Meditation, mindfulness and cognitive flexibility
Author: Adam Moore, Peter Malinowski Publisher: https://web.colby.edu/cogblog/files/2014/05/Moore-Malinowski-2009-.pdf Publish Date: 2009-1-31 Review Date: Status:💥
Annotations
1
a
bstrac
1
This study investigated the link between meditation, self-reported mindfulness and cogni-tive flexibility as well as other attentional functions. It compared a group of meditators
experienced in mindfulness meditation with a meditation-naïve control group on mea-sures of Stroop interference and the ‘‘d2-concentration and endurance test”. Overall the
results suggest that attentional performance and cognitive flexibility are positively related
to meditation practice and levels of mindfulness. Meditators performed significantly better
than non-meditators on all measures of attention. Furthermore, self-reported mindfulness
was higher in meditators than non-meditators and correlations with all attention measures
were of moderate to high strength. This pattern of results suggests that mindfulness is inti-mately linked to improvements of attentional functions and cognitive flexibility. The rele-vance of these findings for mental balance and well-being are discussed.
1
Introduction
2
Cognitive flexibility is here understood as the human ability to adapt cognitive pro-cessing strategies to face new and unexpected conditions and is intrinsically linked to attentional processes (Cañas, Quesada,
Antolí, & Fajardo, 2003). As mindfulness meditation is dependant on the (re-)investment of attention on a moment by mo-ment basis, mindfulness training should hypothetically lead to increased cognitive flexibility and an increased ability to re-spond in a non-habitual fashion.
2
In this study we are employing two well-established measures, the Stroop task (Stroop, 1935) and the d2-test of attention
(Brickenkamp & Zilmer, 1998), to test participants’ ability to suppress interfering information and to focus and direct their
attention.
2
Following these ideas, in our study we are comparing a group of meditators with experience in mindfulness meditation
with a group of non-meditators on several tests of attention and assess them in a quiet experimental situation, but without
inducing a meditative state or state of mindfulness. Furthermore, for estimating their levels of mindfulness, they are required
to complete a mindfulness self-report questionnaire, the Kentucky Inventory of Mindfulness Skills (KIMS, Baer, Smith, &
Allen, 2004). This approach allows us to compare between meditators and non-meditators and additionally to investigate
the relation between mindfulness and cognitive control more closely.
2
The concurrent use of the d2-test of attention and the Stroop task enables us to test the ability to focus, sustain and direct
ones attention and to suppress interfering information. As cognitive flexibility implies the ability to interrupt or deautom-atise automated responses, that is to respond non-habitually, we shall briefly introduce the concepts of automatisation and
deautomatisation in relation to meditative practice.
2
According to S hiffrina ndSchneider (1977) cognitive processes can typically be classified as being either controlled or
automatic. They suggested that automatic processes operate in parallel and independent of attention, ‘‘automatic [processes]
do not require attention, though they may attract it if training is inappropriate, and they do not use up short-term memory
capacity” (Shiffrin & Schneider, 1977, p. 38). Spelke, Hirst, and Neisser (1976) have also stressed that behaviour should only
3
be termed ‘automatic’ if it did not involve certain high-order attentional skills. Additionally, Spelke et al. have shown that
whilst certain processes may be thought of as innately automatic, others may become automatic through practice. It is gen-erally accepted that once automated these processes are thought to be initiated unintentionally and effortlessly. Conse-quently they cannot be easily interrupted or prevented.
3
Numerous researchers (for example Dyer, 1973; Virzi & Egeth,
1985) have proposed that reading is seen as an automatic process, acquired through extensive practice and long-term learn-ing literate adults experienced. The Stroop effect (Stroop, 1935) evidences the difficulty of interrupting the automatic process
of reading of words in proficient readers. During the Stroop task participants are asked to attend to the colour in which words
are printed, not to the semantics of the words. Participant’s responses are significantly slower and less accurate when asked
to identify the colour of an incongruent colour word compared with responses to neutral and congruent words (M
acLeod,1991). In line with Stroop’s account, theories most often explain the Stroop asymmetry as a consequence of automaticity
( seeHasher & Zacks, 1979; LaBerge & Samuels, 1974; Logan, 1980). The automatic semantic activation of the word meaning
must be overridden in order for participants to respond correctly when faced with incongruent words. As reading is auto-matic for proficient readers less attentional resources are required for reading irrelevant (neutral) words than for ink colour
naming. Consequently, the reading of the word appears obligatory and results in an increase in reaction times and errors
when attempting to process incongruent colour words. Increasing the performance in this task would therefore require
the reinvestment of attention (deautomatisation) and a non-habitual response.
3
There is much debate as to whether processes that have become automatised can be brought back under top down con-trol. In view of Shiffrin and Schneider’s (1977) suggestion that automatised processes act independently of attention, it has
been proposed that deautomatisation is achieved through the reinvestment of attention in actions and behaviours (D
eikman,1963, 1966, 2000). Considering the definition of mindfulness as ‘‘bringing one’s complete attention to the present experience
on a moment-to-moment basis” (Marlatt & Kristeller, 1999, p. 68), the potential link to processes of deautomatisation be-comes obvious. If mindfulness-based meditation training improves the ability to invest one’s attention to the present mo-ment, it should be possible to bring processes that became automatised by extensive practice back under cognitive control.
In the current study we thus expected to find a reduction in Stroop interference as well as improvement in focused atten-tion and processing speed (as measure of sustained attention) in a group of buddhist meditators experienced in mindfulness
meditation compared to a meditation-naïve control group.
3
We furthermore expected to find higher self-reported mindfulness in the meditation group and a positive correlation be-tween mindfulness and the different measures of attentional performance and cognitive control. The KIMS consists of four
sub-scales which allows breaking down mindfulness into several components (Baer et al., 2004) .O bserving(1) highlights the
importance of observing and noticing the full variety of external and internal stimuli. As many forms of mindfulness training
include the practice of naming and labelling of the observed phenomena (e.g. Thera, 2005) ,d escribing(2) captures the resulting
ability to verbalise experiences. Acting with awareness (3) implies the ability to pay undivided attention to the activity one is
carrying out (e.g. Hanh, 1987), without getting distracted. Finally, accepting without judgment (4) covers the aspect of being
non-judgemental or non-evaluative about the presentmoment and thus implies the ability to refrain from automatic responses.
3
When investigating these sub-components of mindfulness, we expected that the two facets acting with awareness a
ndaccepting without judgment will exhibit the strongest correlations with task performance as these two facets most clearly
capture the aspects of inhibition of irrelevant and distracting information as well as the deautomatisation of cognitive
f
3
Methods
3
Participants
3
The study used two groups, a meditation group and a matched meditation-naïve control group. Group one consisted of 25
buddhist meditators, who were recruited from a local buddhist centre, where meditation practices that particularly focus on
mindfulness meditation are taught. Most of the buddhist participants were enrolled in intermediate classes and had at least
completed a 6-week beginners course on meditation. The control group consisted of 25 non-meditators. The majority were
recruited from a locally based, multinational credit management company and comprised of a wide spread of professions
including telephone operatives, team leaders, IT technicians, finance workers, account managers, upper management and
marketing executives. Some participants were also recruited from the local student population. The groups did not differ
with respect to gender (12 males, 13 females in each group), age (range: 20–40 years, mean age buddhists: 28.0 years, con-trols 27.5 years, p = .78) and amount of sleep during the night before testing (buddhists: 8.0, control: 7.9 h, p = .67).
3
Material
3
We used the Kentucky Inventory of Mindfulness Skills (KIMS) to assess the level of mindfulness participants experienced
(Baer et al., 2004).
3
The KIMS consists of 39 Likert-scale items, 18 of which are reversely scored. This questionnaire asks
the participants to rate the items with respect to ones g eneralexperience, rather than relating to a specific time point or per-iod. Possible answers are ‘Never or very rarely true’, ‘Rarely true’, ‘Sometimes true’, ‘Often true’ and ‘Very often or always
true’. High scores (1–5 for individual items) indicate high levels of mindfulness. The KIMS items are allocated to the four
4
s ub-scaleso bserving(12 items), d escribing(8 items), acting with awareness (10 items) and accepting without judgment
(9 items). The internal consistencies (Cronbach’s Alpha) for these sub-scales have been reported as .91, .84, .83 and .87,
respectively (Baer et al., 2004). The respective internal consistencies in the current study were .65, .49, .87 and .82.
5
Results
5
As can be seen in Table 2, KIMS total scores were significantly correlated with each of the examined variables (also see
Fig. 1 for the most important correlations). Positive correlations were found with the d2-scores TN (r = .510, p < .001), TN ?E
(r = .620, p < .01), CP (r = .667, p < .01) and the Stroop-score TNP (r = .331, p < .05). This indicates that high levels of mindful-ness are correlated with high processing speed, good attentional and inhibitory control, and a good coordination of speed
5
Table 2
Correlation matrix including all correlations between all KIMS scores (total and sub-scales) with all measures of attentional performance. As control, also the
correlations with age and sleep are included. For abbreviations refer to Table 1.
6
Fig. 1. Scatter graphs illustrating the correlations between the total mindfulness score (KIMS-total) and Stroop errors (top), d2 Concentration Performance
(Centre) and d2 total numbers of items processed minus errors (bottom). Control group (non-meditators) values are printed as black circles, meditator
values are printed as white squares.
A. Moore, P. Malinowski / Consciousness and Cognition 18 (2009) 176–186 181
6
with concurrent accurate performance. Negative correlations were found with the d2-errors E (r = ?. 527,p < .001), E1
(r = ?. 493,p < .001), E2 (r = ?. 398,p < .01) and the Stroop error SE (r = ?. 780,p < .001), signifying that higher levels of
7
mindfulness are linked to reduced errors across measures, suggesting greater attentional control, accuracy of visual scan-ning, inhibitory control, carefulness, cognitive flexibility and quality of performance. These results support the hypothesis
that mindfulness would correlate positively with task performance. It is important to note that all of the SE occurred whilst
participants were attempting to read the colour of incongruent colour words. Thus, SE may be said to result from a failure to
deautomatise.
7
The same pattern is found when examining each of the four mindfulness facets separately. Each of the four facets retain
significant positive correlations with TN, TN ?E, CP and TNP and negative correlations with E, E1, E2 and SE with a few minor
exceptions. It was specifically hypothesised that the acting with awareness and accepting without judgment facets of mind-fulness would have the strongest correlations with task performance. As can be seen in Table 2 this hypothesis is also sup-ported as these facets showed the highest correlations with each of the variables.
7
S everalt-tests were conducted in order to examine inter-group differences. These results are detailed in Table 3. Signif-icant between-group differences were found for all dependent variables (p < .01 or lower), confirming our expectation to find
higher levels of self-reported mindfulness as well as better attentional performance and thus cognitive flexibility on all mea-sures. As the scatter graphs in Fig. 1 also distinguish between meditators and non-meditators they provide a further illus-tration of these differences.
7
Regression analyses were carried out to assess the combined predictive power of the mindfulness facets. As the high in-ter-correlations of these mindfulness sub-scales already suggest, in most cases regression models with only one significant
factor emerged, highlighting the multicollinearity of the sub-scales. Obviously, the significant factor always was the one cor-relating highest with the predicted variable (see Table 2). The only exception to this pattern was that Stroop errors (SE) were
best predicted by a 2-factor solution including the factors KIMS-aware (standardised beta: ?. 57p < .001) and KIMS-observe
(standardised beta, ?. 26,p < .05) and an adjusted explained variance R
2
= .59 (p < .001).
7
Finally, we calculated the inter-correlations between the mindfulness sub-scales separately for the meditation and the
control group. As can be seen in Table 4, these inter-correlations tended to be higher for the non-meditators than for the
meditators. Comparing them to inter-scale correlations reported previously (Baer, Smith, Hopkins, Krietemeyer, & Toney,
2006; Baer et al. 2004) it turns out that particularly the inter-scale correlations for the control group were higher than pre-viously reported.
9
Conclusions