Data Analysis
In order to examine if there is any relationship between body (bodily senses) and
mind (sustained attention and working memory), the correlational analysis was done
for the pre-test measures. The following statistics were calculated with SPSS version
21. Before examining the main concern about the correlations between each of the
bodily senses (interoception and proprioception) and cognitive performance (working memory, RTCV and NOGO ACC) will be examined, the participants’ performance on
each of the three tasks with respect to gender and their ages will be reported first, and
then the inter-domain correlations between the interoception and proprioception. All
results were considered significant at p < .05, two-tailed.
Children’s Performances of Interoception (Heartbeat Detection Task)
As shown in Table 4, the mean score of interoceptive accuracy for all the
participants is 0.68 (SD = 0.16) and that of interoceptive awareness is 729.66 (SD =
578.40). There is an unexpected significant correlation between interoceptive
accuracy and interoceptive awareness (r = .43, p < .01). Note that for the indices of
awareness used in this study, smaller numbers represent better awareness (i.e., less
incongruence between subjective confidence ratings and objective-measured
accuracy) than the larger ones. The above finding thus indicates children better at
counting heartbeat, nevertheless, do worse on judging how good their counts were.
No gender difference is found for either interceptive accuracy or awareness
(Accuracy: F(1,56) = .01, p = .91; Awareness: F(1,56) = 1.97, p = .17). In addition,
neither one is significantly associate to age (Accuracy: r = -.02, p = .89; Awareness: r
= .11, p = .41).
Children’s Performances of Proprioception (Passive Movement Task)
For all participants, the mean score is 0.81 (SD = 0.06) for proprioceptive
accuracy and 663.80 (SD = 431.49) for proprioceptive awareness. Unlike
interoception, there is no correlation between proprioceptive accuracy and awareness
(r = -.06, p = .66). For each of the index, neither gender difference (Accuracy: F(1,56)
= .68, p = .41; Awareness: F(1,56) = .06, p = .80), nor the correlation to age
(Accuracy: r = .11, p = .43; Awareness: r = .11, p = .42) was found.
Children’s Performance of Sustained Attention (SART)
For SART, the measurement of sustained attention, the mean accuracy of NOGO
trials (NOGO ACC) is 0.40 (SD = 0.20) and the mean Reaction Time Coefficient of
Variability of GO trials (RTCV) for GO trials is 0.39 (SD = 0.12). There was a
significant difference in RTCV between male (M = .43, SD = .12) and female (M
= .35, SD = .11), F(1,56) = 6.26, p < .05. Children’s RTCVs were also significantly
correlated with age, r = -.39, p < .01. As for NOGO ACC, neither a gender difference,
F(1,56) = 2.55, p = .11, nor a significant correlation with age, r = .19, p = .15, was
found.
Children’s Performance of Working Memory (OSPAN)
The working memory capacity was measured by the total number of OSPAN. The
mean score was 38.34 (SD = 6.40). There was no significant gender difference
F(1,56) = .63, p = .43 but it correlated with ages significantly, r = .29, p = .03.
Table 4
Descriptive Statistics for Children’s Performance on Proprioception, Interoception, Sustained Attention and Working Memory Capacity at T1 (N=58)
Indexes Mean SD Min Max
Proprioceptive Accuracy 0.81 0.06 0.66 0.92
Proprioceptive Awareness 663.80 431.49 63.60 1771.14
Interoceptive Accuracy 0.68 0.16 0.27 0.96
Interoceptive Awareness 729.66 578.40 36.29 2508.74
GO trial RTCV 0.39 0.12 0.17 0.71
NOGO trial ACC 0.40 0.20 0.04 0.96
Working memory capacity 38.34 6.40 21.00 49.00
Note: RTCV is short for reaction time coefficient of variability.
Correlation between Interoception and Proprioception
The inter-domain correlations between interoception and proprioception for
accuracy and awareness were also examined, respectively. There was a significant
positive correlation (r = .30, p < .05) between proprioceptive and interoceptive
awareness but no any relation was found in accuracy across domains (r = -.13, p
= .34)
Correlations between Bodily Senses and Sustained Attention
As predicted, children’s proprioceptive awareness is positively correlated with their performance on sustain attention. Specifically speaking, children’s scores of
proprioceptive awareness was positively correlated with RTCV (r = .30, p < .05) and
negatively correlated with NOGO accuracy (r = -.28, p < .05). In other words,
children who have better proprioception awareness tend to have higher accuracy and
respond in a more stable way in the sustain attention task, both of which indicating
fewer mind wandering and better sustained attention, than those have worse
awareness. In contrast, there is no significant correlation between interoceptive
awareness and measures of SART (RTCV: r = -.14, p = .29; NOGO ACC: r =.07, p
= .59).
In addition, it shows that proprioceptive accuracy has a significant correlation
with RTCV (r = -.37, p < .01) and a no-significant-trend in relating to NOGO ACC (r
= .24, p = .07). That is to say that children with higher proprioceptive accuracy
showed more consistent speed of responding, and thus have fewer mind wandering,
than those with lower accuracy. However, no such correlation was found between
interoceptive accuracy and sustained attention (Accuracy: r = .00, p = 1.00;
Awareness: r = .05, p = .71).
On the other hand, different from the results above about sustained attention, no
any indexes of bodily senses (interoception or proprioception) correlated with
working memory, neither awareness nor accuracy (ps > .10). All the above results are
shown in Table 5.
Did Interoceptive Awareness Play the Other Role than the Predictor for
Sustained Attention?
In the previous analysis, the direct relation between awareness and sustained
attention was found only in proprioception but not interoception. However,
interoceptive awareness associate with proprioceptive awareness significantly. It is
possible that interoceptive awareness may have effect on the relation between
proprioceptive awareness and sustained attention, so two separate moderation
analyses were carried out for RTCV and NOGO ACC. In all the regression analysis,
the proprioceptive awareness and interoceptive awareness were all mean centered (to
reduce the possibility of multicollinearity among the interaction terms and their
component predictors) and the interaction term between them were computed (Aiken
& West, 1991). The two predictors and the interaction term were simultaneously
entered into regression model. At first, RTCV was analyzed and it showed that the overall regression reached significance (𝑅2 = .16; 𝑅𝑎𝑑𝑗𝑢𝑠𝑡𝑒𝑑 = .11; F(3,54) = 3.34, p
= .03) and proprioceptive awareness (β = .35, p = .01) was significantly predictive for
RTCV which was in line with our relational results that mentioned above. In addition,
the interaction term had a trend toward significance (β = -.23, p = .09) while
interoceptive awareness was not significantly predictive for RTCV (β = -.08, p = .56).
The slopes were illustrated in Figure 5.
In other words, we can see from Figure 5 that the relation between proprioceptive
awareness and RTCV was more obvious in the children who have better interoceptive
awareness. For the children who have lower interoceptive awareness, this kind of relation is not clear. That is, although interoceptive awareness didn’t directly associate
with RTCV, the role it played was the moderator for the effect of proprioceptive
awareness on RTCV.
As for the NOGO ACC, overall regression did not reach significance (𝑅2 = .10;
𝑅𝑎𝑑𝑗𝑢𝑠𝑡𝑒𝑑 = .05; F(3,54) = 2.01, p = .12). Identical with previous relational results,
only proprioceptive awareness was significantly predictive for NOGO ACC (β = -.30,
p = .03). Neither the predictiveness of interoceptive awareness nor the interaction
term was significant or has the trend toward significance. (interoceptive awareness: β
= -.02, p = 91; interaction: β = .15, p = .27)
Table 5
Inter-correlation Matrix among the Scores on Proprioception, Interoception, Sustained Attention and Working Memory Capacity at T1 (N=58) Proprioceptive
Proprioceptive Awareness -.06 1
Interoceptive Accuracy -.13 -.03 1
Interoceptive Awareness -.07 .30* .43** 1
GO trial RTCV -.37** .30* .00 -.14 1
NOGO trial ACC .24∆ -.28* .05 .07 -.66** 1
Working memory capacity -.03 -.04 -.003 .14 -.20 .17 1
Note: RTCV is short for reaction time coefficient of variability.
∆ p<.10. * p<.05. ** p<.01.
Figure 5. The scores of RTCV (reaction time coefficient of variability) for children with high/low interoceptive awareness (IAW) and high/low proprioceptive awareness.
Discussion I
In summary, there are three main findings about the correlational analysis on the pretest scores. First, children’s abilities regarding proprioception are related to their
performance on sustained attention. Specifically speaking, children better at
proprioceptive awareness respond to GO trials with a more constant speed (i.e.,
smaller RTCVs) and have higher accuracy rates for NOGO trials (NOGO ACC) in
SART than those worse. In addition, children better at proprioceptive accuracy also
tend to be better at RTCV. Remind that high stability of reaction time for Go trials and
high accuracy rate for NOGO trials both represent better control of attention.
Although practicing body awareness or somatic attention has been thought to be
crucial for the positive training effect of mindfulness training on attention, my finding, for the first time, demonstrates that children’s body awareness, i.e.,
proprioceptive awareness, is associated to their attention control.
Second, although no correlation between sustained attention and interoception is
found, children’s interoceptive awareness is nevertheless significantly correlated with
proprioceptive awareness, and moderates the relation between proprioceptive
awareness and RTCV. That is, the positive correlation between proprioceptive
awareness and RTCV is statistically significant only for children with high
interoceptive awareness but not for those with low interoceptive awareness.
Third, unlike sustained attention, children’s working memory capacities are
independent to their body awareness. It may be not too surprising if we compare the
natures of the tasks measuring sustained attention, SART, and working memory
capacity, OSPAN. Unlike SART, a working memory capacity task is usually more
challenging and demanding for it intends to measure a person’s maximal cognitive
resources available at a time. It is known that a person is usually more alerted and
tend to have fewer mind wanderings when doing a high-demanding task compared to
a low-demanding one like SART (e.g. Rummel & Boywitt, 2014). Therefore, good
performance on SART requires more attention monitoring and could be more sensitive to one’s meta-awareness than a challenging one like OSPAN. Interestingly,
improving on working memory capacity is the most-frequently-reported effect of
mindfulness training on cognitive functions (Teng & Lien, 2016). The above result
might indicate that this improvement is not through the enhancing body awareness. In
next study, I would further test this conjecture.
In addition to the aspect of awareness, children’s proprioceptive abilities, i.e.,
how accurate they know where they bodies are, also predict their performance on the
sustained attention task (i.e., reducing RTCV in SART). A plausible account is that
children with better proprioception require fewer attention resources to keep their
motor or postural control (such as sitting or standing) during an attention task, so that less interfere children’s performance on attention-related tasks. It is known that
proprioception plays an important role at motor or postural control (e.g. Marsh &
Geel, 2000; Riemann & Lephart, 2002; Teasdale & Simoneau, 2001). In addition, for
the elderly, who has reduced attention resources, sensory integration required for
postural control is also known to depend on their ability of attention (e.g. Redfern,
Jennings, Martin, & Furman, 2001). Therefore, for children who also has very limited
capacity of attention, being endowed with better proprioception may leave more
attentional resource for the attentional task at hand. However, this account would also predict that children’s performance on working memory capacity would be influenced
by their proprioceptive accuracy for the same reason. As the results show, this is not
true. Another possibility is proprioception and attention might involve some common
cognitive process or brain areas, which requires further investigation.
Some other might argue that children with higher proprioception may trade
speed for accuracy. Therefore, it is nothing to do with the so-called body-mind
interaction. Previous studies have shown that there is speed-accuracy trade-off
between GO trials reaction time and NOGO accuracy (e.g. Seli, 2016). My data also
showed that these two indices are correlated to each other (r = .60, p < .001).
However, children’s proprioception does not correlate to the reaction time of GO
trials (accuracy: r = -.03, p = .84; awareness: r = -.19, p = .18). So the trading-off
account cannot explain the relationship between NOGO accuracy and proprioception.
Unexpectedly, instead of directly associating to sustained attention,
interoceptive awareness is positively associated with proprioceptive awareness and
moderates the relation between proprioceptive awareness and sustained attention. The
moderating effect of interoception says that children’s proprioceptive awareness can
predict their sustained attention better when they are relatively good than poor at
interoception, which indicate an indirect influence of interoception on sustain
attention. It is likely that these two kinds of body awareness may come from a general
kind of meta-awareness and jointly influence the attention monitoring. Further
investigation is required to know more about the relationship among the three.
As previously mentioned, the aspects of accuracy and awareness of
interoception were usually found to be independent to each other for adults (Garfinkel
et al., 2015; Meessen et al., 2016). However, my data show that children who score
higher on interoceptive accuracy are worse on awareness than those who score lower
on accuracy. To make the picture clearer, I further split participants into high- and
low-accuracy groups along the median score of interoceptive accuracy. I found that
the correlation no longer exists for the low-accuracy group. That is, only in the
high-accuracy group (the average rate is 0.81), children who scored higher in high-accuracy tend
to have larger gap between their confidence judges and accuracy than those who with
relative lower scores in accuracy (0.55). As the data show, the confidence of high
accuracy group (the average confidence rating is 4.84) did not high enough to match
their accuracy performance, compared to that of the low accuracy group (4.21). It
may be easier for children at 8-11 years old to know that they have messed thing up
than that they have done a good job in the heartbeat counting task.