Content uploaded by Matthew Jordan
Author content
All content in this area was uploaded by Matthew Jordan on Jun 05, 2017
Content may be subject to copyright.
Are Empathy and Concern Psychologically Distinct?
Matthew R. Jordan, Dorsa Amir, and Paul Bloom
Yale University
Researchers have long been interested in the relationship between feeling what you believe others feel—often
described as empathy—and caring about the welfare of others—often described as com- passion or concern.
Many propose that empathy is a prerequisite for concern and is therefore the ultimate motivator of prosocial
actions. To assess this hypothesis, the authors developed the Empathy Index, which consists of 2 novel scales,
and explored their relationship to a measure of concern as well as to measures of cooperative and altruistic
behavior. A series of factor analyses reveal that empathy and concern consistently load on different factors.
Furthermore, they show that empathy and concern motivate different behaviors: concern for others is a
uniquely positive predictor of prosocial action whereas empathy is either not predictive or negatively
predictive of prosocial actions. Together these studies suggest that empathy and concern are psychologically
distinct and empathy plays a more limited role in our moral lives than many believe.
Keywords: empathy, concern, prosocial motivation, cooperation, altruism
Supplemental materials: http://dx.doi.org/10.1037/emo0000228.supp
The great scholars of the Scottish
Enlightenment, Adam Smith and David
Hume, argued for the psychological and
moral centrality of what they called
sympathy but which is now often known as
empathy—the capacity to feel what you infer
others are feeling (Hume, 1978; Smith,
1759). I watch you step on a piece of glass
and I wince; I imagine my friend receiving a
job offer, and I feel pleasure myself. This can
be triggered by being in the presence of
another, as in “emotional contagion,” but it
can also be generated through an act of the
imagination.
There is an extensive body of research taken
to show that such empathic reactions
motivate cooperation, helping, and moral
thoughts and actions more generally (Batson,
2010; Hoffman, 2008; Toi & Batson, 1982;
Zaki, 2016), and there is a large body of
developmental and comparative research
exploring the view that empathy lies at the
root of compassion and morality (Baron-
Cohen, 2002; F. De Waal, 2010).
However, there has also been a backlash
against empathy (see Bloom, 2014, in press,
for discussion). Some philosophers have
argued that empathy is neither necessary nor
sufficient for prosocial behavior (e.g., Prinz,
2011a), while some psychologists have
argued that empathy and compassion are
cognitively and neuro- logically distinct
(DeSteno, 2015; Shamay-Tsoory, Aharon-
Peretz, & Perry, 2009; Singer & Klimecki,
2014), and that it’s compas- sion—a more
distanced concern for others—that is a
primary motivator of moral behavior. These
are the issues we explore here.
One issue that arises in any discussion of
empathy is terminological. Batson (2010)
listed eight different meanings of the term;
Decety and Cowell (2014) noted that it is
used to describe everything “from yawning
contagion in dogs, to distress signaling in
chickens, to patient-centered attitudes in
human medicine.”; whereas De Vignemont
and Singer (2006) suggested “there are
probably nearly as many definitions of
Jordan, Amir, and Bloom
2"
empathy as people working on this topic.”
For instance, some investigators, but not
others, include the capacity for perspective
taking and understanding the mental states of
others as “empathy” and some investigators,
but not others, distinguish empathy from
compassion. In Batson’s influential work,
empathy—or empathic concern—is used to
refer to “other- oriented emotion elicited by
and congruent with the perceived welfare of
someone in need” (Batson, 2011).
Our own usage of the term is narrow. As
others have noted (Prinz, 2011b), Batson’s
definition brings together two concepts that
may or may not be the same psychological
construct; namely, (a) concern for the welfare
of others and (b) emotional congruence
(which could be simple valence matching, a
more specific emotional contagion, or the
product of a more complex inferential
process). To mark this distinction, we follow
other psychologists (e.g., Eisenberg &
Strayer, 1987) and philosophers (e.g.,
Darwall, 1998) in defining empathy as
described above—feeling what others feel—
which is conceptually distinct from concern
or compassion. But our interest here is
empirical, not terminological; some- one
wedded to a different definition of the term
should feel comfortable recasting our project
as exploring the relationship between two
different aspects of empathy, as opposed to
empathy and compassion.
One way to address the role of empathy in
motivating prosocial behavior is to look at
the consequences of individual differences in
empathy, and there is considerable research
along these lines (e.g., Eisenberg & Strayer,
1987; Eisenberg & Fabes, 1990; Hoffman,
AQ: 5 2008). One problem here, however, is
that the most common empathy scales aren’t
valid measures of empathy in the specific
sense described here—in the sense of an
individual sharing the inferred feelings of a
target. Baron-Cohen’s Empathy Quotient
scale (Baron-Cohen & Wheelwright, 2004),
for instance, has items that some might
construe as being related to some sense of
empathy, such as “I find it easy to put myself
in somebody else’s shoes” and “Seeing
people cry doesn’t really upset me” (reverse-
coded), but it also includes items that pertain
to social adroitness, such as “I can easily tell
if someone else wants to enter a
conversation” and “I find it hard to know
what to do in a social situation” (reverse-
coded). One’s score on such a scale thereby
reflects multiple underlying social– cognitive
and emotional processes, not just empathy.
In the studies we report here, we introduce
the Empathy Index, which consists of two
new scales—one focusing on empathy in the
strict sense of emotional contagion, the other
focusing on the tendencyfor behavioral
contagion, often seen as related to empathy
(see Table 1 below for items). The empathy
subscale contains items that generally track
the extent to which participants tend to feel
what those around them are feeling. The
behavioral contagion scale asks about the
extent to which participants might do what
those around them are doing. Although
empathy and behavioral contagion (as we
define them here) share the feature of being
automatically triggered, they differ in that
empathy tracks and produces affect (e.g.,
someone else’s sadness) and behavioral
contagion tracks and produces behavior (e.g.,
yawning at the sight of a yawn).
Here, we explore the relationship between the
two novel sub- scales of the Empathy Index
and the four subscales of the Interpersonal
Reactivity Index (IRI; Davis, 1983), which is
the most widely used measure of empathy in
the broad sense that captures several aspects
of social cognition. These subscales, each
with seven items, include perspective-taking,
tailored to capture people’s interest in taking
Jordan, Amir, and Bloom
3"
the perspectives of others (sometimes called
“cognitive empathy”); fantasy, their tendency
to identify with fictional characters; personal
distress, which measures how anxious or
distressed people get during emergencies
(and has occasionally been used as a measure
of empathy in the sense in which we are
using it [e.g., Decety & Yoder, 2016]); and
empathic concern, which focuses on feelings
for others.
Although empathic concern is occasionally
viewed as an empathy scale in a more narrow
sense (as reflected in its name), it can also be
seen as a more general measure of concern
for the welfare of others, as it includes items
that assess how much one cares about other
people, without addressing empathy in a
direct sense. For instance, items include “I
am often quite touched by things that I see
happen,” “I care for my friends a great deal,”
and “I feel sad when I see a lonely stranger in
a group.” In our studies below, then, we refer
to empathic concern simply as concern.
In Studies 1–3, we explore the connections
between these sub- scales to assess the
relationship between our new measures of
empathy (in the narrow sense) and behavioral
contagion, and the existing subscales by
Davis (1983), looking particularly at the
relationship between the novel measures of
the Empathy Index and his measure of
concern. Many distinct conceptual groupings
of these social–cognitive abilities and
propensities have been pro- posed, and we
hope to shed light on which cluster together
and which do not through the use of these six
subscales. In Studies 2–3, we also explore the
relationship between these scales and two
prosocial behaviors: cooperation (Study 2)
and altruism (Study 3).
TABLE 1
Individual Items from the Four Subscales of the Interpersonal Reactivity Index (IRI)
Plus the Two Scales of the Empathy Index
Interpersonal Reactivity Index (IRI)
Perspective Taking subscale
I sometimes find it difficult to see things from the "other guy's" point of view.
I try to look at everybody's side of a disagreement before I make a decision.
I sometimes try to understand my friends better by imagining how things look from their perspective.
If I'm sure I'm right about something, I don't waste much time listening to other people’s arguments.
I believe that there are two sides to every question and try to look at them both.
When I'm upset at someone, I usually try to "put myself in his shoes" for a while.
Before criticizing somebody, I try to imagine how I would feel if I were in their place.
Fantasy subscale
I really get involved with the feelings of the characters in a novel.
I am usually objective when I watch a movie or play, and I don't often get completely caught up in it.
Becoming extremely involved in a good book or movie is somewhat rare for me.
After seeing a play or movie, I have felt as though I were one of the characters.
When I watch a good movie, I can very easily put myself in the place of a leading character.
When I am reading an interesting story or novel, I imagine how I would feel if the events in the story were
happening to me.
I daydream and fantasize, with some regularity, about things that might happen to me.
Jordan, Amir, and Bloom
4"
Concern (Empathic Concern) subscale
I often have tender, concerned feelings for people less fortunate than me.
Sometimes I don’t feel very sorry for other people when they are having problems.
When I see someone being taken advantage of, I feel kind of protective towards them.
Other people’s misfortunes do not usually disturb me a great deal.
When I see someone being treated unfairly, I sometimes don’t feel very much pity for them.
I am often quite touched by things that I see happen.
I would describe myself as a pretty soft-hearted person.
Personal Distress subscale
In emergency situations, I feel apprehensive and ill-at-ease.
I sometimes feel helpless when I am in the middle of a very emotional situation.
When I see someone get hurt, I tend to remain calm.
Being in a tense emotional situation scares me.
I am usually pretty effective in dealing with emergencies.
I tend to lose control during emergencies.
When I see someone who badly needs help in an emergency, I go to pieces.
Novel subscales
Empathy subscale
If I see someone who is excited, I will feel excited myself.
I sometimes find myself feeling the emotions of the people around me, even if I don’t try to feel what
they’re feeling.
If I’m watching a movie and a character injures their leg, I will feel pain in my leg.
If I hear a story in which someone is scared, I will imagine how scared I would be in that situation and
begin to feel scared myself.
If I hear an awkward story about someone else, I might feel a little embarrassed.
I can’t watch shows in which an animal is being hunted, because I feel nervous as if I am being hunted.
If I see someone fidgeting, I’ll start feeling anxious too.
Behavioral Contagion subscale
If I see someone else yawn, I am also likely to yawn.
If I see someone vomit, I will gag.
I catch myself crossing my arms or legs just like the person I’m talking to.
If I see a video of a baby smiling, I find myself smiling.
If I see someone suddenly looking away, I’ll automatically look in the direction they are looking.
If I’m watching someone walking on a balance beam, I will lean when they lean.
If I’m having a conversation with someone and they scratch their nose, I will also scratch my nose.
Note. Four subscales from the original IRI are included plus the two novel subscales we developed to measure
empathy and behavioral contagion.
Jordan, Amir, and Bloom
5"
Study 1
Methods
We recruited 95 participants (46 males, Mage
36.52 years) through Amazon’s Mechanical
Turk to complete the six subscales described
above. To participate, Mechanical Turk
workers had to live in the United States and
have a 90% approval rating. We randomly
presented the 42 items from the six subscales
(see Table 1).
There is no consensus on how to determine
the appropriate sample size for a factor
analysis, although rules of thumb typically
use the participants-to-items ratio. Using the
composite data from Studies 1–3, we have a
participants-to-items ratio of greater than 10
to 1, which is higher than most published
factor analysis findings (Costello & Osborne,
2005). Furthermore, we separately analyzed
the factor structure in our three studies to
determine whether or not the structure
replicated across samples.
Results
All six subscales were highly reliable (all
Cronbach’s alpha .71); therefore, we
performed a factor analysis on the six
subscales composite scores (see online
supplement for item-level analysis, which is
nearly identical). We factor analyzed the
subscale composite scores using the principal
factor method and oblique promax rotation
because it is plausible that all other-regarding
tendencies and social–cognitive abilities are
correlated with one another and we wanted to
allow for the factors to correlate if necessary
(Russell, 2002). That said, the results of the
factor analysis hold using different factor and
rotation methods (see online supplement).
We extracted two factors: Factor 1 contained
our empathy and behavioral contagion
subscales as well as the personal distress
subscale from the IRI, whereas Factor 2
contained the perspective taking and concern
subscales from the IRI. The fantasy subscale
was the only complex subscale, loading
similarly and nearly significantly on both
factors. Given the contents of each factor, we
refer to Factor 1 as the Contagion factor and
Factor 2 as the Other- regarding factor.
Figure 1 below shows the factor loadings
plot.
As depicted in Figure 1, there were three
main findings of interest. First, the two novel
scales of the Empathy Index—empathy and
behavioral contagion—were highly related to
one another, and, more interestingly, were
also related to Davis’s (1983) sub- scale of
personal distress. Other scholars, such as
(Schroeder, Dovidio, Sibicky, Matthews, &
Allen, 1988) have also claimed that empathy
and personal distress may be importantly
related.
Second, and more interesting, concern was
not related to either of our two new
measures; individuals who are prone to feel
the experiences of others, as measured by the
Empathy Index sub- scales, were not
particularly likely to feel concern for others,
as assessed by Davis’s (1983) concern scale.
Finally, to our surprise, concern was related
to perspective taking, the capacity to reason
about the thoughts and feelings of others, and
we will elaborate on this in the general
discussion. We should note that although
concern and perspective taking load on the
same factor, it doesn’t follow that they are
the same construct or process (and the same
Jordan, Amir, and Bloom
6"
goes for the subscales on the other factor).
After all, there are many reasons two
processes might be correlated other than
using the same cognitive machinery. Rather
than using this factor analysis to try to
determine which apparently distinct
constructs are actually the same (i.e., load on
the same factor), our goal was to see which
constructs are not the same (i.e., load on
different factors).
Our main question of interest here is how
empathy, in the narrow sense explored in the
Empathy Index is related to concern and
prosocial behavior. In Studies 2 and 3, we
use these measures to explore how empathy
and concern give rise (or fail to give rise) to
pro-social behavior. Study 2 examines the
emotional underpinnings of cooperation
during a Public Goods Game in which
individuals can pay a cost for the benefit of
the group. Study 3 explores the role of
empathy and concern in a situation in which
individuals pay a cost for the benefit of
another who cannot reciprocate.
Figure 1: The graph above shows the factor loadings for each of the six subscales after
performing a factor analysis using the principal factor method and oblique promax rotation. See
the online article for the color version of this figure.
Jordan, Amir, and Bloom
7"
Study 2
It is often argued that empathy can motivate
prosocial behavior, but the evidence here is
somewhat mixed. There is some evidence
suggesting that individuals who score high on
standard empathy measures are more
cooperative, but there is also considerable
evidence to the contrary (see Prinz, 2011b,
for a critical review). Furthermore, studies
that explore individual differences typically
employ scales such as those adopted by
Davis (1983) and Baron- Cohen &
Wheelwright (2004), and, as we’ve seen,
such scales blur together empathy and
concern/compassion. And so a positive
correlation between these measures and
prosocial behavior might reflect a causal role
of these other sentiments, not empathy.
There is much stronger evidence that certain
empathy-related inductions can lead people
to be more generous and cooperative to
others (see Batson, 2010, for review). But
even here, it’s unclear precisely what is
causing the effect. It’s an open question as to
whether it is an Adam Smith-like effect of
feeling what others are feeling that drives
prosocial behavior or a more general
compassion and concern for others. For
instance, in Batson, Batson, et al. (1995),
subjects were more likely to endorse
providing a medical treatment for a young
girl, even though others were ahead of her in
line, when they were given the prompt:
“Try to imagine how the child who is interviewed
feels about what has happened and how it has
affected this child’s life. Try to feel the full impact
of what this child has been through and how he or
she feels as a result.”
The effect of this prompt might be due to a
narrow influence of empathy, where subjects
feel what they imagine to be the child’s
pain—but its effect might also result from
participants thinking more about her mental
states (without necessarily feeling them) or
feeling an elevated concern for the girl
(without, again, feeling what she is feeling).
We begin to explore these issues in Study 2
by assessing the relationship between the IRI,
our Empathy Index, and cooperation in the
Public Goods Game (PGG; Dawes, 1980), a
social dilemma in which individuals can
contribute to a group pot in such a way that it
is group-beneficial, but individually costly to
contribute. Prior work has suggested that the
sort of inductions described above may
promote cooperation in these kinds of
laboratory social dilemmas (Batson, Batson,
et al., 1995; Batson & Moran, 1999). If it is
the case that empathy in the narrow sense is
the motivation for cooperation, we should
find that our empathy scale is particularly and
uniquely predictive of contributions in the
PGG. However, if a general concern for
others is the primary motivation for
cooperating, we should see that concern, as
captured in Davis’s (1983) measure of
empathic concern—is uniquely predictive of
contributions to the public good.
Methods
We recruited 193 participants (109 male,
Mage 38.11 years) from Amazon’s
Mechanical Turk. To participate, Mechanical
Turk workers had to live in the United States,
have a 90% approval rating, and could not
have participated in Study 1 (or other studies
like this one that we have run in the past).
Participants read instructions for the PGG
and completed comprehension questions
before being matched with other real
Mechanical Turk workers and making their
contribution decision. After making their
PGG decision, participants completed the IRI
and the Empathy Index subscales, after which
the study concluded.
Jordan, Amir, and Bloom
8"
In our PGG (Rand, 2012), participants were
placed into groups of four and endowed with
40 cents each, from which they could
contribute whatever portion they wanted to a
common pot. The experimenter then doubled
the total common pot contributions and
evenly dispersed the common pot to all the
group members. We made it clear to
participants that for each cent they
contributed, they would only receive [1/2]
cent back (see online supplement for
instructions and comprehension questions).
The interesting thing about the PGG is that if
everyone donates, everyone gains, but any
individual can gain more by withholding. For
instance, if everyone contributes their entire
endowment, each individual would receive
twice what he or she started with— 80
cents— but if three members of a group
contribute their entire endowment and the
fourth member contributes nothing, those
who contributed everything receive 60 cents
each whereas the individual who contributed
nothing receives 100 cents. It is always
payoff maximizing, then, to contribute
nothing to the group pot. Participants played
the PGG with real other group members for
real stakes; no deception was used.
We assumed the smallest meaningful
correlation between sub- scales and
cooperation was approximately 0.2. To have
90% power to detect a correlation of that
size, we needed about 200 participants.
Results
Here we report results for those who
completed the comprehension check, but we
find similar results for all analyses reported
below if we include those 29 participants
who failed this check (see online supplement
for details). For ease of interpretation, we use
the fraction of the endowment contributed to
the public good as our dependent measure.
The mean contribution to the public good
was 51.7% of the initial endowment
confidence interval (CI): [46.8, 59.7], with
64.8% of the subjects contributing something
and 44.6% contributing their entire
endowment. In separate regressions, we
found that concern (b 0.144, CI: [0.065,
0.224], p .001, R2 0.074) was strongly
positively predictive of cooperation, as was
empathy (b 0.114, CI [0.009, 0.219, p .033,
R2 0.028). However, when both concern (b
0.129, CI [0.044, 0.215], p .003) and empathy
(b 0.054, CI [ 0.056, 0.164], p .337) were
included in the same model (R2 0.079), only
concern predicted contributions to the public
good. When we included all six subscales,
concern (b 0.134, CI [0.026, 0.242], p .015)
was the only subscale that remained
significantly predictive (R2 0.086, all other ps
0.315; we found no evidence of
multicollinearity, see online supplement for
details of multicollinearity diagnostics). We
obtained similar results when we fit separate
regression models predicting cooperation
using the Contagion (b 0.097, CI [0.012,
0.182], p .025, R2 0.031) and Other-regarding
(b 0.135, CI [0.049, 0.220, p .002, R2 0.056)
factors we extracted in Study 1. As above,
when both factors were included (R2 0.062),
the Other-regarding factor (b 0.113, CI
[0.016, 0.209], p .022) remained predictive,
while the Contagion factor did not (b 0.046,
CI [ 0.049, 0.141], p .343).
As in Study 1, we performed a factor analysis
on our six subscales to assess the replicability
of the factor structure. Usingthe same
principal factor method and oblique rotation,
we found identical results (shown below in
Figure 2). Empathy, behavioral contagion,
and personal distress loaded on one factor,
whereas concern and perspective taking
loaded on the other factor, and fantasy loaded
on both factors. Hence, we replicated the
main finding of our initial factor analysis.
Jordan, Amir, and Bloom
9"
The unique predictive power of concern (it
accounted for 76% of the total variance
explained by all six subscales) suggests that,
at least in our study, the level of concern for
others is more relevant to cooperation during
the PGG than feeling what others are feeling.
But there are other interpretations of our data.
For one thing, the one-shot and anonymous
PGG we used isn’t a particularly emotionally
evocative task. This implies that, although we
did find a sensible relationship between
concern and cooperation, such a relationship
is susceptible to explanation in terms of a
third variable. Perhaps being in a cooperative
environment leads both to more cooperation
and greater concern for those around you.
Relatedly, this study was entirely
correlational, which makes interpreting the
causal relationship between concern and
cooperation very difficult. Consequently, in
Study 3, we aim to examine the causal role
between prosocial behaviors and empathy
and concern, by experimentally manipulating
the salience of the target in need.
Figure 2: The factor structure from Study 1 replicated with a distinct sample in Study 2. The factor loadings
above are the result of principal factor analysis and promax rotation. See the online article for the color version
of this figure.
Jordan, Amir, and Bloom
10"
Study 3
In Study 3 we turn our attention to altruism—
paying a cost for another’s benefit without the
possibility of reciprocation (Trivers, 1971)—
which has been the behavior most focused on
by empathy researchers (Batson, 2010; Batson,
Klein, Highberger, & Shaw, 1995; Batson et
al., 1997; F. De Waal, 2010; F. B. De Waal,
2008; Preston & De Waal, 2002; Toi &
Batson, 1982). To explore this, we shift the
focus to charitable donations, and
experimentally manipulate how emotionally
evocative the target of donations is. The
individual differences we have been measuring
with the six subscales reflect a kind of
sensitivity to different kinds of social
emotional stimuli. Here, this means that more
emotionally salient targets of giving should
induce stronger responses among individuals
who score high on the relevant subscales—
whether that is concern, empathy, or another
subscale. We used the identifiable victim
paradigm (Small & Loewenstein, 2003; Small,
Loewenstein, & Slovic, 2007) as a means of
presenting targets of giving that would induce
different degrees of emotional responses.
Previous work on the identifiable victim
paradigm has found increased generosity when
individuals are exposed to a specific individual
in need, with how emotionally evocative a
salient target is being one of the primary
difference between an identifiable victim and
statistics. Specifically, this is often seen as
reflecting an empathic response—being
exposed to a specific individual motivates one
to imagine what that individual feels, which
prompts generous behavior. If so, when
exposed to an identifiable victim, one might
expect to find that those high on our measure
of empathy will be more likely act
altruistically because identifiable victims are
empathy-inducing targets and empathy
motivates prosocial behavior (see De Waal,
2008, for this argument).
In contrast, to the extent that a more general
form of concern for others is relevant,
identifiable victims are concern-inducing, and
concern motivates pro-social behavior, we
might expect those high on concern to be more
generous in the presence of an identifiable
victim. By taking advantage of the fact that
identifiable victims are more emotionally
evocative than statistics about suffering, but
that both pieces of information motivate
altruistic giving, we will be able to tease apart
whether empathy or concern is more relevant
for motivating prosocial actions.
Methods
Based on Study 2 in which the correlations
were closer to r .3, we determined that we
would need about 100 participants in each
condition, for a total of about 200 participants,
to have 90% power to detect similar
correlations in Study 3.
We recruited 192 participants (110 male, Mage
34.73) through Amazon’s Mechanical Turk.
To participate, Mechanical Turk workers had
to live in the United States, have a 90%
approval rating, and could not have
participated in Study 1 or 2 (or other studies
like this one that we have run in the past).
Participants were randomly assigned to either
the Identifiable Victim condition—in which
they read details about one suffering and
emotionally evocative child in the developing
world—or the Statistics condition—in which
they read about a number of poverty and
disease statistics in the developing world. We
followed the identifiable victim protocol from
Small et al. (2007) exactly, except that the
statistics were updated to reflect current data
and the name and photo of the identifiable
victim were updated to one currently listed on
the Save the Children charity website.
Jordan, Amir, and Bloom
11"
Participants were given an endowment of 30
cents and told that they could give however
much they wanted to a charitable cause and
keep the rest as bonus; what participants kept
for themselves was sent as bonuses, while
what they gave was sent to the charity. After
they made their donation decision, participants
answered five questions about the charitable
cause (as in previous identifiable victim
studies): (a) How upsetting is this situation to
you? (b) How sympathetic did you feel while
reading the description of the cause? (c) How
much do you feel it is your moral
responsibility to help out with this cause? (d)
How touched were you by the situation
described? (e) To what extent do you feel that
it is appropriate to give money to aid this
cause?
These five questions that participants answered
about the charitable cause track a set of
feelings and motivations that are not all
obviously either concern or empathy—with the
exception of Question 2, which directly taps
sympathy. For instance, the first question
“How upsetting is this situation to you?” could
track how concerned participants were about
the identifiable victim or how negatively
participants felt because of their empathy for
the individual in need.
After making their donation decision and
answering the five feelings questions,
participants completed the four subscales of
the IRI and the two subscales of the Empathy
Index mentioned above.
Results
The overall mean contribution was 35.9% CI
[30.3%, 41.6%] of the $0.30 endowment, with
57.3% donating something and 21.4% giving
their entire endowment. We first regressed
donations on all of the subscales and found
only concern (b 0.200, CI [0.119, 0.282], p
.001) and empathy were predictive (b 0.106,
CI [ 0.219, 0.007], p .066, R2 0.136, all other
ps 0.381). Therefore, we focused further
analysis on the subscales of interest.
We examined the main effect of condition,
(mean-centered) concern scores and (mean-
centered) empathy scores on amount donated.
We found no difference in donations between
the Identifiable Victim condition (M 38.2%,
CI [29.7%, 46.7%]) and the Statistics
condition (M 33.7%, CI [26.2%, 41.2%]) (b
0.016, CI [0.122, 0.091], p .769). However, we
did find main effects of concern (b 0.183, CI
[0.113, 0.253], p .001) and empathy (b 0.111,
CI [ 0.190, 0.032], p .006) such that more
concern leads to more giving and more
empathy leads to less giving (R2 0.13). (We
should note that the negative relationship
between empathy and donations is consistent
with a suppressor effect that resulted from
controlling for concern. How- ever, we found
no evidence of multicollinearity [all VIFs
under 2.4]—a common means of diagnosing
suppressor effects—be- tween the subscales
and donations, so it is unclear whether this is a
true suppressor effect or something more
complicated. Details regarding the
multicollinearity diagnostics are available in
the online supplement.)
Next, we examined how the subscales
interacted with condition to predict donations.
We did not find a concern x empathy condition
interaction (b 0.693, CI [ 0.221, 1.607],p
.137), nor did we find an empathy x condition
interaction(b 0.111, CI [ 0.048, 0.271], p
.169). However, we did find a significant
concern x condition interaction (b 0.167,CI [
0.308, 0.026], p .020) such that the coefficient
of concern on donations was significantly
smaller in the Statistics condition (b 0.108, CI
[0.014, 0.201], p .024) than in the Identifiable
Victim condition (b 0.275, CI [0.168,
Jordan, Amir, and Bloom
12"
0.382],p .001; R2 0.156). See Figure 3 below.
Figure 3: The figures above show the relationships between empathy, concern and donations across
conditions. For graphical purposes, we have broken our continuous measures of empathy and concern into
quartiles. A: Donations are shown as a function of concern (quartiles) across the two conditions. B: Donations
are shown as a function of empathy (quartiles) across the two conditions. See the online article for the color
version of this figure.
These results suggest a number of
conclusions. First, in both the Identifiable
Victim condition and the Statistics condition,
concern was a better predictor of donations
than empathy. Second, although empathy was
predictive of donations, it was negatively
predictive, meaning that those who were
more empathic gave less to the charitable
cause. Third, the magnitude of the
relationships between concern and donations
increased in the presence of an identifiable
victim relative to the Statistics condition—
the positive relationship between concern and
donations was more positive in the
Identifiable Victim condition. This suggests
that the presence of a salient target for our
social emotions enhances the effects of our
social emotions, particularly when we feel
concerned, as one would expect if these
emotions are evolutionarily prepared. That is,
differences in concern lead to greater changes
in prosocial motivations in a more
ecologically relevant setting in which the
target has a name and face. We find the same
pattern of results for the Other-regarding and
Contagion factors (see online supplement).
Furthermore, despite the fact that those high
on concern donated more in both conditions,
this effect was most pronounced when
directed at an identifiable victim, which
suggests that the concern measure isn’t
tracking an abstract utilitarian sentiment that
scales with the number of people in need.
Rather, concern appears to be tracking a less
reflective other-regarding concern that scales
with the salience of the individual in need,
which leads to a more pronounced effect of
concern on donations when the victim is
identifiable.
Jordan, Amir, and Bloom
13"
But why do concern and empathy predict
donations? To answer this question, we used
the composite scores from the five feelings
questions (which were highly reliable, .911).
We asked participants about the charitable
cause as a means of looking for a mediation
between our subscales and donations. That is,
we wanted to determine whether concern
leads to higher donations or empathy leads to
lower donations because those who feel
concern or empathy are more moved by the
charitable cause, which in turn leads them to
give differently.
Furthermore, through mediation analysis, we
can shed some light both on how participants
interpret similarly ambiguous inductions used
in prior research (Batson, 2010; Batson,
Batson, et al., 1995; Batson et al., 1997; Toi
& Batson, 1982) and how general feelings of
empathy and concern translate into specific
altruistic motivations and actions. In fact, a
preliminary analysis revealed that feelings
scores were positively predicted by both
empathy (b 0.246, CI [0.052, 0.440], p .013)
and concern (b 0.719, CI [0.579, 0.860], p
.001), which suggests that both scales are
capturing emotions relevant to the altruistic
decision participants were faced with.
In the main analyses reported above, we
found direct effects of both concern and
empathy, so we conducted Sobel-Goodman
mediation analyses, using the sgmediation
package for Stata (Ender, 2006). Our goal
was to examine whether feelings about the
charitable cause mediated either or both of
these direct effects and included concern as a
covariate in the empathy analysis and vice
versa in the concern analysis. Furthermore, as
is common in other work on identifiable
victims, we found no difference in feelings
across conditions, t(190) 0.472, p .637 (Small
et al., 2007). Therefore, we collapsed across
conditions for these analyses.
We began with the empathy mediation
analysis (which included concern as a
covariate throughout) and found a negative
direct effect of empathy on donations (b
0.111, CI [ 0.190, 0.033], p .006). We then
examined the extent to which empathy
predicted feelings and found no relationship
(b 0.091, CI [ 0.264, 0.082], p .302). Finally,
we examined the indirect effect of empathy
on donations (b 0.089, CI [ 0.156, 0.023], p
.009) when controlling for the effect of
feelings on donations (b 0.245, CI [0.190,
0.300], p .001). Feelings mediated only 20%
of the effect of empathy on donations, which
a Sobel test for mediation revealed was not
significant (b 0.022, SE 0.022, z 1.027, p
.304). We should note that when concern is
not included as a covariate, we do find a
significant indirect effect of empathy on
donations through feelings (b 0.361, SE
0.148, z 2.432, p .015) such that more
empathy leads to more feelings, which in turn
lead to less giving. The details of this
analysis are available in the online
supplement.
Next, we conducted the same statistical
procedure but focusing on whether feelings
mediated the relationship between concern
and donations (with empathy included as a
covariate throughout). We began by
examining the direct effect of concern on the
fraction donated across both conditions,
which was strongly positive (b 0.184, CI
[0.115, 0.254], p .001). We then found that
feelings for the charitable cause were
positively predicted by concern (b 0.751, CI
[0.598, 0.904], p .001). Finally, we examined
the indirect effect of concern (b 0.001, CI [
0.071, 0.072], p .982) when controlling for
feelings (b 0.245, CI [0.190, 0.300], p .001).
Feelings mediated 99.6% of the effect of
concern on donations, which a Sobel test for
mediation confirmed was signif- icant (b
0.184, SE 0.028, z 6.508, p .001). Through
these two mediation analyses, we have some
Jordan, Amir, and Bloom
14"
evidence that although concern and empathy
both lead to being more moved by the
charitable cause (they are both positively
correlated with the feelings measure), such
feelings only translate to more giving if they
stem from concern (feelings mediate the
relationship between concern and donations
but not empathy and donations).
We also constructed some more elaborate
path models that contained multiple
mediators and the best fitting models always
started with a path from perspective taking to
concern (rather than beginning with
empathy). See the online supplement for
details.
Finally, as in Studies 1 and 2 above, we
factor analyzed the six subscales to ensure
the factor structure was stable and replicable.
Figure 4 below shows that, in a third distinct
sample, we found the same factor structure
with empathy, behavioral contagion, and
personal distress loading on one factor,
concern and perspective taking loading on
another, and fantasy loading on both. In
addition to replicating across three samples,
we aggregated all of our data (N 480) and
conducted several combined analyses. Most
importantly, in our combined dataset this
factor structure replicates across several
subpopulations that vary in their mean
subscale scores, like age, gender, and so
forth. See the online supplement for those
analyses and a matrix showing the
relationship between each of subscales and
the two factors.
Together, these results show that to the extent
that the identifiable victim in this study
evoked stronger emotional responses, being
prone to feeling empathy in response to a
salient and identifiable victim does not only
fail to motivate altruism, but in fact inhibits
altruistic behavior. Contrary to the effects of
empathy, concern for others appears to be a
strong motivator of altruistic behavior and is
a particularly potent motivator when the
target of giving is salient and identifiable.
Furthermore, the fact that feelings about the
charitable cause mediate the relationship
between concern and donations suggests that
concern leads to donations because of how
much more upsetting the situation is to those
high on concern. That is, when concerned
individuals are more moved by a charitable
cause, they donate more because they feel a
greater sense of moral responsibility to help
those in need.
Surprisingly, we did not replicate the typical
identifiable victim effect in which an
identifiable victim elicits more giving than
statistics about a charitable cause. There are a
number of potential explanations for this
failure, among which are the experimental
experience of Mechanical Turk participants
that would have ex- posed them to other
charitable giving studies, making them non-
naïve. That said, the goal of using the
identifiable victim paradigm was to
manipulate how salient and emotionally
evocative the target of the donation was, and
thereby activate the already-present
individual differences in these social
emotions. In that respect, the identifiable
victim paradigm accomplished this goal well.
Jordan, Amir, and Bloom
15"
Figure 4: The factor structure from Studies 1 and 2 replicated again in Study 3. Factor loadings above are the
result of principal factor analysis and promax rotation. See the online article for the color version of this figure.
General Discussion
Across three studies, our results motivate a
distinction between concern for others—what
Davis (1983) called “empathic concern”—
and a more narrow sense of empathy defined
as feeling what others feel, as measured in
our new Empathy Index. They are
psychologically distinct and do not underlie
the same prosocial behaviors. In our studies,
we find that concern, and not empathy, is the
primary motivator of moral thoughts and
actions. In Study 1, we examined the
relationship between the four subscales—
concern, perspective taking, fantasy, and
personal dis- tress—from the commonly
used IRI and the two novel subscales of our
Empathy Index—empathy and behavioral
contagion. The primary goal of Study 1 was
to determine whether, among a suite of other
social emotional capacities, feeling what
others feel (as measured by empathy) was
especially related to caring about what others
feel (as measured by concern). A factor
analysis revealed that the two are unrelated,
with empathy and concern each having
loadings near zero on the other’s factor.
Empathy was joined by personal distress and
our other novel subscale, behavioral
contagion, on the Contagion factor, whereas
concern was joined by perspective taking on
the Other-regarding factor.
Although the contents of the Contagion
factor were expected given the relationship
between feeling what others feel generally
(empathy) and feeling distressed when others
are in distress (personal distress), we were
surprised by the close and reliable
relationship between caring about what
others feel and understanding the contents of
others’ minds (as measured by perspective
taking).
Jordan, Amir, and Bloom
16"
In Study 2, we explored the relationship
between individual differences in the six
subscales and cooperation—a willingness to
pay an individual cost for the benefit of the
group—as measured by a PGG. We found
that both concern and empathy were
positively related to cooperation, but that
concern was uniquely predictive, remaining a
strong predictor even in the presence of the
other five subscales. Furthermore, of the total
variance the subscales accounted for, concern
accounted for the considerable majority. This
suggests that caring about the welfare of
others translates into being willing to pay a
cost for their benefit. However, in an
anonymous, one-shot PGG like the one we
used, the typical cues to emotional and
mental states of the others in the group are
not available, and are therefore unlikely to be
salient motivators. Perhaps further research
will provide insight into the role of these six
subscales in both imputing and responding to
the emotional and mental states of others.
In Study 3 we examined how the four
subscales of the IRI and the two subscales of
the Empathy Index were related to altruism,
as measured by how much participants
donated to a charitable cause. To determine
how individual differences in concern and
empathy (as well as the other subscales) were
causally related to giving, we manipulated
how salient and emotionally evocative the
target of altruism was using the identifiable
victim paradigm. We found that concern was
strongly positively predictive of donations,
whereas empathy was negatively predictive
of donations.
As expected, in Study 3, the salience of the
target mattered: in the presence of an
identifiable victim, the relationship between
concern and giving became more positive
(and the relationship between empathy and
giving became more negative, although not
statistically significantly so). The fact that
participants’ feelings about the charitable
cause mediated the relationship between
concern and donations suggests that those
high on concern gave because the identifiable
victim was concern-inducing, which resulted
in being moved by the charity. Along similar
lines, the fact that feelings about the charity
does not mediate the relationship between
empathy and donations suggests that an
empathic response does not lead people to be
more moved to donate to charitable causes.
Rather, empathic responses appear to be
debilitating, likely because identifiable
victims induce feelings of personal distress,
as the results from Study 1 would have
predicted. It is important to note, however,
that the role of trait empathy in
motivating prosocial actions may change
depending on the state induced by the target
of empathy. That is, to the extent that
empathic people are motivated by the
emotions they catch from others, it is
possible that empathic people would be more
likely to act prosocially in (the uncommon)
situations where the target of helping doesn’t
induce distress (e.g., empathic people may be
more likely to help a calm person in need
than a frantic or nervous person in need).
Surprisingly, we did not replicate the
identifiable victim effect. That is, participants
did not give more in the presence of an
identifiable victim than when presented with
statistics about suffering (although the null
result was in the right direction). This is
puzzling given how robust and reliable an
effect it is.
Taken together, these three studies suggest
that (a) feeling what others feel is
psychologically distinct from caring about
what others feel, (b) caring about what others
feel is a much stronger motivator of prosocial
thoughts and actions than feeling what others
feel, and (c) the use of our Empathy Index
Jordan, Amir, and Bloom
17"
subscales in conjunction with the IRI allow
for a more fine-grained analysis of social
emotional and cognitive abilities.
Furthermore, implicit in those findings is the
fact that our Empathy Index is reliably
measuring an important social emotional
construct: it tracked nicely with personal
distress throughout all three studies,
positively (although not uniquely or robustly)
predicted cooperation in Study 2 and was
negatively related to altruism in Study 3.
Hence our empathy subscale does appear to
be measuring the extent to which participants
feel what others feel, and is sensibly related
to how participants responded to social
situations.
There are other questions that our data raise
but do not answer. For instance, the
clustering between concern and perspective
taking is suggestive of a potentially
interesting set of interrelated psychological
processes that lead to prosocial thoughts and
actions. It appears that those who care about
the welfare of others are also more likely to
think about the contents of others’ minds.
Furthermore, although our experiments were
not designed to test the order of processes,
our path analyses suggest that perspective
taking leads to concern, which in turn
promotes prosociality. Future research may
shed more light on whether there is a casual
cognitive or developmental link between
perspective taking and concern. Perhaps the
most fruitful means of addressing the
relationship between having concern for
others and attending to the mental states of
others is to look at the emergence of these
processes in development.
Our studies suffer from some other
limitations. Despite the replicability of the
factor structure we observed, both across
studies and across subpopulations (see
supplement), our sample was exclusively
drawn from the United States. Therefore, it is
hard to make claims about the universality of
the dissociation between feeling what others
feel and caring about what others feel. Future
work may address this issue by examining
the factor structure across diverse populations
and small-scale societies.
Ever since it was acknowledged as a real
phenomenon, human prosociality has been
generally viewed as a puzzle. A number of
theories have been put forth in an effort to
describe the proximate mechanisms
responsible for inducing and supporting
cooperation and altruism. Although empathy
has figured prominently in many of these
theories, we now have some reason to doubt
its efficacy as a motivation for prosocial
behavior.
Jordan, Amir, and Bloom
18"
References
Baron-Cohen, S. (2002). The extreme male brain theory of autism. Trends in Cognitive Sciences, 6, 248–254.
http://dx.doi.org/10.1016/ S1364-6613(02)01904-6
Baron-Cohen, S., & Wheelwright, S. (2004). The empathy quotient: An investigation of adults with Asperger
syndrome or high functioning autism, and normal sex differences. Journal of Autism and Developmen- tal
Disorders, 34, 163–175. http://dx.doi.org/10.1023/B:JADD .0000022607.19833.00
Batson, C. D. (2010). Empathy-induced altruistic motivation. In M. M. P. R. Shaver (Ed.), Prosocial motives,
emotions, and behavior: The better angels of our nature (pp. 15–34). Washington, DC: American Psycho-
logical Association. http://dx.doi.org/10.1037/12061-001
Batson, C. D. (2011). Altruism in humans. Oxford, United Kingdom: Oxford University Press.
Batson, C. D., Batson, J. G., Todd, R. M., Brummett, B. H., Shaw, L. L., & Aldeguer, C. M. (1995). Empathy
and the collective good: Caring for one of the others in a social dilemma. Journal of Personality and Social
Psychology, 68, 619–631. http://dx.doi.org/10.1037/0022-3514.68.4.619
Batson, C. D., Klein, T. R., Highberger, L., & Shaw, L. L. (1995). Immorality from empathy-induced altruism:
When compassion and jus- tice conflict. Journal of Personality and Social Psychology, 68, 1042– 1054.
http://dx.doi.org/10.1037/0022-3514.68.6.1042
Batson, C. D., & Moran, T. (1999). Empathy-induced altruism in a pris- oner’s dilemma. European Journal of
Social Psychology, 29, 909–924. http://dx.doi.org/10.1002/(SICI)1099-0992(199911)29:7 909::AID- EJSP965
3.0.CO;2-L
Batson, C. D., Sager, K., Garst, E., Kang, M., Rubchinsky, K., & Dawson, K. (1997). Is empathy-induced
helping due to self–other merging? Journal of Personality and Social Psychology, 73, 495–509. http://dx
.doi.org/10.1037/0022-3514.73.3.495
Bloom, P. (2014, September/October). Against empathy. The Boston Re- view. Retrieved from
http://bostonreview.net
Bloom, P. (in press). Against empathy: The case for rational compassion. New York, NY: Harper Collins.
Costello, A., & Osborne, J. (2005). Best practices in exploratory factor analysis: Four recommendations for
getting the most from your analysis. Practical Assessment, 10, 1–7.
Darwall, S. (1998). Empathy, sympathy, care. Philosophical Studies, 89, 261–282.
http://dx.doi.org/10.1023/A:1004289113917
Davis, M. H. (1983). Measuring individual differences in empathy: Evi- dence for a multidimensional
approach. Journal of Personality and Social Psychology, 44, 113–126. http://dx.doi.org/10.1037/0022-3514
.44.1.113
Dawes, R. M. (1980). Social dilemmas. Annual review of psychology, 31, 169 –193.
Decety, J., & Cowell, J. M. (2014). The complex relation between morality and empathy. Trends in Cognitive
Sciences, 18, 337–339. http://dx.doi .org/10.1016/j.tics.2014.04.008
Jordan, Amir, and Bloom
19"
Decety, J., & Yoder, K. J. (2016). Empathy and motivation for justice: Cognitive empathy and concern, but not
emotional empathy, predict sensitivity to injustice for others. Social Neuroscience, 11, 1–14. http://
dx.doi.org/10.1080/17470919.2015.1029593
DeSteno, D. (2015). Compassion and altruism: How our minds determine who is worthy of help. Current
Opinion in Behavioral Sciences, 3, 80–83.
de Vignemont, F., & Singer, T. (2006). The empathic brain: How, when and why? Trends in Cognitive
Sciences, 10, 435–441. http://dx.doi.org/ 10.1016/j.tics.2006.08.008
De Waal, F. B. (2008). Putting the altruism back into altruism: The evolution of empathy. Annual Review of
Psychology, 59, 279–300. http://dx.doi.org/10.1146/annurev.psych.59.103006.093625
De Waal, F. (2010). The age of empathy: Nature’s lessons for a kinder society. New York, NY: Random
House LLC.
Eisenberg, N., & Fabes, R. (1990). Empathy: Conceptualization, measure- ment, and relation to prosocial
behavior. Motivation and Emotion, 14, 131–149. http://dx.doi.org/10.1007/BF00991640
Eisenberg, N., & Strayer, J. (1987). Critical issues in the study of empathy. In N. E. J. Strayer (Ed.), Empathy
and its development (pp. 3–13). New York, NY: Cambridge University Press.
Ender, P. (2006). Sgmediation: Program to compute Sobel–Goodman mediation tests. Los Angeles, CA:
Academic Technology Services, Statistical Consulting Group
Hoffman, M. L. (2008). Empathy and prosocial behavior. Handbook of Emotions, 3, 440–455.
Hume, D. (1978). 1739. A treatise of human nature. London, UK: John Noon. Preston, S. D., & de Waal, F. B.
(2002). Empathy: Its ultimate and
proximate bases. Behavioral and Brain Sciences, 25, 1–20.Prinz, J. (2011a). Against empathy. The Southern
Journal of Philosophy, 49,
214–233. http://dx.doi.org/10.1111/j.2041-6962.2011.00069.xPrinz, J. (2011b). Is empathy necessary for
morality. Empathy: Philosoph-
ical and Psychological Perspectives, 1, 211–229.Rand, D. G. (2012). The promise of Mechanical Turk: How
online labor
markets can help theorists run behavioral experiments. Journal of The- oretical Biology, 299, 172–179.
http://dx.doi.org/10.1016/j.jtbi.2011.03 .004
Russell, D. W. (2002). In search of underlying dimensions: The use (and abuse) of factor analysis in
Personality and Social Psychology Bulletin. Personality and Social Psychology Bulletin, 28, 1629–1646.
http://dx .doi.org/10.1177/014616702237645
Schroeder, D. A., Dovidio, J. F., Sibicky, M. E., Matthews, L. L., & Allen, J. L. (1988). Empathic concern and
helping behavior: Egoism or altru- ism? Journal of Experimental Social Psychology, 24, 333–353. http://
dx.doi.org/10.1016/0022-1031(88)90024-8
Jordan, Amir, and Bloom
20"
Shamay-Tsoory, S. G., Aharon-Peretz, J., & Perry, D. (2009). Two systems for empathy: A double
dissociation between emotional and cognitive empathy in inferior frontal gyrus versus ventromedial pre-
frontal lesions. Brain: A Journal of Neurology, 132, 617–627. http:// dx.doi.org/10.1093/brain/awn279
Singer, T., & Klimecki, O. M. (2014). Empathy and compassion. Current Biology, 24, R875–R878.
http://dx.doi.org/10.1016/j.cub.2014.06.054 Small, D. A., & Loewenstein, G. (2003). Helping a victim or
helping the
victim: Altruism and identifiability. Journal of Risk and Uncertainty, 26,
5–16. http://dx.doi.org/10.1023/A:1022299422219Small, D. A., Loewenstein, G., & Slovic, P. (2007).
Sympathy and cal-
lousness: The impact of deliberative thought on donations to identifiable and statistical victims. Organizational
Behavior and Human Decision Processes, 102, 143–153. http://dx.doi.org/10.1016/j.obhdp.2006.01.005
Smith, A. (1759). (1976): The theory of moral sentiments. Indianapolis, IN: Liberty Classics.
Toi, M., & Batson, C. D. (1982). More evidence that empathy is a source of altruistic motivation. Journal of
Personality and Social Psychology, 43, 281–292. http://dx.doi.org/10.1037/0022-3514.43.2.281
Trivers, R. L. (1971). The evolution of reciprocal altruism. The Quarterly Review of Biology, 46, 35–57.
http://dx.doi.org/10.1086/406755
Zaki, J. (2016). Empathy is a moral force. In K. G. Gray, J. (Ed.), The atlas of moral psychology. New York,
NY: Guilford Press.
- A preview of this full-text is provided by American Psychological Association.
- Learn more
Preview content only
Content available from Emotion
This content is subject to copyright. Terms and conditions apply.
