Data Analysis
The following statistics were also calculated with SPSS version 21. First, I will
examine if there is any difference in these indices between the BMAA group and the
control group before training. Second, the training effect will be examined through
one-way ANOVA with group (BMAA/control) as the between-subject factor and the
improvement ratio for each measurement as the dependent variables. Finally, to
examine the hypothesis about body-mind interaction, a series of multiple regression
was preformed to see if those who had better proprioception would benefit more from
BMAA training than those who did not. All results were considered significant at p
< .05, two-tailed.
The Baseline Difference between the BMAA Group and the Control Group
There was no significant difference between the BMAA group and the control
group in gender (p = .77) or in age (BMAA: M = 9.59 years old, SD = 0.75;
CONTROL: M = 9.38 years old, SD = 0.74, p = .33) at T1.
As shown in Table 6, there was also no significant difference across groups in
proprioceptive accuracy (p = .48), interoceptive accuracy (p = .56), interoceptive
awareness (p = .32), RTCV (p = .49) and working memory (p = .28) at T1. However,
there was a tendency that the BMAA group (M = .45, SD = .21) performed better than
the control (M = .35, SD = .17) on accuracy rate for the NOGO trials (NOGO ACC) in
SART, F(1, 46) = 3.12, p = .08. In contrast, the BMAA group scored worse on
proprioceptive awareness (M = 768.48, SD = 499.12) than the control group (M =
531.41, SD = 324.19) at T1 (F(1, 46) = 3.56, p = 0.07).
The Training Effects of BMAA
Next, in order to examine whether the BMAA group has significantly improved
more than the contrast group in bodily senses and cognitive abilities measured after
intervention, a one-way ANOVA was used to compared improvement ratios for each
of the measures across groups. The ratios were the difference scores (T2-T1) divided
by scores at T1.
As shown in Table 6, the BMAA group was significantly improve more in
working memory (ratio: M = .14, SD = .13) than the contrast group (M = .04, SD
= .18), F(1,46) = 4.88, p = .03. The BMAA group also had a tendency to improve
more for RTCV (ratio: M = -.06, SD = .27) than the contrast group (ratio: M = .11, SD
= .40), F(1,46) = 3.11, p = .08. In other words, there was a significant training effect
of BMAA in improving children participants’ working memory capacities and had a
tendency training effect on sustained attention, compared to the control group. No
group effects for other measures including proprioception (accuracy and awareness),
interoception (accuracy and awareness) and NOGO ACC was found (ps > .10).
Table 6
T1 scores, T2 scores and Improvement Ratio on Proprioception, Interoception, Sustained Attention and Working Memory Capacity for the BMAA Group (N = 27) and the Control Group (N = 21), respectively
Indexes Groups Time 1 M(SD) Time 2 M(SD) Improvement ratio Training effect (p-value)
SART - GO trial ACC BMAA 0.96 (0.07) 0.97 (0.03) 0.02 (0.09) .17
Proprioceptive Awareness BMAA 768.48 (499.12) 625.46 (366.62) 0.13 (0.86) .86 CONTROL 531.41 (324.19) 527.49 (402.89) 0.09 (0.66)
Interoceptive Accuracy BMAA 0.69 (0.15) 0.7 (0.19) 0.06 (0.44) .79
CONTROL 0.67 (0.17) 0.68 (0.21) 0.04 (0.26)
Interoceptive Awareness BMAA 802.71 (625.01) 1080.32 (1203.06) 1 (2.41) .47 CONTROL 639.35 (459.42) 691.59 (590.57) 0.55 (1.76)
Note: Improvement ratio: (T2 scores – T1 scores)/ T1 scores.
Who Would Benefit More from BMAA Training?
Based on the hypothesis and the features of BMAA training, I examine, with a
series of linear regression analysis, whether children with better proprioception would
benefit more on sustained attention than the worse ones from the training.
The regression models included T1 cognitive measures as controlled variable,
effect of proprioceptive accuracy at T1 (PACT1), effect of proprioceptive awareness at
T1 (PAWT1), and difference scores for a cognitive measures (working memory
capacity, RTCV and NOGO ACC for SART, respectively) were used as the dependent
variable.
In the terms regarding the effect of PACT1 and PAWT1, children were split into
two groups according to the median for each, and the better half was coded as +.5 and
the remains -.5, respectively, according to mean-center recommendations of Kraemer
& Blaséy (2004), The significance of these terms indicate that individual’s condition
of PAC and PAW at the beginning could influence how much they would improve on
cognitive functions through BMAA. A series of linear regression models was then
tested for each of the cognitive measures mentioned. The models were illustrated as
follows:
dRTCV = RTCVT1 + PACT1(H/L) + PAWT1(H/L)
dNOGO =NOGOT1 + PACT1(H/L) + PAWT1(H/L)
dWMC = WMCT1+ PACT1(H/L) + PAWT1(H/L)
As shown in Table 7, the overall model, the effect of PAC and PAW for NOGO
ACC were all significant (overall: p = .01; PAC: p = .02; PAW: p = .05). The models,
however, were not significant for RTCV (p = .17) and working memory (p = .38) for the BMAA group. Besides, for reader’s interest, not any effect of bodily sense was
found for the control group (ps > .10).
The above results indicate that children who came with better PAC improved
more on NOGO ACC through BMAA training than children with poorer PAC. As for
PAW, it showed an opposite pattern: Children with poorer PAW before training
benefited more on NOGO ACC via the training than children with better PAW.
Table 7
Multiple Regression Analyses Regarding the Effects of Proprioception at T1 on Sustained Attention and Working Memory, respectively
ScoreT1 PACT1 PAWT1 Model statistics
Groups Dependent Variables β t p β t p β t p R2 df F p
BMAA GO trials RTCV -.31 -1.59 .13 -.36 -1.83 .08 .12 .65 .52 .20 3, 23 1.86 .17 NOGO trial ACC -.16 -.92 .37 .43 2.58 .02 -.37 -2.06 .05 .39 3, 23 4.89 .01 Working memory -.12 -.62 .54 .03 .15 .88 .31 1.54 .14 .12 3, 23 1.07 .38 CONTROL GO trials RTCV -.40 -1.67 .11 -.33 -1.37 .19 .09 .40 .69 .19 3, 17 1.30 .31 NOGO trial ACC -.44 -2.01 .06 .37 1.70 .11 .05 .26 .80 .25 3, 17 1,93 .16 Working memory -.46 -1.94 .07 -.28 -1.20 .25 .22 .99 .34 .21 3, 17 1.53 .24 Note: RTCV is short for reaction time coefficient of variability.
Discussion II
In sum, as predicted, BMAA-C is effective on improving children’s working
memory capacities and sustained attention indicated by the decreased variability of
response speed (lower RTCV). However, contrary to my prediction, no training effect
on bodily senses, particularly on bodily awareness, was found. However, for the
BMAA group, children who scored better than the median on proprioceptive accuracy
before training improved more for NOGO accuracy, another index for sustained
attention, than the worse. Similarly, children with poorer proprioceptive awareness
before training also improved more for NOGO accuracy than the counterpart.
The finding that the training effect of BMAA-C on working memory capacity
(WMC) is in line with previous findings for adults showing that WMC is the most
reliable effect of MBCPs in general (e.g. Gothe et al., 2013; Taylor-Piliae et al., 2010)
and BMAA in specific (Teng & Lien, 2016). There are several plausible explanations why BMAA or, more general, MBCPs could facilitate practitioners’ working memory
capacities. First, some might argue that BMAA practice is kind of a multi-tasking
practice which exerted on working memory. Note that practitioners have to monitor
several places of their bodies (such as keeping the eyes downwards, doing some kinds
of movement, and keeping some body parts relaxed at the same time) to meet the
principle. Therefore, working memory capacities could be enhanced by repeatedly
exercise. If so, the practitioners would be exhausted after the training, which is not the
case as I observed.
Second, as BMAA movements are mostly practiced while eyes are closed, it is
likely that proprioception is heavily involved in these practices, and as Alloway and
Alloway (2015) suggests, proprioception-demanding trainings could benefit trainees’
working memory capacities. However, my data does not support this explanation since my participants’ proprioceptive accuracy was not improved as their working
memory capacities were.
Third, the BMAA practice might help our participants to learn an effective way to quiet or void one’s mind, which could further enhance their cognitive capacities in
two ways. They can either restore their attention more efficiently or/and do a
challenging task more concentratedly as the occurrence of emotion and mind
wandering are both gradually reduced. Note that this explanation would predict
attention restoration rather than exhaustion. More studies are required to test which
one is the case.
My results also showed, for the first time, that BMAA could improve children’s
performance on sustained attention by showing a more stable response speed than the
control group. In addition, unlike the previous studies about other MBCPs such as
yoga (e.g. Rangan et al., 2009), I use an objective behavior measure for sustained
attention (i.e., SART) rather than pencil-and-paper tests. This finding also adds a
piece of supporting evidence, with a different objective measure, to the positive effect
of MBCPs on sustained attention.
As for NOGO ACC, another index of SART, no training effect for the BMAA
group was found compared to the control group. However, further analysis shows an
interesting result that children with better proprioceptive accuracy benefit more on
NOGO ACC from the training than those who had poorer accuracy in pretest, which
is partially support my fourth hypothesis concerning about whether bodily senses
moderate the training effect of BMAA on executive functions. It is probably because
that the BMAA practices were heavily exerted on proprioceptive ability so that
children with poor proprioceptive accuracy need more time to get the benefit from the
training.
In contrast, children with poorer proprioceptive awareness (PAW) improved
NOGO ACC more than those who had better PAW in the pretest. Since PAW was
significantly correlated with NOGO ACC in pretest, I wonder whether children with
poorer PAW had more room to improve on PAW, and led to the improvement on
sustained attention indicated by NOGO ACC. To test this hypothesis, I examined
whether the differential score (T2-T1) of NOGO ACC and PAW correlated with each
other. However, there is no significant correlation between them in the BMAA group
(r = -.12, p = .55). To further know whether there is a contingency between the
improvement on PAW and NOGO ACC, children in the BMAA group were further
separated into four groups according to whether they improved on PAW and NOGO
ACC: children were categorized as PAW-improved group if their PAW scores at T2
was lower than that at T1 (remember that for PAW, the lower the better), and
PAW-not-improved group, otherwise. Similarly, they were labelled as NOGO
ACC-improved group if their NOGO ACC scores at T2 higher than that of T1, and NOGO
ACC-not-improved group, otherwise. I found that higher proportion of the low-PAW
groups had improvement on PAW than the high-PAW group, Fisher's Exact Test, p
= .04 (one-tailed). Furthermore, proportion of the PAW-improved group improved on
NOGO ACC was higher, marginally significant, than that of the PAW-not-improved
group, Fisher's Exact Test, p = .08 (one-tailed). As for the control group, no such
trend was found.
Contrary to my prediction, no training effect was found for bodily awareness
including interoception and proprioception although during BMAA-C program.
However, although the control group outperformed the BMAA group on PAW at T1
with a marginal significance, the BMAA group caught up with the control group after
training. Further studies should examine this issue more carefully with well-matched
participants and perhaps with a longer training time for children.