2p = .423 (see Figure 7).
The effect of Type of trial did
not reach significance for the 4 electrodes. For the electrode FC1, we
obtained the following results: F (1, 15) = 0.004, p =
.952, η 2p = .000.
———————————–
Insert Figure 7 about here
———————————–
Source localization
analysis
A K-means clustering (K = 33) algorithm using EEGLAB Study Statistics
was performed only on the components that represented higher theta power
during the Stroop task in the same regions, as previously observed based
on the results of our power spectral analysis. To identify the best
clusters, first, the outlier clusters recognized automatically by EEGLAB
were excluded, and then the clusters including sufficient numbers of
participants and components (i.e., at least 8 participants representing
more than 50% of the data) were considered. From the results of kmean
clustering, 6 of 35 clusters were identified and contained more than
50% of the data. As shown in Figure 9, the sources of cerebral
activation originated from the prefrontal, frontal and central areas,
more precisely regions close to the ACC, thalamus and posterior
cingulate cortex.
———————————–
Insert Figure 8 about here
———————————–
ECG indices
We conducted 3 separate repeated-measures ANOVAs with Time on task (T1,
T2, T3 and T4) as the within-subjects factor and Group as the
between-subjects factor on the HRV parameters (SDNN, HF & LF). When
testing an effect involving Time on task as a repeated-measure factor,
we applied a Greenhouse-Geisser correction to consider any violation of
the sphericity assumption.
The results for the SDNN index showed a significant interaction of Time
on task X Group: F (2.073, 60.130) = 3.889, p = .025,ƞ 2p = .118 (see Figure 9, panel
A). This interaction disappeared when conducting ANCOVA with BMI as a
covariate: F (2.09, 58.62) = 2.198; p = 0.118;ƞ 2p = .073.
For the HF, the interaction of Time on task X Group and the simple
effect of Group did not reach significance: F(2.746, 79.639) =.589, p =
.609, ƞ 2p = .020 and F(1 29) = .516, p = .478,ƞ 2p = .017, respectively.
However, the effect of Time on task was significant: F (2.746,
79.639) = 2.964, p =.041,ƞ 2p = .093. HF increased
throughout the mental task, regardless of the task.
Finally, concerning the results for LF, the interaction of Time on task
X Group reached significance: F (2.477, 70.959) = 5.104, p= .005, ƞ 2p = .150 (see Figure
9, panel B). This interaction was still significant when conducting
ANOVA with BMI as a covariate: F (2.58, 72.24) = 3.40; p =
.028; ƞ 2p = .108. A breakdown
of this interaction showed that LF for the control group decreased,
whereas LF for the experimental group increased between T1 and T2:F (1, 29) = 5.915; p = .021;ƞ 2p = .169.
———————————–
Insert Figure 9 about here
———————————–