Content uploaded by Yoona Kang
Author content
All content in this area was uploaded by Yoona Kang on May 05, 2014
Content may be subject to copyright.
Journal of Experimental Psychology: General
The Nondiscriminating Heart: Lovingkindness Meditation
Training Decreases Implicit Intergroup Bias
Yoona Kang, Jeremy R. Gray, and John F. Dovidio
Online First Publication, August 19, 2013. doi: 10.1037/a0034150
CITATION
Kang, Y., Gray, J. R., & Dovidio, J. F. (2013, August 19). The Nondiscriminating Heart:
Lovingkindness Meditation Training Decreases Implicit Intergroup Bias. Journal of
Experimental Psychology: General. Advance online publication. doi: 10.1037/a0034150
The Nondiscriminating Heart: Lovingkindness Meditation Training
Decreases Implicit Intergroup Bias
Yoona Kang
Yale University
Jeremy R. Gray
Michigan State University
John F. Dovidio
Yale University
Although meditation is increasingly accepted as having personal benefits, less is known about the broader
impact of meditation on social and intergroup relations. We tested the effect of lovingkindness meditation
training on improving implicit attitudes toward members of 2 stigmatized social outgroups: Blacks and
homeless people. Healthy non-Black, nonhomeless adults (N⫽101) were randomly assigned to 1 of 3
conditions: 6-week lovingkindness practice, 6-week lovingkindness discussion (a closely matched active
control), or waitlist control. Decreases in implicit bias against stigmatized outgroups (as measured by
Implicit Association Test) were observed only in the lovingkindness practice condition. Reduced
psychological stress mediated the effect of lovingkindness practice on implicit bias against homeless
people, but it did not mediate the reduced bias against Black people. These results suggest that
lovingkindness meditation can improve automatically activated, implicit attitudes toward stigmatized
social groups and that this effect occurs through distinctive mechanisms for different stigmatized social
groups.
Keywords: meditation, lovingkindness, compassion, implicit bias, prejudice
Various forms of meditation are gaining mainstream popularity
in Western therapeutic and health professions (Bishop et al., 2004).
More than 20 million American adults, about one out of 11,
meditated in 2007 (Barnes, Bloom, & Nahin, 2008). This wide-
spread interest is fueled by empirical evidence on meditation
supporting enhanced individual well-being, including reduction of
stress (Astin, 1997), anxiety (Evans et al., 2008), and depression
(Teasdale et al., 2000); improvement of mood (Fredrickson, Cof-
fey, Pek, Cohn, & Finkel, 2008); and even enhancement of aca-
demic performance (Hall, 1999). Although meditation practice has
a variety of benefits associated with personal well-being, evidence
of its broader implications on interpersonal and intergroup out-
comes is limited (cf. Hutcherson, Seppala, & Gross, 2008). The
present study aimed to expand the current breadth of investigation
that is largely limited to meditation effects within individuals and
to examine the impact of meditation across individuals. Specifi-
cally, we examined changes in implicit intergroup attitudes using
lovingkindness meditation.
Lovingkindness meditation is intended to cultivate warm and
friendly feelings toward the self and others (Hutcherson et al.,
2008). In practice, individuals contemplate warm feelings they
have toward a person they care about the most (e.g., a family
member). They then extend these positive thoughts first to them-
selves and then to a growing circle of others, eventually to all
sentient beings. In support of the view that lovingkindness medi-
tation may enhance interpersonal connectedness, Fredrickson et al.
(2008) demonstrated that participants who attended a 7-week
lovingkindness meditation course, relative to waitlisted controls,
reported greater increases in social support and positive relations
with close others. Hutcherson et al. (2008) obtained further support
for the idea that lovingkindness meditation can change general
interpersonal orientations, such that a 7-min lovingkindness med-
itation increased explicit positivity toward the self, close others,
and unacquainted individuals, and increased implicit positivity
toward the self and close others but not unacquainted others. In the
present research, we investigated whether practicing lovingkind-
ness meditation could influence attitudes at an intergroup level,
specifically implicit prejudice.
Traditionally, prejudice has been measured explicitly using self-
reports, which are subject to deliberate, and often strategic, control
in their expression. By contrast, implicit attitudes, which are typ-
ically measured with response-latency techniques such as the
Implicit Association Test (IAT; Greenwald, Poehlman, Uhlmann,
& Banaji, 2009), represent automatically activated associations.
Because people typically adjust their explicit expressions to appear
unbiased, implicit attitudes are often better predictors of subse-
quent behavior in socially sensitive domains (Greenwald et al.,
2009). Also, because of the speed and ease with which evaluative
Yoona Kang, Department of Psychology, Yale University; Jeremy R.
Gray, Department of Psychology, Michigan State University; John F.
Dovidio, Department of Psychology, Yale University.
We are most grateful to Lama Tsondru for his guidance and support. We
thank Kathryn Redford, Daniel Millstein, Ranjani Prabhakaran, Anna
Newheiser, Hyeyoung Shin, and Kevin Callender for their contributions.
Correspondence concerning this article should be addressed to Yoona
Kang, Department of Psychology, Yale University, P.O. Box 208205, New
Haven, CT 06520. E-mail: yoona.kang@yale.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Experimental Psychology: General © 2013 American Psychological Association
2013, Vol. 142, No. 4, 000 0096-3445/13/$12.00 DOI: 10.1037/a0034150
1
judgments are activated, almost a century’s worth of empirical
investigation has led to the consensus that implicit group attitudes
are extremely difficult to control. However, some findings suggest
that it is possible for people to inhibit implicit biases longer term
through the development of automatic goals— goals that are pur-
sued with sufficient frequency that they become chronically ac-
cessible (Bargh, Gollwitzer, Lee-Chai, Barndollar, & Trotschel,
2001; Moskowitz & Ignarri, 2009). Lovingkindness meditation
establishes a deep sense of positive interconnectedness to others,
regardless of their group membership, that is incompatible with
social biases and thus can potentially inhibit automatically acti-
vated biases. Consistent with the idea that it is extended practice
of lovingkindness meditation that is required to inhibit implicit
intergroup biases, Hutcherson et al. (2008) found that a 7-min
lovingkindness meditation was not sufficient to reduce implicit
bias toward unacquainted others.
In the present research, we examined the broad potential of
lovingkindness meditation, practiced more extensively, to have
additional significant social impact on improving implicit negative
attitudes toward members of culturally stigmatized groups. In the
present study, 101 participants were randomly assigned (a) to
practice lovingkindness meditation for 6 weeks (lovingkindness
practice), (b) to discuss ideas about lovingkindness meditation for
6 weeks (lovingkindness discussion), or (c) to participate in lov-
ingkindness meditation at a later date (waitlist control partici-
pants). Participants completed measures of implicit attitudes to-
ward Blacks and homeless people, assessed with IATs (Greenwald
et al., 2009), at the beginning and end of the study period.
We selected Blacks and homeless people as the target groups to
examine the breadth of impact of lovingkindness meditation. Im-
plicit prejudices toward Blacks and homeless people are both
strong and pervasive and may both involve affective regulatory
processes. In general, feelings of fear and vulnerability underlie
prejudice against Blacks (e.g., Rudman & Lee, 2002), whereas
biases against homeless people often involve disgust and con-
tempt, which may lead to dehumanization (Harris & Fisk, 2006).
Therefore, increased social perception, or humanization, may play
a more important role in bias reduction toward homeless people
than toward Blacks. In addition, the social norms about the ex-
pression of bias toward these groups (and thus explicit attitudes)
differ (Crandall, Eshlemen, & O’Brien, 2002). Because of the
historical, political, and social emphasis on the immorality and
illegality of racial discrimination, there are generally strong forces
against prejudice toward Blacks, present even in early socialization
(Olson, Dweck, Spelke, & Banaji, 2011). By contrast, prejudice
against homeless people is more normatively acceptable (Crandall
et al., 2002). Therefore, cognitive factors, such as cognitive control
that relates to recognition of societal and personal standards of
egalitarian treatment toward Blacks (Crandall et al., 2002), may
play a more central role in implicit bias against Blacks than against
homeless people.
Thus, in the present research, we had three objectives. First, we
sought to investigate whether lovingkindness meditation could
reduce implicit bias toward a stigmatized group. Second, we tested
whether such an effect would be generalizable across two different
target groups: Blacks and homeless people. And third, we exam-
ined potentially different mechanisms that contribute to reductions
in implicit biases toward these two groups: cognitive control and
stress.
Based on research demonstrating how implicit biases can be
inhibited through extensive practice creating strong incompatible
associations (Kawakami, Dovidio, Moll, Hermsen, & Russin,
2000) or goals (Bargh et al., 2001; Moskowitz & Ignarri, 2009),
we hypothesized that only those who actively engage in loving-
kindness meditation practice—and not those in the discussion or
waitlist control—would show reduced bias in implicit attitudes
toward Blacks and homeless people.
We also hypothesized that the effect of lovingkindness medita-
tion on implicit attitudes toward Blacks and homeless people may
involve distinct mechanisms, and accordingly we tested two po-
tential underlying mechanisms that may account for the hypothe-
sized effects for both groups (Kang, Gruber, & Gray, 2012).
Specifically, we predicted that decreases in bias would be medi-
ated by increased cognitive control and decreased psychological
stress. First, extended meditation practice (Tang et al., 2007) or
even a shorter term 3-day lovingkindness practice (Hunsinger,
Livingston, & Isbell, 2012) can enhance individuals’ cognitive
control. Cognitive control is generally thought to be important in
regulating automatic intergroup bias, and practice in exerting cog-
nitive control over biases through meditation may enable people to
inhibit the activation of implicit biases against outgroup members
(Monteith, Arthur, & Flynn, 2010). Second, lovingkindness medi-
ation may reduce bias by changing more affective, stress-related
reactivity. Stress can promote cognitive biases such as stereotyping
(e.g., Baron, Inman, Kao, & Logan, 1992; Friedland, Keinan, &
Tytiun, 1999) or implicit prejudice (Frantz, Cuddy, Burnett, Ray,
& Hart, 2004; Terbeck et al., 2012). Given that lovingkindness
meditation decreases stress (Carson et al., 2005) and increases
positive emotions (Fredrickson et al., 2008) associated with adap-
tive stress coping (Fredrickson, Tugade, Waugh, & Larkin, 2003),
lovingkindness mediation may reduce bias by decreasing stress.
Method
Participants
Healthy non-Black, nonhomeless adults (N⫽107) who had no
previous experience with lovingkindness meditation were re-
cruited for compensation (up to $130). Participants were recruited
using flyers advertising for a study on “the effect of meditation on
cognitive and affective functioning” posted in New Haven and
surrounding communities. Participants were randomly assigned to
one of three conditions: lovingkindness practice, lovingkindness
discussion, or waitlist control. Six participants (three from the
lovingkindness practice, two from the lovingkindness discussion,
and one from the waitlist control condition) did not complete the
study. The final sample consisted of 101 participants (65 women;
mean age ⫽25.20 years, SD ⫽5.20; 22 Asian, 62 White, six
Hispanic, one Middle Eastern, and 10 other; see Table 1).
Procedure
Participants completed a pretest within 7 days prior to the
intervention, learned their random condition assignment, attended
the intervention, and completed the posttest within 7 days after the
conclusion of the intervention. Approximately 1 week past the
posttest, lovingkindness meditation courses were offered to all
lovingkindness discussion and waitlist control participants. Tests
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2KANG, GRAY, AND DOVIDIO
were administered in a psychology laboratory, and the intervention
classes were held in a classroom.
The lovingkindness practice intervention consisted of six
weekly lovingkindness meditation classes. Each class lasted for an
hour, with 17–19 participants per group. Each meditation class
included 30-min sitting meditation (modified from Carson et al.,
2005; Fredrickson et al., 2008), checking on participants’ progress
and answering questions, followed by open discussion. Partici-
pants also were assigned to practice lovingkindness meditation at
home using a guided meditation mp3 file (http://marc.ucla.edu/
mpeg/05_Loving_Kindness_Meditation.wma) at least 20 min per
day, 5 days per week.
The lovingkindness discussion intervention was closely
matched to the lovingkindness practice condition and was con-
ducted in an identical classroom setting. The lovingkindness dis-
cussion intervention consisted of six weekly discussion classes.
Each class lasted for 40 min, with 17–19 participants per group.
Prior to attending each class, participants received reading mate-
rials on the ideas behind lovingkindness meditation. Each discus-
sion included a brief presentation of the reading material, open
discussion, and question-and-answer sessions. The lovingkindness
meditation practice and discussion materials are available by re-
quest from Yoona Kang.
1
Discussion participants were explicitly
told not to do any meditation practice throughout the period of
intervention. Neither the lovingkindness practice nor discussion
class involved attempts to negate or inhibit the original negative
group associations, which may ironically enhance bias (Legault,
Gutsell, & Inzlicht, 2011). Instead, participants learned to extend
lovingkindness to all beings without explicitly invoking specific
social groups as target for the loving feelings.
Participants in the waitlist control did not have any further
contact with the instructor or course materials until the posttest.
An accredited meditation instructor with an extensive experi-
ence with practicing and teaching lovingkindness meditation for
over 30 years led all lovingkindness practice and discussion ses-
sions. He was unaware of the study hypotheses and received
monetary compensation for his time. He orally reported that both
discussion and meditation sessions would be beneficial for the
participants, consistent with the Buddha’s teachings. Class atten-
dance was monitored in each class. Each day using a secure online
website, lovingkindness practice participants reported the minutes
of meditation they had engaged in since the last report. In addition,
we controlled for potential differences in class interaction by
having those in the lovingkindness practice and discussion groups
report how much (from 1 ⫽not at all to5⫽extremely often) they
had interacted with other participants in class. Participants also
rated their overall experience in class (from 1 ⫽extremely nega-
tive to 7 ⫽extremely positive).
Measures
Implicit attitudes, explicit attitudes, cognitive control, and stress
were assessed at the beginning and end of the study period. All
participants provided demographic information upon completion
of the posttest.
Standard IAT procedures were used to assess implicit attitudes
toward two stigmatized outgroups, Blacks (vs. Whites) and home-
less persons (vs. college students). The materials and procedure for
the race IAT were based on the task used by Nosek & Smyth
(2007); the materials for the homeless IAT were based on previous
work by Harris and Fiske (2006). IATs were administered using
the DirectRT program (Empirisoft, New York, NY) on a PC
desktop and were scored following the guidelines of Greenwald,
Nosek, and Banaji (2003). In addition, we assessed the favorability
of participants’ explicit attitudes toward homeless people, among
other groups, with a feeling thermometer (from 0 ⫽extremely
negative to 100 ⫽extremely positive).
Cognitive control, one of the potential mediators, was assessed
using the multi-source interference task (MSIT; Bush & Shin,
2006; see Shehzad, DeYoung, Kang, Grigorenko, & Gray, 2012,
for materials and procedural details; ␣⫽.75 in present study). We
measured psychological stress, a second potential mediator, using
the 14-item perceived stress scale (PSS; Cohen, Kamarck, &
Mermelstein, 1983). Participants reported their experience of
1
Contact Yoona Kang at yoona.kang@yale.edu
Table 1
Participant Demographics, Baseline Characteristics, and Intervention-Related Statistics
Variable
Lovingkindness practice
(n⫽35)
Lovingkindness discussion
(n⫽33)
Waitlist control
(n⫽33) Statistic
Demographic
Age in years (SD) 25.69 (5.17) 24.42 (5.06) 25.45 (5.45) F⫽0.55
Female (%) 57.1 63.6 72.7
2
⫽1.81
White (%) 60.0 57.6 66.7
2
⫽3.73
Education in years (SD) 17.37 (2.47) 16.82 (2.70) 16.67 (2.12) F⫽0.79
Baseline characteristics
Cognitive control–MSIT (SD) 314.52 (65.28) 288.94 (94.33) 309.24 (95.06) F⫽0.84
Psychological stress–PSS (SD) 40.23 (7.53) 38.76 (8.09) 37.70 (8.81) F⫽0.83
Intervention-related
Class attendance 4.86 (1.44) 4.91 (1.18) — F⫽0.03
Meditation minutes 553.84 (239.56) — — —
Interaction with others 2.29 (0.80) 2.03 (0.67) — F⫽1.98
Overall experience 5.38 (0.92) 5.20 (0.89) — F⫽0.65
Note. Mean values are displayed with standard deviations in parentheses where applicable. MSIT ⫽multisource interference task; PSS ⫽Perceived Stress
Scale.
ⴱ
p⬍.05.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
LOVINGKINDNESS MEDITATION AND IMPLICIT BIAS CONTROL
stress in the past week (from 0 ⫽never to4⫽very often). Internal
consistency scores across the experiment were high (␣
mean
⫽.89;
␣
range
⫽.88 –.89).
Results
Lovingkindness practice, discussion, and waitlist control partic-
ipants did not significantly differ with respect to age, gender,
ethnicity, or education (ps⬎.40; see Table 1), nor in baseline
measures of cognitive control and psychological stress (ps⬎.40).
Participants in the lovingkindness practice and discussion condi-
tions did not significantly differ in their average attendance (p⫽
.872), the degree to which they interacted with other participants in
class (p⫽.165), or the ratings of their overall experience in class
(p⫽.424).
First, we examined the impact of practicing lovingkindness
mediation on explicit attitudes toward Blacks and homeless peo-
ple. For all participants, pre- and posttest explicit attitude scores
were significantly correlated (rs⫽.569 and .604 for Blacks and
homeless, respectively, ps⬍.001). We predicted that those who
practiced lovingkindness meditation for 6 weeks would show a
significant improvement in explicit attitudes toward Blacks and
homeless people, while those in the other two control conditions,
lovingkindness discussion and waitlist controls, would not (see
Table 2 for means). To test this prediction, we compared posttest
explicit attitude ratings across conditions for both race and home-
less, separately, while treating pretest explicit attitude ratings as
covariates. We tested two planned comparison orthogonal con-
trasts. First, we contrasted the meditation practice condition versus
the average of the lovingkindness discussion and waitlist control
conditions (weighted coefficients ⫽⫹2, ⫺1, ⫺1), and second, we
contrasted the lovingkindness discussion and waitlist control con-
ditions (weighted coefficients ⫽0, ⫹1, ⫺1). There was no sig-
nificant effect for explicit attitudes toward Blacks (ps⬎.50) or
homeless people (ps⬎.50). Across all conditions, explicit atti-
tudes toward homeless people were more negative than those
toward Blacks at both pretest, M⫽52.96, SD ⫽23.83, versus
M⫽65.20, SD ⫽18.23, t(100) ⫽5.185, p⬍.001, and posttest,
M⫽53.64, SD ⫽21.26, versus M⫽65.47, SD ⫽20.21, t(99) ⫽
5.601, p⬍.001. Additionally, we computed explicit attitude
difference scores by subtracting explicit attitude scores for Blacks
from Whites and homeless people from college students, sepa-
rately for pre- and posttests. When we compared posttest differ-
ence scores across conditions while treating pretest difference
scores as covariates, there was no significant effect for explicit
attitudes toward Blacks minus Whites (ps⬎.30) or homeless
people minus college students (ps⬎.60).
Next, we examined the impact of practicing lovingkindness
mediation on implicit attitudes toward Blacks and homeless peo-
ple. Implicit and explicit intergroup attitudes were not correlated
(rs at pre- and posttest between implicit attitude scores and explicit
scores for Blacks ⫽.176, .172, homeless people ⫽.056, .133, for
Blacks minus Whites ⫽.050, .048, and college students minus
homeless people ⫽.010, .026, ps⬎.08). Across all participants,
implicit attitudes toward Blacks and homeless people were posi-
tively correlated at both the pretest, r(101) ⫽.587, p⬍.001, and
at posttest, r(101) ⫽.690, p⬍.001. Initially, at the time of the
Table 2
Descriptive Statistics for the Study Variables
Variable
Lovingkindness practice
(n⫽35)
Lovingkindness discussion
(n⫽33)
Waitlist control
(n⫽33)
Implicit bias (IAT): Blacks
Pre ⫺0.012 (0.29) 0.062 (0.31) 0.092 (0.20)
Post ⫺0.163 (0.33) 0.050 (0.31) 0.096 (0.20)
Statistic F⫽12.692
ⴱⴱ
F⫽0.077 F⫽0.021
Implicit bias (IAT): Homeless
Pre 0.128 (0.23) 0.109 (0.23) 0.173 (0.23)
Post ⫺0.021 (0.29) 0.056 (0.31) 0.149 (0.31)
Statistic F⫽12.442
ⴱⴱ
F⫽2.086 F⫽0.352
Explicit attitude: Blacks
Pre 61.91 (19.73) 65.91 (17.18) 68.45 (17.59)
Post 64.00 (19.81) 65.97 (19.04) 66.55 (22.18)
Statistic F⫽0.645 F⫽0.000 F⫽0.271
Explicit attitude: Homeless
Pre 56.29 (26.58) 47.44 (22.39) 54.73 (22.18)
Post 56.29 (20.45) 51.03 (22.26) 53.36 (21.42)
Statistic F⫽0.000 F⫽0.886 F⫽0.139
Cognitive control (MSIT)
Pre 314.523 (65.28) 288.934 (94.33) 309.237 (95.06)
Post 314.067 (70.99) 295.534 (88.98) 292.776 (93.48)
Statistic F⫽0.002 F⫽0.261 F⫽1.069
Psychological stress (PSS)
Pre 40.229 (7.53) 38.758 (8.09) 37.700 (8.81)
Post 36.314 (8.45) 39.061 (7.51) 40.576 (8.86)
Statistic F⫽6.281
ⴱ
F⫽0.059 F⫽2.959
Note. Mean values are displayed with standard deviations in parentheses where applicable. IAT ⫽Implicit
Association Test; PSS ⫽Perceived Stress Scale; MSIT ⫽multisource interference task.
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4KANG, GRAY, AND DOVIDIO
pretest, participants across groups tended to have negative implicit
attitudes toward Blacks, M⫽0.046, t(100) ⫽1.72, p⫽.089, and
homeless people, M⫽0.136, t(100) ⫽6.00, p⬍.001. There were
no differences across the three experimental conditions on either
the race (p⫽.271) or the homeless (p⫽.510) pretest IATs.
We predicted that those who practiced lovingkindness medita-
tion for 6 weeks would show a significant decline in implicit
prejudice toward Blacks and homeless people, while those in the
other two control conditions, lovingkindness discussion and wait-
list controls, would not. To test this, we compared posttest IAT d
scores across three conditions, while treating pretest IAT dscores
as covariates. Two planned comparison orthogonal contrasts, as
was used in the explicit attitude analyses, were conducted for both
race and homeless IAT dscores, separately. For race IAT, implicit
bias against Blacks decreased as a function of lovingkindness
meditation practice, as predicted. Regression analyses revealed
that those in the lovingkindness practice condition showed signif-
icantly less bias against Blacks at posttest (M⫽⫺0.16, SD ⫽
0.33), compared with those in the lovingkindness discussion (M⫽
0.05, SD ⫽0.31) or waitlist conditions (M⫽0.10, SD ⫽0.20),
⫽⫺.270, t(98) ⫽⫺3.769, p⬍.05. Race IAT scores at posttest
for those in the lovingkindness discussion and waitlist control
conditions did not differ, ⫽⫺.032, t(98) ⫽⫺0.417, p⫽.677.
For homeless IAT, there was a significant effect of lovingkind-
ness meditation practice on implicit bias against homeless people.
Regression analyses revealed that those in the lovingkindness
practice condition showed less bias against homeless people at
posttest (M⫽⫺0.02, SD ⫽0.29) compared with those in the
lovingkindness discussion (M⫽0.06, SD ⫽0.31) or waitlist
conditions (M⫽0.15, SD ⫽0.31), ⫽⫺.173, t(98) ⫽⫺2.314,
p⬍.05. Homeless IAT scores at posttest for those in the loving-
kindness discussion and waitlist control conditions did not differ,
⫽⫺.050, t(98) ⫽⫺0.647, p⫽.519.
We next tested whether the effect of lovingkindness meditation
on implicit biases against Blacks and homeless people was ex-
plained by two potential mediators: cognitive control and psycho-
logical stress. Across conditions, greater increases in cognitive
control were associated with less implicit bias only toward Blacks
at posttest compared with at pretest, r(101) ⫽.202, p⫽.043, but
not significantly so with implicit bias toward homeless people,
r(101) ⫽.087, p⫽.389. Contrary to prediction, however, when
the lovingkindness practice condition was contrasted with the
lovingkindness discussion and waitlist control conditions, there
was no difference in cognitive control at posttest, controlling for
pretest, as a function of meditation conditions, ⫽.062, t(98) ⫽
0.761, p⫽.448. Thus, cognitive control was not plausibly a
mediator of the effect of the intervention on implicit attitudes
toward Blacks and homeless people.
Next, we tested the indirect effect from lovingkindness medita-
tion to improved implicit attitudes toward Blacks and homeless
people with psychological stress as a mediator. There was a
significant effect of lovingkindness meditation practice on stress
reduction, such that controlling for pretest stress ratings, those in
the lovingkindness practice condition reported decreased stress at
posttest (M⫽36.31, SD ⫽8.45), compared with those in the
lovingkindness discussion (M⫽39.06, SD ⫽7.51) or waitlist
conditions (M⫽40.58, SD ⫽8.86), ⫽⫺.249, t(98) ⫽⫺2.766,
p⬍05. Psychological stress at posttest for those in the loving-
kindness discussion and waitlist control conditions did not differ,
⫽⫺.094, t(98) ⫽⫺1.020, p⫽.310.
For the race IAT, the effect of lovingkindness practice on bias
reduction was not mediated by decreased stress. When changes in
psychological stress, condition, and pretest race IAT scores were
considered simultaneously as predictors of implicit attitudes at
posttest, the effect of condition on bias reduction stayed signifi-
cant, ⫽⫺.266, t(97) ⫽⫺3.547, p⫽.001, and changes in stress
did not predict the bias at posttest, ⫽.016, t(97) ⫽0.214, p⫽
.831. For the homeless IAT, supportive of the hypothesized me-
diation, when changes in psychological stress, condition, and pre-
test homeless IAT scores were considered simultaneously as pre-
dictors of implicit attitudes at posttest, condition no longer
predicted improved implicit attitudes, ⫽⫺.128, t(97) ⫽⫺1.668,
p⫽.099, while changes in psychological stress did, ⫽.156,
t(97) ⫽2.024, p⬍.05 —with decreased stress predicting de-
creased bias. Results from the bootstrapping procedure (Figure 1)
supported our prediction: The indirect effect from lovingkindness
meditation to decreased stress to reduced implicit bias against
homeless people was, as predicted, significant (bias-corrected con-
fidence interval [⫺.8360, ⫺.0138], which does not contain 0).
2
Discussion
Practicing lovingkindness meditation significantly decreased
implicit bias toward Blacks and homeless people, whereas discuss-
ing lovingkindness in a similar group setting did not. Although
previous studies have demonstrated the effect of lovingkindness
meditation on increased explicit positivity toward close others
(Fredrickson et al., 2008) and strangers (Hutcherson et al., 2008),
explicit attitudes toward Blacks and homeless people did not
improve as a function of lovingkindness meditation in the current
study. Because lovingkindness meditation practice does not relate
directly to intergroup relations, participants might not have recog-
nized the relevance to their conscious attitudes toward such strong
and highly acceptable form of social bias. Another possibility is
that explicit attitudes toward more categorically distinct groups,
such as racial minorities and homeless people, may be more
resistant to change than groups that are more open to decategori-
zation or recategorization (Nier et al., 2001) such as unacquainted
others. On the positive side, however, the lack of change in explicit
attitudes suggests that participants’ responses were not determined
by demand characteristics, evaluation concerns, or experimenter
expectancy effects. Explicit measures are particularly susceptible
to these influences, and explicit intergroup attitudes did not
change.
Our data indicated, though, that lovingkindness meditation had
greater impact on implicit attitudes, which are automatically acti-
vated, rather than explicit attitudes, which are more controlled by
conscious categorization processes. Implicit and explicit inter-
group attitudes are only weakly correlated generally (Greenwald et
al., 2009) and in the present study, and rely on different cognitive
2
Changes in bias did not mediate the effect of meditation on stress
reduction. When changes in homeless IAT scores, condition, and pretest
stress ratings were considered simultaneously as predictors of post stress
ratings, the effect of Condition on bias reduction stayed significant, ⫽
⫺.240, t(97) ⫽⫺2.598, p⬍.05, and changes in homeless IAT scores did
not predict the posttest stress ratings, ⫽.050, t(97) ⫽0.525, p⫽.601.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
LOVINGKINDNESS MEDITATION AND IMPLICIT BIAS CONTROL
processes (Gawronski, Peters, Brochu, & Strack, 2008; Rydell &
McConnell, 2006). We posit that extended practice in lovingkind-
ness meditation that increases connectedness to others generally
may have directly addressed some of the foundations of implicit
bias against culturally stigmatized groups.
In addition, we found on both the explicit and the implicit (IAT)
measures that bias was stronger against homeless people than
Blacks. Perhaps because of the difference in cultural forces con-
cerning bias against these two groups and the preexisting charac-
teristics of those who are interested in meditation (Tanner et al.,
2009), implicit bias against Blacks was only marginally significant
for all participants at the time of pretest in the present study. Thus,
there may have been more room for attitude change for implicit
bias against homeless people than against Blacks, and the change
from pretest to posttest for the lovingkindness practice condition
was from essentially neutral to positive for implicit attitudes to-
ward Blacks and from negative to neutral for homeless people.
Nevertheless, we note that that the magnitude of impact of prac-
ticing lovingkindness meditation for improving implicit attitudes
was similar for the two groups (
2
for the change from pretest to
posttest for the race IAT ⫽.272 and for the homeless IAT ⫽.268).
Although implicit intergroup attitudes are difficult to change,
growing evidence shows implicit bias reduction through uncon-
scious and effortless activation of countergoal (Sassenberg &
Moskowitz, 2005) or conscious and effortful affirmations of coun-
terstereotypes (Kawakami et al., 2000). However, counterstereo-
types affirmation training to explicitly motivate prejudice reduc-
tion can sometimes ironically increase prejudice (Legault et al.,
2011). Lovingkindness meditation may offer an effective alterna-
tive that may embed improved intergroup attitudes along with a
host of other personal benefits such as stress reduction (Astin,
1997) or mood improvement (Fredrickson et al., 2008), sustaining
improved intergroup orientations without arousing rebound or
backlash effects. We note, however, that we only tested for re-
duced implicit bias shortly after the completion of the study period.
Therefore, the current study does not demonstrate the durability of
the effect of lovingkindness meditation.
We further tested two potential mechanisms of such effect,
using psychological stress and cognitive control as potential me-
diators. Previous research has revealed that IAT scores can be
artificially elevated by stress and anxiety (Frantz et al., 2004;
Terbeck et al., 2012). Thus, one way that practicing lovingkind-
ness meditation might reduce implicit bias is by diminishing stress.
Indeed, decreases in stress mediated the effect of lovingkindness
meditation on implicit bias toward the homeless people, as we
predicted. However, although implicit bias toward homeless peo-
ple and Blacks were significantly correlated, decreases in stress did
not mediate reductions in implicit bias toward Blacks. Thus, al-
ternatives to lovingkindness meditation, such as mere relaxation
exercises, are unlikely to have the robust and generalized effect for
bias reduction that we observed for lovingkindness meditation.
However, the precise mechanism by which practicing lovingkind-
ness meditation reduced implicit bias toward Blacks remains un-
clear.
We hypothesized that lovingkindness meditation would improve
implicit attitudes toward Blacks by increasing cognitive control,
and as predicted, increases in cognitive control were correlated
with reductions in implicit bias toward Blacks but not toward
homeless people. However, cognitive control did not increase with
meditation practice and thus did not qualify as a mediator of the
lovingkindness meditation effect. The lack of increase in cognitive
control as a function of lovingkindness meditation is in contrast
with the previous finding that showed enhanced cognitive control
with lovingkindness meditation practice (Hunsinger et al., 2012).
Attending a 3-day lovingkindness meditation course improved
performances in Stroop task. It is, however, important to note that
the participants’ cognitive control was measured immediately fol-
lowing the lovingkindness meditation practice in the previous
study, likely to have measured more state-like changes in cognitive
control. By contrast, the present study assessed changes in cogni-
tive control 3–7 days after the completion of lovingkindness
courses, likely to have measured more trait-like changes that are
more stable over time.
Although mindfulness meditation has been shown to increase
cognitive control (Tang et al., 2007), lovingkindness meditation
likely involves relatively more affective (vs. cognitive) modes of
processing. Mindfulness meditation and lovingkindness medita-
tion may differentially influence cognitive and affective down-
stream consequences, which then may result in different benefits in
the context of social relations. For example, Desbordes et al.
(2012) reported that mindfulness meditation decreases overall
emotional reactivity to all positive, negative, and neutral stimuli,
indexed by generalized decreases in activity within the right
amygdala. In contrast, compassion meditation, a close sibling of
lovingkindness meditation, increased emotional sensitivity, in-
dexed by heightened amygdala activity, to stimuli depicting human
suffering, a pattern of response previously associated with em-
pathic concern and social emotions. These results suggest that in
the context of social relations, benefits of mindfulness meditation
may be the increased cognitive control over negative emotional
reactivity, whereas lovingkindness meditation may cultivate feel-
ings of connectedness by creating new positive associations. Be-
cause implicit biases toward homeless people have particularly
strong affective components such as disgust (Harris & Fiske,
2006), increasing affective (stress) control may play a particularly
strong role in reducing implicit bias toward them. Cognitive fac-
tors, perhaps related to recognition of societal and personal egal-
itarian goals regarding Blacks (Crandall et al., 2002), may play a
larger role in implicit bias against Blacks, but our measure of
cognitive control may not represent the most pertinent type of
cognitive activity that is central to this process. Cognitive control
is a multifaceted phenomenon, and MSIT, like the Stroop task,
may measure only one aspect of cognitive control related to
automatic response inhibition (Bush & Shin, 2006). Indeed, while
Figure 1. Meditation model. Psychological stress mediated the effect of
6-week lovingkindness mediation practice on implicit bias against home-
less people,
ⴱ
p⬍.05,
ⴱⴱ
p⬍.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6KANG, GRAY, AND DOVIDIO
several plausible processes have been hypothesized, direct evi-
dence of processes that mediate implicit bias, and particularly
changes in implicit bias, is currently limited (Rudman, 2004).
Therefore, we recommend those conducting future studies to use
different measures of cognitive control that assess not only the
ability to inhibit undesirable reactions but also the capacity to
allocate attention as intended.
Future research might also consider the generalizability and
longer term psychological and behavioral consequences of these
effects of practicing lovingkindness meditation. In addition, in
order to test whether these implicit changes will have any impor-
tant real-life consequences, future studies may utilize more robust
sets of behavioral indicators that assess the benefits of lovingkind-
ness meditation in actual intergroup behaviors, such as positivity
of nonverbal behavior in interracial interactions. Furthermore, in
theory, lovingkindness meditation is intended to cultivate feelings
of lovingkindness for all beings, including one’s enemies. There-
fore, in future studies, researchers may also use other motivation-
ally relevant target groups and examine the effects of lovingkind-
ness meditation on attitude changes among Israelis toward
Palestinians, or use reviled social groups that involve strong neg-
ative associations, such as terrorists or pedophiles. Finally, because
lovingkindness meditation involves complex processes, future
studies might focus on other potential mechanisms such as positive
emotions, cognitive flexibility, self-compassion, interpersonal se-
curity, empathy, perspective taking, and changes in individuals’
notions of ingroup as a function of lovingkindness meditation.
Nevertheless, the current research offers direct evidence that lov-
ingkindness meditation can be used as a powerful tool not only for
promoting personal health but also as an intervention for promot-
ing healthy intergroup relations.
References
Astin, J. A. (1997). Stress reduction through mindfulness meditation.
Psychotherapy and Psychosomatics, 66, 97–106. doi:10.1159/
000289116
Bargh, J. A., Gollwitzer, P. M., Lee-Chai, A., Barndollar, K., & Trotschel,
R. (2001). The automated will: Nonconscious activation and pursuit of
behavioral goals. Journal of Personality and Social Psychology, 81,
1014 –1027. doi:10.1037/0022-3514.81.6.1014
Barnes, P. M., Bloom, B., & Nahin, R. (2008). Complementary and
alternative medicine use among adults and children: United States, 2007.
National Health Statistics Report, 12, 1–23.
Baron, R. S., Inman, M. L., Kao, C. F., & Logan, H. (1992). Negative
emotion and superficial social processing. Motivation and Emotion, 16,
323–346. doi:10.1007/BF00992970
Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. D., Car-
mody, J.,...Devins, G. (2004). Mindfulness: A proposed operational
definition. Clinical Psychology: Science and Practice, 11, 230 –241.
Bush, G., & Shin, L. M. (2006). The multi-source interference task: An
fMRI task that reliably activates the cingulo-frontal-parietal cognitive/
attention network. Nature Protocols, 1, 308 –313. doi:10.1038/nprot
.2006.48
Carson, J. W., Keefe, F. J., Lynch, T. R., Carson, K. M., Goli, V., Fras,
A. M., & Thorp, S. R. (2005). Loving-kindness meditation for chronic
low back pain: Results from a pilot trial. Journal of Holistic Nursing, 23,
287–304. doi:10.1177/0898010105277651
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of
perceived stress. Journal of Health and Social Behavior, 24, 385–396.
doi:10.2307/2136404
Crandall, C. S., Eshleman, A., & O’Brien, L. (2002). Social norms and the
expression and suppression of prejudice: The struggle for internalization.
Journal of Personality and Social Psychology, 82, 359 –378. doi:
10.1037/0022-3514.82.3.359
Desbordes, G., Negi, L. T., Pace, T. W. W., Wallace, B. A., Raison, C. L.,
& Schwartz, E. L. (2012). Effects of mindful-attention and compassion
meditation training on amygdala response to emotional stimuli in an
ordinary, non-meditative state. Frontiers in Human Neuroscience, 6,
292. doi:10.3389/fnhum.2012.00292
Evans, S., Ferrando, S., Findler, M., Stowell, C., Smart, C., & Haglin, D.
(2008). Mindfulness-based cognitive therapy for generalized anxiety
disorder. Journal of Anxiety Disorders, 22, 716 –721. doi:10.1016/j
.janxdis.2007.07.005
Frantz, C. M., Cuddy, A. J., Burnett, M., Ray, H., & Hart, A. (2004). A
threat in the computer: The race implicit association test as a stereotype
threat experience. Personality and Social Psychology Bulletin, 30, 1611–
1624. doi:10.1177/0146167204266650
Fredrickson, B. L., Coffey, K. A., Pek, J., Cohn, M. A., & Finkel, S. M.
(2008). Open hearts build lives: Positive emotions, induced through
loving-kindness meditation, build consequential personal resources.
Journal of Personality and Social Psychology, 95, 1045–1062. doi:
10.1037/a0013262
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, 365–376. doi:10.1037/0022-3514.84.2.365
Friedland, N., Keinan, G., & Tytiun, T. (1999). The effect of psychological
stress and tolerance of ambiguity on stereotypic attributions. Anxiety,
Stress & Coping, 12, 397– 410. doi:10.1080/10615809908249318
Gawronski, B., Peters, K. R., Brochu, P. M., & Strack, F. (2008). Under-
standing the relations between different forms of racial prejudice: A
cognitive consistency perspective. Personality and Social Psychology
Bulletin, 34, 648 – 665. doi:10.1177/0146167207313729
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding
and using the Implicit Association Test: I. An improved scoring algo-
rithm. Journal of Personality and Social Psychology, 85, 197–216.
doi:10.1037/0022-3514.85.2.197
Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R.
(2009). Understanding and using the Implicit Association Test: III.
Meta-analysis of predictive validity. Journal of Personality and Social
Psychology, 97, 17– 41. doi:10.1037/a0015575
Hall, P. D. (1999). The effect of meditation on the academic performance
of African American college students. Journal of Black Studies, 29,
408 – 415. doi:10.1177/002193479902900305
Harris, L. T., & Fiske, S. T. (2006). Dehumanizing the lowest of the low:
Neuroimaging responses to extreme out-groups. Psychological Science,
17, 847– 853. doi:10.1111/j.1467-9280.2006.01793.x
Hunsinger, M., Livingston, R., & Isbell, L. (2012). The impact of loving-
kindness meditation on affective learning and cognitive control. Mind-
fulness. Advance online publication. doi:10.1007/s12671-012-0125-2
Hutcherson, C. A., Seppala, E. M., & Gross, J. J. (2008). Loving-kindness
meditation increases social connectedness. Emotion, 8, 720 –724. doi:
10.1037/a0013237
Kang, Y., Gruber, J., & Gray, J. R. (2013). Mindfulness and de-
automatization. Emotion Review, 5, 192–201. doi:10.1177/
1754073912451629
Kawakami, K., Dovidio, J. F., Moll, J., Hermsen, S., & Russin, A. (2000).
Just say no (to stereotyping): Effects of training in the negation of
stereotypic associations on stereotype activation. Journal of Personality
and Social Psychology, 78, 871– 888. doi:10.1037/0022-3514.78.5.871
Legault, L., Gutsell, J., & Inzlicht, M. (2011). Ironic effects of anti-
prejudice messages: How motivational interventions can reduce (but
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
LOVINGKINDNESS MEDITATION AND IMPLICIT BIAS CONTROL
also increase) prejudice. Psychological Science, 22, 1472–1477. doi:
10.1177/0956797611427918
Monteith, M. J., Arthur, S. A., & Flynn, S. M. (2010). Self-regulation and
bias. In J. F. Dovidio, M. Hewstone, P. Glick, & V. M. Esses (Eds.),
Sage handbook of prejudice, stereotyping, and discrimination (pp. 493–
507). London, United Kingdom: Sage. doi:10.4135/9781446200919.n30
Moskowitz, G. B., & Ignarri, C. (2009). Implicit volition and stereotype
control. European Review of Social Psychology, 20, 97–145. doi:
10.1080/10463280902761896
Nier, J. A., Gaertner, S. L., Dovidio, J. F., Banker, B. S., Ward, C. M., &
Rust, M. C. (2001). Changing interracial evaluations and behavior: The
effects of a common group identity. Group Processes & Intergroup
Relations, 4, 299 –316. doi:10.1177/1368430201004004001
Nosek, B. A., & Smyth, F. L. (2007). A multitrait–multimethod validation
of the Implicit Association Test: Implicit and explicit attitudes are
related but distinct constructs. Experimental Psychology, 54, 14 –29.
doi:10.1027/1618-3169.54.1.14
Olson, K. R., Dweck, C. S., Spelke, E. S., & Banaji, M. R. (2011).
Children’s responses to group-based inequalities: Perpetuation and rec-
tification. Social Cognition, 29, 270 –287. doi:10.1521/soco.2011.29.3
.270
Rudman, L. A. (2004). Sources of implicit attitudes. Current Directions in
Psychological Science, 13, 79 – 82. doi:10.1111/j.0963-7214.2004
.00279.x
Rudman, L. A., & Lee, M. R. (2002). Implicit and explicit consequences of
exposure to violent and misogynous rap music, Group Processes &
Intergroup Relations, 5, 133–150. doi:10.1177/1368430202005002541
Rydell, R. J., & McConnell, A. R. (2006). Understanding implicit and
explicit attitude change: A systems of reasoning analysis. Journal of
Personality and Social Psychology, 91, 995–1008. doi:10.1037/0022-
3514.91.6.995
Sassenberg, K., & Moskowitz, G. B. (2005). Do not stereotype, think
different! Overcoming automatic stereotype activation by mindset prim-
ing. Journal of Experimental Social Psychology, 41, 506 –514. doi:
10.1016/j.jesp.2004.10.002
Shehzad, Z., DeYoung, C. G., Kang, Y., Grigorenko, E. L., & Gray, J. R.
(2012). Interaction of COMT val
158
met and externalizing behavior:
Relation to prefrontal brain activity and behavioral performance. Neu-
roImage, 60, 2158 –2168. doi:10.1016/j.neuroimage.2012.01.097
Tang, Y. Y., Ma, Y., Wang, J., Fan, Y., Feng, S., Lu, Q.,...Posner, M. I.
(2007). Short-term meditation training improves attention and self-
regulation. Proceedings of the National Academies of Science of the
United States of America, 104, 17152–17156.
Tanner, M. A., Travis, F., Gaylord-King, C., Haaga, D. A., Grosswald, S.,
& Schneider, R. H. (2009). The effects of the transcendental meditation
program on mindfulness. Journal of Clinical Psychology, 65, 574 –589.
doi:10.1002/jclp.20544
Teasdale, J. D., Segal, Z. V., Williams, J. M., Ridgeway, V. A., Soulsby,
J. M., & Lau, M. A. (2000). Prevention of relapse/recurrence in major
depression by mindfulness-based cognitive therapy. Journal of Consult-
ing and Clinical Psychology, 68, 615– 623. doi:10.1037/0022-006X.68
.4.615
Terbeck, S., Kahane, G., McTavish, S., Savulescu, J., Cowen, P. J., &
Hewstone, M. (2012). Propranolol reduces implicit negative racial bias.
Psychopharmacology, 222, 419 – 424. doi:10.1007/s00213-012-2657-5
Received February 16, 2013
Revision received July 16, 2013
Accepted July 17, 2013 䡲
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8KANG, GRAY, AND DOVIDIO
