Content uploaded by Alexander J. Stell
Author content
All content in this area was uploaded by Alexander J. Stell on Oct 27, 2015
Content may be subject to copyright.
LKM, BIAS & OTHER-REGARDING EMOTIONS
1
Brief loving-kindness meditation reduces racial bias, mediated by positive other-regarding
emotions
Alexander J. Stell & Tom Farsides
University of Sussex, Brighton, UK
This is a pre-print version of a manuscript accepted for publication in Motivation & Emotion,
September 2015
APA Style Online First Citation:
Stell, A. J., & Farsides, T. (2015). Brief loving-kindness meditation reduces racial bias, mediated
by positive other-regarding emotions. Motivation and Emotion, 1-8. Advance online publication.
doi:10.1007/s11031-015-9514-x
LKM, BIAS & OTHER-REGARDING EMOTIONS
2
Abstract
The relationship between positive emotions and implicit racial prejudice is unclear. Interventions
using positive emotions to reduce racial bias have been found wanting, while other research
shows that positive affect can sometimes exacerbate implicit prejudice. Nevertheless, loving-
kindness meditation (LKM) has shown some promise as a method of reducing bias despite
increasing a broad range of positive emotions. A randomised control trial (n = 69) showed that a
short-term induction of LKM decreased automatic processing, increased controlled processing,
and was sufficient to reduce implicit prejudice towards the target’s racial group but not towards a
group untargeted by the meditation. Furthermore, the reduction in bias was shown to be mediated
by other-regarding positive emotions alongside increased control and decreased automaticity on
the IAT. Non-other-regarding positive emotions conversely showed no correlation with bias. The
study is the first to show that a short-term positive emotional induction can reduce racial
prejudice, and aids the understanding of how positive emotions functionally differentiate in
affecting bias.
Keywords: loving-kindness meditation, implicit social cognition, prejudice, positive other-
regarding emotions, implicit association task
LKM, BIAS & OTHER-REGARDING EMOTIONS
3
Brief loving-kindness meditation reduces racial bias, mediated by positive other-regarding
emotions
Short-term interventions using emotional inductions to reduce racial prejudice are usually
ineffective (Lai et al. 2014). Other research has shown that in comparison to negative affect,
general positive affect can lead to increased prejudiced thoughts and feelings toward racial out-
groups. Huntsinger et al. (2009) found that music that put white participants in a positive mood
elicited more stereotypic bias towards black people compared to music that primed a negative
mood. Similary, Bodenhausen et al. (1994) found that individuals induced to feel happy were
more likely to employ stereotypes in social judgements than those experiencing a neutral state.
One explanation for such findings is that positive states such as happiness may have the effect of
increasing reliance on internal knowledge structures such as heuristics (see Shiota 2014).
Nevertheless, research is beginning to show that positive emotions are not all alike in their
effects on implicit cognition. Griskevicius et al. (2010) compared the extent to which
participants’ persuasion processing was affected by six distinct positive emotions. Whilst those
induced with contentment, anticipatory enthusiasm, attachment love or amusement were more
persuaded by a weak argument, nurturant love and awe showed the opposite effect (Griskevicius
et al. 2010). This finding was taken by the authors to indicate that awe and nurturant love
contraindicate positive emotions’ usual reliance on automatic decision-making. If positive
emotions have divergent effects on automatic processing, it is possible they will also affect
implicit prejudice in varying ways. The present study explores the relative effects on bias of two
categories of positive emotions; those that are other-regarding or those that are non-other
regarding.
LKM, BIAS & OTHER-REGARDING EMOTIONS
4
The technique known to Buddhists as loving-kindness meditation (LKM) aims to self-
regulate an affective state of unconditional kindness towards the self and others. Although
techniques vary, practitioners of LKM typically repeat phrases such as ‘may you be happy and
healthy’ while visualizing a person (the target) experiencing the outcome of such wishes
(Salzberg 1995).
LKM is known to engender positive well-being outcomes for the individual, including
increases in general positive affect. This effect has been measured by self-report as well as with
autonomic processes such as vagal tone (Kok et al. 2013). Supporting traditional accounts that
regard LKM as a tool for interindividual harmony (Salzberg 1995), a few studies have now
highlighted LKM’s capacity to positively affect social cognition. Hutcherson et al. (2008)
showed that LKM significantly increased self-reported as well as implicit positive affect towards
a photo of a neutral stranger. Leiberg et al. (2011), demonstrated that a day-long induction of
LKM increased cooperative behaviour in a specially developed pro-social economic game.
Similarly, Weng et al. (2013) showed that LKM increased altruistic redistribution of funds to a
‘victim’ figure encountered outside the training context.
If LKM increases prosociality, inter-group processes would appear to be an area of
human life likely to benefit from such an effect. Hunsinger et al. (2012) investigated the effect of
LKM on explicit racial prejudice, comparing long-term meditators to a passive control (non-
meditators). Although this study found significantly less self-reported prejudice in long-term
LKM practitioners, the authors note the difficulty in drawing causal conclusions due to the
participants being a self-selecting sample, presumably drawn to the practice of meditation for
reasons that might also motivate the desire to express egalitarian values to a greater extent.
Nevertheless, more recent experimental work (Kang et al. 2013), has documented that a 6-week
LKM training has the capability to reduce implicit racial bias as measured by the implicit
association task (IAT) in comparison to both an active and a wait-list control group. Whilst such
LKM, BIAS & OTHER-REGARDING EMOTIONS
5
work highlights the potential value of LKM as a method to decrease racial bias, it is unclear
which mechanism or set of mechanisms mediates the effect. Kang et al. (2013) proposed two
mediators – cognitive control and stress – but both were found to be ineffective. As LKM’s most
apparent outcome is that it increases positive emotions, its effectiveness appears to contradict
past research which shows that positive emotions are either unrelated to bias or that they increase
it. The current paper attempts to solve this question by proposing that LKM attenuates racial bias
to the extent that it increases positive other-regarding emotions, and it is these emotions which
mediate the effect on bias. Unlike most positive emotions, which increase self-focused attention
(e.g. Abele et al. 2005) other-regarding positive emotions are those in which the locus of the
emotion is turned outward, toward others. Such emotions may have evolved to facilitate fitness-
enhancing activities such as cooperation, reciprocity and bonding (Shiota 2014). Gratitude, for
instance, is a response to another’s benevolence while awe and elevation are elicited by another’s
virtue, whether or not the self benefits (Horberg et al. 2011). Compassionate and nurturant forms
of love likewise centre on the wellbeing of the other (Haidt 2003).
Functional-adaptive approaches suggest that positive other-regarding emotions are
normally expressed in the presence of and elicited by close others. However, by encouraging
such emotions towards non-close others, LKM may engender kin-like responses towards those
targeted by the meditation. Thus, people who are not usually placed within one’s ‘Moral circle’
(Singer 2011) may be experienced as such due to the emotion’s associated action-tendency.
It is not known whether LKM can affect bias after a short exposure to the practice.
Although Hutcherson et al. (2008) showed that just a few minutes of LKM can affect positive
evaluations of a single target and Kang et al. (2013) showed that a 6-week intervention can
reduce prejudice, it remains uncertain whether brief LKM can reduce prejudice. An additional
aim of the current study then is to ascertain whether a short-term induction is efficacious in
reducing prejudice.
LKM, BIAS & OTHER-REGARDING EMOTIONS
6
Additionally, it is uncertain what the effect of targeting LKM to specific individuals or
groups is. Kang et al. (2013) for example allowed their participants to choose freely their
meditation targets. This would have made it likely that targets were of the same race. The present
study explores this issue by presenting a focal racial group – black people – as the target of
meditation and compares scores to a peripheral race untargeted by the meditation; specifically,
Asian people.
Finally, it is unclear whether LKM affects bias by decreasing automatic processing,
increasing controlled processing, or a combination of the two. Other meditative techniques are
believed to increase controlled processing (Moore and Malinowski 2009), yet changes in
automatic responding may also be predicted by LKM’s highly affective nature. Using a method
known as process disassociation procedure (PDP) analysis (Huntsinger et al. 2009), we will
measure LKM-induced changes in both controlled and automatic processing.
Present study
In this study we aimed to ascertain whether a short-term LKM induction was capable of
reducing implicit racial bias towards both a focal and a peripheral racial out-group. Additionally,
in order to explore possible mechanisms by which LKM may affect bias, we examined the
mediating effect of positive other-regarding emotions. Specifically, we predicted that although
LKM would lead to general increases in positive emotions, only those that were other-regarding
would successfully mediate LKM’s effect on bias. Finally we predicted that LKM would
decrease automatic and increase controlled processing on the IAT.
Method
Participants
LKM, BIAS & OTHER-REGARDING EMOTIONS
7
Seventy-one undergraduate students participated in a study advertised as “investigating
the effect of imagery on categorization”. Intended sample size was calculated a priori using a
power analysis in which we sought to detect a medium effect size with .80 power. This rough
effect size was based on prior research on LKM and bias (Kang et al. 2013, effect size between
LKM and control conditions). Participation was rewarded by entry into a random draw for two
cash prizes of £25. Only white people were sampled to avoid frequently observed inter-ethnic
differences in implicit bias (Nosek et al. 2006). To limit the risk of introducing biases in the
recruitment process, the data from non-whites signed up to the study was collected but not
analysed. This affected two study sign-ups. No participant reported meditating for more than 30
minutes per week. Two participants were excluded for making errors in over 40% of IAT trials1
leaving a total of sixty-nine participants (50 women; Mage = 23.7 years, SD = 4.24).
Material
Implicit Association Task (IAT). Implicit racial bias was measured using the race
Implicit Association Test (IAT), following Greenwald et al.’s (2003) recommendations.
Category word “black” and category term “white” were presented in either corner at the
top of the screen. Below each was either the category term “good” or the category term “bad”.
Pairings of the words in each corner varied. A series of attribute terms (e.g., “wonderful”) and
photographs of either black or white people were presented in the middle of the screen.
Participants used one or other computer key to identify them as belonging to a category shown in
the top-left or the top-right. How long participants took to press the correct key following
presentation of the target word or photograph was recorded. Racial bias is thought to be indicated
by faster identification of positive attribute terms when the positive category term was paired
with the “white” (in-group) category term than when it was paired with the “black” (out-group)
1 Inclusion of these participants’ data did not alter the interpretation of any of the substantive
findings.
LKM, BIAS & OTHER-REGARDING EMOTIONS
8
one, and also by faster identification of negative attribute terms when the negative category term
was paired with the out-group category term than when it was paired with the in-group one.
To explore whether LKM induces a race-specific reduction in bias, all participants
completed two IATs: one with “black” people as the out-group (IAT-1) and another with
“Asian” people as the out-group (IAT-2). The order in which these IATs were administered was
also counterbalanced and was found to have no effect on the dependent variables.
Positive other-regarding and positive non-other-regarding emotions. To assess the
effect of the manipulation on self-reported positive emotion, positive items from the modified
Differential Emotions Sub-scale (mDES; Fredrickson et al. 2003) were administered. This
measure asks participants to rate their strongest experience, during the manipulation, of each of
11 specific emotions on a 5-point scale (e.g. awe: “During the visualization exercise, I felt awe,
wonder, or amazement” 0 = not at all to 4 = extremely). As previous research on implicit bias
has shown that the specificity of an emotion in terms of its capability to provoke functional-
adaptive responses is more important than some general quality like valence (Dasgupta et al.
2009) we sub-divided these emotions into either other regarding or non-other-regarding based on
existing theoretical approaches (Haidt 2003; Horberg et al. 2011). Positive other-regarding
emotions were gratitude, elevation, love and awe (α = .92). Non-other-regarding emotions were:
amusement, buoyancy, hope, curiosity, happiness, pride and contentment (α = .85).
Face Images. For the LKM and Imagery manipulations, target images were one of eight
possible gender-matched black people acquired from the Centre of Vital Longevity’s face
database (Minear and Park 2004). Images for the IAT tasks were taken from the Multi-Category
Implicit Association Test (MC-IAT; Nosek et al. 2013) which feature colour, custom-morphed
images of white, black and Asian faces.
Procedure and Conditions
LKM, BIAS & OTHER-REGARDING EMOTIONS
9
Participants were randomly allocated to either a loving-kindness meditation (LKM) or a
visualization (Imagery) condition. In each condition, instructions lasting approximately 7
minutes were given over headphones. Replicating Hutcherson et al.’s (2008) induction,
participants were asked to close their eyes, relax and take some deep breaths.
In the LKM condition (n = 34), participants were then instructed to imagine people who
‘deeply cared for them’ standing on either side of them, sending them love. Then after
approximately 4 minutes, they were asked to open their eyes and redirect the feelings of love
towards gender-matched black person shown in a photograph, and then wish them health,
happiness and wellbeing (Salzberg, 1995).
In the Imagery condition (n = 35), participants were first instructed to think about the
physical characteristics of two acquaintances for whom they had no strong feelings, after which
they were asked to open their eyes and pay close attention to the physical features of the same
gender-matched black person show in a photograph as was used in the LKM condition.
Thus, participants in each condition underwent a closely matched procedure but with
only those in the LKM condition imagining receiving and sending loving thoughts and feelings.
After the manipulation, participants were presented with the counter-balanced IATs
followed by the mDES. Finally, all participants were debriefed.
(Insert Table 1 about here)
Results
Positive emotions. To gauge whether LKM increased positive emotions, a 2 (treatment:
LKM/Imagery) × 2 (emotion locus: other-regarding/non-other-regarding) mixed-model ANOVA
was conducted with the first factor between and the second factor within subjects. Effects were
found for treatment, F(1, 67) = 34.22, p < .001, = .34, emotion locus, F(1, 67) = 9.75, p =
.003, = .13, and emotion locus × treatment, F(1, 67) = 35.52, p < .001, = .35. In line with
η
P
2
η
P
2
η
P
2
LKM, BIAS & OTHER-REGARDING EMOTIONS
10
our predictions, follow-up t-tests revealed that, compared to Imagery, LKM increased both other-
regarding (Imagery: M = .81, SD = .80; LKM: M = 2.46, SD = 1.14), t(67) = 7.02, p < .001, d =
1.67, and non-other regarding (Imagery: M = 1.60, SD = 0.68; LKM: M = 2.21, SD = 0.86),
t(67) = 3.27, p = .002, d = 0.78 positive emotions (see Table 1). Paired t-tests qualified the
significant emotion locus × treatment interaction by revealing that those induced with Imagery
exhibited larger amounts of non-other regarding (M = 1.60, SD = 0.68) than other-regarding (M
= 0.81, SD = 0.79) emotions , t(33) = -7.33, p < .001, d = 1.07, whereas this pattern was reversed
for the LKM group: participants induced with LKM exhibited larger amounts of other-regarding
(M = 2.46, SD = 1.14) than non-other-regarding (M = 2.21, SD = 0.86) emotions, although this
difference fell just short of significance, t(33) = 1.80, p = .08, d = 0.25. Thus, participants
induced with LKM experienced similar or heightened levels of positive affect directed towards
others as they felt pertaining to themselves, whereas the Imagery group exhibited significantly
more self- than other-directed positive affect.
Implicit Racial Prejudice. In order to test whether the LKM group exhibited a race-
specific reduction in bias, IAT d bias scores were submitted to a 2 (treatment: LKM/Imagery) ×
2 (Race: Black/Asian) mixed-model ANOVA with the first factor between and the second factor
within subjects. The d outcome measure was calculated in accordance with Greenwald et al.
(2003) by subtracting SD corrected latencies for in-group/positive responses from out-
group/positive responses and in-group/negative responses from out-group/negative responses.
Effects were found for race, F(1, 67) = 19.37, p < .001, = .22, and race × treatment, F(1, 67)
= 6.34, p = .01, = .09. No significant overall treatment effect emerged, F(1, 67) = 1.90, p =
.17, = .03 . These results were qualified by follow-up independent t-tests. For IAT-1 with
black people as the out-group, results indicated, as predicted a significant decrease in overall bias
for the LKM group (M = 0.33, SD = 0.41) in comparison to the Imagery group (M = 0.57, SD =
η
P
2
η
P
2
η
P
2
LKM, BIAS & OTHER-REGARDING EMOTIONS
11
0.40), t(67) = -2.44, p = .02, with a moderate effect size (Cohen’s d = 0.59). However, in IAT-2
(with Asians as out-group), there was no significant differences in bias between the LKM (M =
0.24, SD = 0.47) and Imagery (M = 0.23, SD = 0.34) groups, t(67) = 0.04, p = .97, d = 0.02 (see
Table 1). Therefore, LKM was only effective at reducing bias towards the focal target group;
bias towards the peripheral group was unaffected.
PDP Analysis. To test whether LKM affected controlled or automatic processing (or
both), PDP analysis was used in accordance with Huntsinger et al. (2009). Two separate
estimates for controlled and automatic processing during the IATs were created. The controlled
estimate (C) was computed by subtracting the probability of incorrect responses on the
incompatible blocks from the probability of correct responses on the compatible blocks. The
automatic estimate (A) was computed by taking the probability of incorrect responses on the
incompatible blocks and dividing it by (1 – C). A 2 (treatment: LKM/Imagery) × 2 (processing
type: automatic/controlled) mixed-model ANOVA found effects for processing type, F(1, 67) =
68.45, p < .001, = .51, and processing type × treatment, F(1, 67) = 10.59, p = .002, = .14.
Follow-up independent t-tests showed that, as predicted, automatic processing was significantly
lower in the LKM (M = .55, SD = .25) than in the Imagery (M = .67, SD = .21) condition, t(67) =
-2.20, p = .03, d = 0.53. Furthermore, controlled processing was observed to be significantly
higher in the LKM (M = .90, SD = .06) than in the Imagery (M = .82, SD = .15) condition, t(67)
= 2.79, p = .01, d = 0.62. As the data for the controlled estimate exhibited non-normality and
heterogeneous variance, a Mann-Whitney U-test was also employed. The effect remained
significant, U = 391.50, p = .01. These results indicate that LKM increased controlled and
decreased automatic processing (see Table 1).
Mediation analysis. Mediation analysis was performed using Preacher and Hayes’ (2008)
INDIRECT macro for SPSS, which has been shown to have more power in smaller samples than
η
P
2
η
P
2
LKM, BIAS & OTHER-REGARDING EMOTIONS
12
normal theory tests such as the Sobel test. Significance tests for indirect effects are reported as
bias corrected and accelerated 95% confidence intervals (Preacher and Hayes 2008) while direct
effects between IV, mediator and DV are reported using critical alpha values. First, to test
whether our primary candidate mediator, positive other-regarding emotions, accounted for the
effect of LKM on implicit prejudice, we used the overall bias score from IAT-1 as dependent
variable and conducted a single mediation analysis with treatment group as independent variable,
and either positive other-regarding or positive non-other-regarding emotions as proposed
mediators. LKM was associated with increases in other-regarding emotions (B = 1.66, SE = 0.24,
p < .001) and other-regarding emotions were associated with decreases in bias (B = -0.13, SE =
0.05, p = .02). Results of 5000 bias-corrected and accelerated bootstrapped samples confirmed a
significant indirect effect of LKM on bias though other-regarding positive emotions with 95%
bootstrapped confidence intervals (BCI) showing no overlap with zero (B = -0.21, SE = 0.10,
95% BCI [-0.44, -0.05]). Contrastingly, whilst positive non-other-regarding emotions were
increased by LKM, (B = 0.61, SE = 0.19, p = .002), there was no relationship between these
emotions and bias, (B = -0.05, SE = 0.06, p = .40) and accordingly, the indirect effect of LKM
on bias through non-other-regarding emotions was nonsignificant (B = 0.001, SE = 0.04, 95%
BCI [-0.09, 0.07]). Finally, we took an exploratory step and added the controlled and automatic
estimates from the PDP analysis as candidate mediators in a multiple mediator model that
included positive other-regarding emotions. Regression coefficients and BCIs for the final model
are presented in Figure 1. LKM was associated with decreases in automatic processing (B = -.12,
SE = .24, p < .001), and this mediator was associated with increases in bias (B = .43, SE = .20
, p = .03). LKM was associated with increases in controlled processing (B = -.08, SE = .24, p <
.001). However, although this candidate mediator evidenced significant zero-order negative
correlations with bias (r = -.28, p = .02), when controlling for the effect of the other mediators in
the multiple mediation model, it fell over the threshold for significance (B = .66, SE = .40, p =
LKM, BIAS & OTHER-REGARDING EMOTIONS
13
.10). Nevertheless, bootstrapped estimates of indirect effects for both processing types were
significant (Automatic: B = -.05, SE = .04, 95% BCI [-.14, -.002], Controlled: B = -.05, SE = .03,
95% BCI [-.14, -.003]), indicating that, alongside inducing positive other-regarding emotions,
LKM reduces bias through reducing automatic as well as increasing controlled processing.
Additionally, it was observed that the effect of adding these three successful mediators to the
mediation model meant that the direct path from treatment group to implicit bias became non-
significant (B = -0.05, SE = 0.12, p = .71). This suggested that the mediators accounted for much
of the effect of LKM on bias.
(Insert Figure 1 about here)
Discussion
We found that just seven minutes of loving-kindness meditation directed to a member of
a racial out-group was sufficient to reduce racial bias towards that out-group. The practice was
only effective in reducing bias for the focal race; implicit bias scores towards a peripheral racial
group – Asian people – were not impacted. Additionally, LKM’s effects on bias were mediated
by the presence of other-regarding but not non-other-regarding positive emotions. Furthermore,
LKM appears to gain efficacy both by increasing controlled processing and by decreasing
automatic processing.
The current study is the first to successfully find a short-term positive emotion induction
that reduces racial prejudice as measured by the IAT. Furthermore we have found support for
the possibility that it is the social locus of the positive emotion (whether or not it is other-
regarding) that is crucial in whether the emotion will decrease bias. Finally, we have shown,
again for the first time, how improved appraisals of others that result from LKM practice extend
beyond the metalized recipient (target) of LKM, towards those categorically related (the target’s
race).
LKM, BIAS & OTHER-REGARDING EMOTIONS
14
No previously published study has demonstrated that a short-term emotional induction
has the power to reduce racial bias as measured by the IAT (see Lai et al. 2014). Furthermore,
given that successful (non-affective) interventions in Lai et al. (2014) achieved a reduction of
bias by in some way priming negative construals of the in-group, LKM seems to constitute a
technique that may be more easily scaled up to programs promoting prejudice reduction in the
world outside the laboratory. It seems reasonable to suggest that LKM, with its reported benefits
for the individual, is a more acceptable method of bias reduction than, for example, interventions
that include priming reviled white figures such as Adolf Hitler and Ted Bundy (Lai et al. 2014;
intervention 5).
We found a decrease in prejudice towards a focal racial group but no difference in scores
for a group not targeted by the intervention, Asian people. This result appears to depart from
Kang et al. (2013), who found that prejudice towards black people was decreased by a 6-week
LKM training, despite this group not being explicitly directed to in the training. One possible
reason for this difference, that should be tested in future research, is that LKM enacts both a
specific effect on the target of meditation and their group, as well as a diffuse, but perhaps
weaker, effect on other groups. This diffuse effect may not have been measurable due to the
short-term nature of our intervention. Finding a specific effect of LKM is notable in the context
of previous work on the subject, where it has been unclear to what extent implicit positive
appraisals extend beyond the person who is the target of meditation (Hutcherson et al. 2008).
Presently, we have shown that the effects of LKM appear to extend at least to the target’s racial
group.
Based on a social-functional understanding of emotions, we hypothesized that one sub-set
of positive emotions would serve to mediate LKM-induced reductions of bias. Specifically, we
predicted that positive other-regarding emotions – that is, emotions that have a direct linkage
with positive social aspects of inter-individual functioning – would predict bias reduction, but
LKM, BIAS & OTHER-REGARDING EMOTIONS
15
that non-other-regarding positive emotions would not evidence such effects. Our results
indicated this to be the case. Although LKM was demonstrated to significantly increase a broad
range of other-focused and non-other-focused positive emotions, the latter category showed no
relationship with changes in implicit bias. This indicates that positive emotions need to implicate
others if they are to affect bias; feeling good per se – whilst an outcome of LKM practice – is
probably not the mechanism by which it attenuates prejudice. Such observations then provide
reason to doubt the sufficiency of adapting valence-based theories to prejudice reduction (e.g.
Johnson and Fredrickson 2005) as it appears it is the social locus of the emotion, and not
valence, that matters.
Making the distinction between other- and non-other regarding positive emotions also
adds to the debate on the relationship between positive emotion and either automatic or
controlled cognitive processes. The process-dissociation procedure and mediation analysis
showed that positive emotions induced by LKM served to increase controlled and reduce
automatic processing, thereby reducing bias. That LKM increases control on the IAT may relate
to the practice’s close association with mindfulness, a construct linked to elevated attentional
performance and cognitive flexibility (Moore and Malinowski 2009). A more puzzling question
regards LKM’s role in decreasing automaticity when past research has shown positive emotions’
capacity to increase automatic thinking. One possibility is that this was due to relatively
pronounced other-regarding emotions which, instead of directing attention towards internal
knowledge structures (e.g. biases), focused attention outwards towards the elicitor. Some support
for this interpretation is found in Griskevicius et al.’s (2010) study in which six out of eight
positive emotions increased susceptibility to a weak argument. Awe and nurturant love, however
– two other-regarding emotions – had the opposite effect. This finding was interpreted in terms
of these emotions’ capacity, compared to the other positive emotions measured, to decrease
reliance on heuristics thereby reducing the persuasiveness of the argument.
LKM, BIAS & OTHER-REGARDING EMOTIONS
16
It is important to consider the reasons why LKM succeeded in reducing bias despite
positive emotions’ (even those that are other-regarding) bad track record in this context. Recent
work focusing on elevation for example has shown it to be ineffective in changing racial bias,
even when it is elicited in the context of an admirable black person (Lai et al. 2014; Lai et al.
2013). One possibility is that LKM may differ from other positive emotion inductions by
recruiting multiple other-regarding emotions. Effect sizes are generally much smaller for positive
than negative emotional inductions (Westermann et al. 1996), so it is possible that the cumulative
effect of the various relevant positive emotions is enough to make the difference. Another,
related possibility is that LKM might gain efficacy by affecting sentiments (Prinz 2007) towards
the target directly. Sentiments are not emotions themselves but an orientation (e.g. liking,
disliking) towards a person or object that modulates the occurrence of specific emotions. Love
for instance appears to defy categorization as single, distinct emotion (Scherer 2005) and might
be better classified as a sentiment. LKM may work by modulating sentiments towards targets,
which produce emotions, rather than increasing emotions per se. Nevertheless, future research
will need to employ experimental manipulations of specific emotions and (or) sentiments to
investigate the relationship between positive emotions and bias reduction.
The present study helps identify the effect of positive emotions on implicit bias in loving-
kindness meditation, and isolates positive other-regarding emotions, alongside changes in
cognitive processing, as putative mechanisms towards inter-group harmony.
LKM, BIAS & OTHER-REGARDING EMOTIONS
17
References
Abele, A., Silvia, P., & Zöller-Utz, I. (2005). Flexible effects of positive mood on self-focused
attention. Cognition & Emotion, 19(4), 623–631. doi:10.1080/02699930441000391
Bodenhausen, G. V., Kramer, G. P., & Süsser, K. (1994). Happiness and stereotypic thinking in
social judgment. Journal of Personality and Social Psychology, 66(4), 621–632.
doi:10.1037/0022-3514.66.4.621
Dasgupta, N., DeSteno, D., Williams, L. A., & Hunsinger, M. (2009). Fanning the flames of
prejudice: The influence of specific incidental emotions on implicit
prejudice. Emotion, 9(4), 585-591. doi:10.1037/a0015961
Fredrickson, B. L., Tugade, M. M., Waugh, C. E., & Larkin, G. R. (2003). What good are
positive emotions in crisis? A prospective study of resilience and emotions following the
terrorist attacks on the United States on September 11th, 2001. Journal of Personality
and Social Psychology, 84(2), 365-376. doi:10.1037/0022-3514.84.2.365
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the implicit
association test: I. An improved scoring algorithm. Journal of Personality and Social
Psychology, 85(2), 197-216. doi:10.1037/0022-3514.85.2.197
Haidt, J. (2003). The moral emotions. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith
(Eds.), Handbook of affective sciences (pp. 852– 870). Oxford, UK: Oxford University
Press.
Horberg, E. J., Oveis, C., & Keltner, D. (2011). Emotions as moral amplifiers: An appraisal
tendency approach to the influences of distinct emotions upon moral judgment. Emotion
Review, 3(3), 237-244. doi:10.1177/1754073911402384
Hunsinger, M., Livingston, R., & Isbell, L. (2012). Spirituality and intergroup harmony:
Meditation and racial Prejudice. Mindfulness, 5(2), 139–144. doi:10.1007/s12671-012-
0159-5
LKM, BIAS & OTHER-REGARDING EMOTIONS
18
Huntsinger, J. R., Sinclair, S., & Clore, G. L. (2009). Affective regulation of implicitly
measured stereotypes and attitudes: Automatic and controlled processes. Journal of
Experimental Social Psychology, 45(3), 560–566. doi:10.1016/j.jesp.2009.01.007
Hutcherson, C. A., Seppala, E. M., & Gross, J. J. (2008). Loving-kindness meditation increases
social connectedness. Emotion, 8(5), 720-724. doi:10.1037/a0013237
Johnson, K., & Fredrickson, B. (2005). "We all look the same to me": Positive emotions
eliminate the own-race bias in face recognition. Psychological Science, 16(11), 875-881.
doi:10.1111/j.1467-9280.2005.01631.x
Kang, Y., Gray, J. R., & Dovidio, J. F. (2014). The nondiscriminating heart: Lovingkindness
meditation training decreases implicit intergroup bias. Journal of Experimental
Psychology: General, 143(3), 1306–1313. doi:10.1037/a0034150
Kok, B. E., Coffey, K. A., Cohn, M. A., Catalino, L. I., Vacharkulksemsuk, T., Algoe, S. B., …
Fredrickson, B. L. (2013). How positive emotions build physical health: Perceived
positive social connections account for the upward spiral between positive emotions and
vagal tone. Psychological Science, 24(7), 1123–1132. doi:10.1177/0956797612470827
Lai, C. K., Haidt, J., & Nosek, B. A. (2013). Moral elevation reduces prejudice against gay men.
Cognition & Emotion, 28(5), 781–794. doi:10.1080/02699931.2013.861342
Lai, C. K., Marini, M., Lehr, S. A., Cerruti, C., Shin, J.-E. L., Joy-Gaba, J. A., … Nosek, B. A.
(2014). Reducing implicit racial preferences: I. A comparative investigation of 17
interventions. Journal of Experimental Psychology: General, 143(4), 1765–1785.
doi:10.1037/a0036260
Leiberg, S., Klimecki, O., & Singer, T. (2011). Short-term compassion training increases
prosocial behavior in a newly developed prosocial game. PloS one, 6(3): e17798.
doi:10.1371/journal.pone.0017798
LKM, BIAS & OTHER-REGARDING EMOTIONS
19
Minear, M., & Park, D. C. (2004). A lifespan database of adult facial stimuli. Behavior Research
Methods, Instruments, and Computers, 36, 630-633. doi:10.3758/BF03206543
Moore, A., & Malinowski, P. (2009). Meditation, mindfulness and cognitive flexibility.
Consciousness and Cognition, 18(1), 176–186. doi:10.1016/j.concog.2008.12.008
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2007). The Implicit Association Test at age 7:
A methodological and conceptual review. In J. A. Bargh (Ed.), Social psychology and the
unconscious: The automaticity of higher mental processes (pp. 265– 292). New York,
NY: Psychology Press.
Nosek, B. A., Sriram, N., Smith, C. T., & Bar-Anan, Y. (2013). The multi-category Implicit
Association Test. Unpublished manuscript.
Prinz, J. J. (2007). The emotional construction of morals. Oxford, England: Oxford University
Press.
Salzberg, S. (1995). Loving-kindness: The revolutionary art of happiness. Boston, MA:
Shambhala Publications.
Scherer, K. R. (2005). What are emotions? And how can they be measured? Social Science
Information, 44(4), 695–729. doi:10.1177/0539018405058216
Singer, P. (2011). The expanding circle: Ethics, evolution, and moral progress. Princeton, NJ:
Princeton University Press.
Shiota, M. N. (2014). The evolutionary perspective in positive emotion research. In M. M.
Tugade, M. N. Shiota, & L. D. Kirby (Eds.). Handbook of positive emotions (pp. 44-59).
Guilford Publications.
Weng, H. Y., Fox, A. S., Shackman, A. J., Stodola, D. E., Caldwell, J. Z., Olson, M. C., ... &
Davidson, R. J. (2013). Compassion training alters altruism and neural responses to
suffering. Psychological science, 24(7) 1171–1180. doi:10.1177/0956797612469537
LKM, BIAS & OTHER-REGARDING EMOTIONS
20
Westermann, R., Spies, K., Stahl, G., & Hesse, F. W. (1996). Relative effectiveness and validity
of mood induction procedures: A meta‐analysis. European Journal of Social
Psychology, 26(4), 557-580. doi:10.1002/(SICI)1099-0992(199607)26:4<557::AID-
EJSP769>3.0.CO;2-4
LKM, BIAS & OTHER-REGARDING EMOTIONS
21
Table 1. Participant demographics and descriptive statistics of study variables
Variable
Loving-Kindness
(n = 34)
Neutral Imagery
(n = 35)
Statistic
Demographic
Age in years (SD)
23.29 (3.83)
24.09 (4.61)
Female (%)
74
72
Study Variable
Implicit Bias (IAT):
black people
.33 (.41)
.57 (.40)
t = -2.44*
Implicit Bias (IAT):
Asian people
.24 (.47)
.23 (.33)
t = .04
Automaticity on
IAT-1
.55 (.25)
.67 (.21)
t = -2.20*
Cognitive control on
IAT-1
.90 (.06)
.82 (.15)
t = 2.79*
Positive other-
regarding emotions
2.46 (1.14)
.81 (.80)
t = 7.02***
Positive non-other-
regarding emotions
2.21 (.86)
1.60 (.68)
t = 3.27**
Note: Mean values are displayed with standard deviations in parentheses where
applicable. Values for self-reported emotions are 5-point likert scale scores.
IAT = Implicit Association Task.
*** p < .001, ** p < .01, * p < .05
LKM, BIAS & OTHER-REGARDING EMOTIONS
22
Figure 1. Multiple Mediation Model showing variables that mediate LKM’s effect on implicit
racial bias. Values are unstandardised regression coefficients with bootstrapped confidence
intervals for each mediator’s specific indirect effect in square brackets.
*** p < .001, ** p < .01, * p < .05, + p = .10
Loving-Kindness Meditation
Positive Other-regarding
Emotions
Implicit Racial Bias
Automatic Processing
Controlled Processing
-18 [-.44, -.02]
-.05 [-.14, -.002]
-.05 [-.12, -.003]
1.65***
-.12*
.08**
-.24*
-.05 (ns)
-11*
-44*
-66
+
