Running head: DEVELOPMENT OF EMPATHY
The Developmental Origins of a Disposition toward Empathy: Genetic and
The Hebrew University of Jerusalem
University of Wisconsin - Madison
Carol Van Hulle
University of Wisconsin - Madison
JoAnn L. Robinson
University of Connecticut-Storrs
Soo Hyun Rhee
Institute for Behavioral Genetics, Boulder
Correspondence should be addressed to Ariel Knafo, Psychology Department, The Hebrew
University of Jerusalem, Mount Scopus, Jerusalem 91905, Israel. Telephone: +972 2 5883426.
Fax: +972 2 5881159. E-mail: firstname.lastname@example.org.
Development of Empathy 2
We investigated the development of a disposition towards empathy and its genetic and
environmental origins. Young twins' (N=409 pairs) cognitive (hypothesis testing) and
affective (empathic concern) empathy and prosocial behavior in response to simulated pain by
mothers and examiners were observed at multiple time points. Children's mean level of
empathy and prosociality increased from 14 to 36 months. Positive concurrent and
longitudinal correlations indicated that empathy was a relatively stable disposition,
generalizing across ages, across its affective and cognitive components, and across mother
and examiner. Multivariate genetic analyses showed that genetic effects increased, and shared
environmental effects decreased, with age. Genetic effects contributed to both change and
continuity in children's empathy, whereas shared environmental effects contributed to stability
and non-shared environmental effects contributed to change. Empathy was associated with
prosocial behavior, and this relationship was mainly due to environmental effects.
Development of Empathy 3
The Developmental Origins of a Disposition toward Empathy: Genetic and
Compassion is a dimension of morality that emphasizes concern for the well-being of others
in distress. It is an important aspect of interpersonal responsibility and ethical behavior.
Empathy and prosociality are essential to the expression of compassion. Developmentally,
empathy is present in the first years of life (Zahn-Waxler, Radke-Yarrow, Wagner, &
Chapman, 1992a), and may play a role in facilitating caring actions reflecting concern for the
well-being of others (Hoffman, 1975; Eisenberg & Fabes, 1998). Such actions include
helping, providing physical comfort or reassurance, sharing, and sympathizing. Theorists with
different approaches (Batson, in press; Eisenberg et al., 1987; de Waal, 2007) characterize
empathy as an ideal candidate mechanism to underlie caring behaviors in response to
another’s pain, need, or distress.
In this study, we propose that empathy is an enduring disposition, which is relatively
stable across time and consistent across contexts, and across its cognitive and affective
aspects. Empathy includes both cognitive and affective components (Decety & Jackson, 2004;
Preston & de Waal, 2002; de Vignemont & Singer, 2006). The cognitive aspect of empathy
entails an ability to effectively comprehend a distressing situation and to recognize another’s
emotions and assume that person’s perspective. In young children, it appears in the form of
hypothesis-testing or inquisitiveness, whereby the child actively tries to understand the other’s
problem. The affective aspect of empathy requires an individual to experience a vicarious
emotional response to others’ expressed emotions. In young children, it is seen in emotional
expressions of concern for the victim. Both cognitive and affective components are aspects of
the same complex construct of empathy, and they are not conceived as independent. For
example, Hoffman (1988) referred to empathy as a vicarious affective response that relies on
a developed cognitive sense of others.
Development of Empathy 4
Our research examined the development of empathy as a stable disposition in the second
and third years of life. We assessed both cognitive and affective components of empathy as
well as prosocial behaviors that reflect children’s concern for the well-being of others. We
addressed the development of empathy and prosocial behavior in male and female children.
Using data from monozygotic and dizygotic twins, observed at multiple time points over the
second and third years of life, we examined the genetic and environmental contributions to
empathic development and to the relationship between empathy and prosocial behavior.
Empathy as a Stable Disposition
Children's empathic responding to others has been linked to several stable personality
or temperament traits, such as behavioral inhibition (Young et al., 1999), positive affectivity
(Volbrecht et al., 2007), effortful control (Valiente et al., 2004), and concentration and
impulse control (Rothbart, Ahadi, & Hershey, 1994). Based on longitudinal research, empathy
has been conceptualized as part of a broader altruistic or prosocial personality disposition
whose roots can be found in childhood (Eisenberg, et al., 1999). The first goal of this study
was to investigate empathy as a stable disposition beginning in the first years of life.
A trait approach to empathy suggests that there is an underlying common empathy
disposition that is manifest in both the cognitive (hypothesis testing) and affective (empathic
concern) aspects of empathy. Indeed, hypothesis testing and empathic concern are positively
correlated (Gill & Calkins, 2003; Young et al., 1999; Zahn-Waxler et al., 2001; but see
Volbrecht et al., 2007). In addition, empathy viewed as a trait entails a degree of continuity
across time, as evidenced by longitudinal stability in children's empathy (van der Mark et al.,
2002; Volbrecht, et al., 2007; Zahn-Waxler, Robinson, et al., 1992b; Zahn-Waxler et al.,
2001). Finally, cross-situational consistency is another important feature of an empathy trait.
Indeed, children's empathic concern and hypothesis testing towards an unfamiliar examiner
correlate with the same behavior performed towards the mother (Moreno, Klute, & Robinson,
Development of Empathy 5
in press; Robinson, Zahn-Waxler, et al., 2001; Young et al., 1999; but see van der Mark et al.,
2002, for empathic concern in Dutch girls) and in response to hearing another child cry (Gill
& Calkins, 2003).
Based on these considerations, we expected hypothesis testing and empathic concern,
towards both mothers and an unfamiliar examiner, to correlate with each other and to load on
a common empathy factor at all ages. We also expected this empathy factor to show
considerable continuity across time. Taken together, these expectations imply that there is a
general, underlying empathy factor.
The presence of an underlying empathy factor does not preclude the important
influence of situational constraints, and although we expected some consistency in children's
behaviors towards their mother and an unfamiliar examiner, such consistency would not be
expected to be perfect, because children's dispositions may be manifest differently across
contexts (see Mischel & Shoda, 1995). For example, some children may tend to react
empathetically to their mother's pain but not to that of an unfamiliar examiner. Similarly, the
affective and cognitive aspects of empathy may develop at different rates reflecting partially
non-overlapping brain regions (Singer, 2006), which may be reflected in some children's
empathy being more focused on empathic concern, and other children's on hypothesis testing.
We were interested in the development of specific aspects of empathy (e.g., empathic
concern) as well.
Genetic and Environmental Influences on Empathy
The view of empathy as a stable individual differences disposition raises the issue of
the sources for these individual differences. The second goal of this study was to investigate
the relative genetic and environmental contributions to individual differences in empathy.
Only two studies have addressed directly the issue of the origins of individual
differences in children's observed empathy in the first years of life using a genetically
Development of Empathy 6
informative design. These studies have relied on the twin research design, which allows
differentiation between genetic and environmental influences as it compares monozygotic
(MZ) twins, who share all of their genes, with dizygotic (DZ) twins, who share on average
half of their genes. Assuming that twins of both types share their environments (e.g., the
family environment) to the same extent, higher similarity in MZ versus DZ twins indicates
genetic influence. An estimate of heritability (the proportion of individual differences in the
study population under specific conditions that is due to genetic variability) is computed.
Twin similarity that is beyond this genetic effect can be attributed to the Shared Environment
(environmental influences shared by twins), whereas differences between twins not due to
genetic differences are ascribed to Non-shared Environment or to measurement error (Plomin,
DeFries, McClearn, & McGuffin, 2001).
Volbrecht et al. (2007) estimated the heritability of empathic concern of children aged
19-25 months towards their mother at .30, and that of hypothesis testing at .40. A preliminary
study based on a partial sample (60%) of the present study also found moderate genetic
effects on empathic concern and hypothesis testing (Zahn-Waxler et al., 2001). Both studies
also found evidence of both shared and non-shared environmental influences. We thus
expected empathy to be influenced by both genetic and environmental effects.
Empathy is viewed as a broad disposition that generalizes across its cognitive and
affective components, and across situations. Therefore, at each age, we investigated the
genetic and environmental contributions to a common factor (e.g., Rijsdijk & Sham, 2002) of
empathy, which encompasses both hypothesis testing and empathic concern towards both the
children's mother and an unfamiliar examiner. However, children's responses are not expected
to correlate perfectly across contexts and across modalities of empathy. In addition to stability
and consistency in children's overall behavior, meaningful variations in how individuals
respond to specific contexts may exist (Mischel & Shoda, 1995). Therefore, in addition to
Development of Empathy 7
general genetic and environment effects relevant to a common factor of empathy, we also
investigated the genetic and environmental effects unique to specific instances of empathy.
The Development of Genetic and Environmental Influences on Empathy
As noted, the idea of an empathy disposition entails substantial longitudinal stability.
This stability can be the result of either stable genetic influences, or stable environmental
effects, or both (Knafo & Plomin, 2006a). The third goal of this study is to estimate the
genetic and environmental contributions to stability in empathy across time. Genetics and the
environment can also contribute to change with time. Because children's empathic abilities
undergo changes in the first few years of life, it is important to study the contributions to
change and stability. Previous research suggests that genetic effects contribute to continuity,
but additional new genetic effects accounting for change also emerged as children grow up
(Knafo & Plomin, 2006a; Plomin et al., 1993; Zahn-Waxler et al., 2001). Shared
environmental effects can also account for both continuity and change. Zahn-Waxler et al.
(2001) reported overall continuity in the shared environment effects accounting for children’s
empathy. Finally, non-shared environmental influences typically contribute to change rather
than continuity (Knafo & Plomin, 2006a; Zahn-Waxler et al., 2001). Our longitudinal genetic
design will enable us to observe the dynamics of genetic and environmental effects in the
children's earliest years of life.
Empathy and Prosocial Behavior
Empathy has been described as the motivating impetus behind prosocial behavior in
humans and other animals (de Waal, 2007). The cognitive and affective understanding of the
suffering of others that empathy entails is characterized by a negative experience, and can
lead to behavioral efforts to alleviate the distress of the other. In this sense, prosocial
behaviors, whose intention is to promote the welfare of others (Eisenberg & Fabes, 1998), can
Development of Empathy 8
be seen as a potential outcome of empathy. Investigating the relationship between prosocial
behavior and empathy is the fourth goal of this paper.
Research on children shows positive relationships between empathy towards a person
in distress and prosocial behavior towards that person (e.g., Eisenberg & Fabes, 1998; Young
et al., 1999; Zahn-Waxler, Radke-Yarrow, et al. 1992). In addition to the relationship between
the empathy experienced at a certain moment and the desire to help the victim, research on
children aged five to 13 found relationships between children's overall tendency for empathy
and their prosocial behavior (e.g., Barnett & Thompson, 1985; Roberts & Strayer, 1996). We
expected that, even at younger ages more empathic children would be more likely to behave
prosocially than less empathic children. .
As is the case with empathy, there is evidence of moderate genetic influences on
individual differences in children's prosocial behavior (Hur & Rushton, 2007; Knafo &
Plomin, 2006a, 2006b; Scourfield, John, Martin, & McGuffin, 2004; Zahn-Waxler et al.,
2001). Because empathy and prosocial behavior are positively related at the phenotypic level,
and because they are both expected to be heritable to a certain extent, we sought to estimate
the extent to which this relationship is due to genetic factors that affect both prosocial
behavior and empathy. If, for example, the core affective basis of prosocial behavior is the
empathy we feel towards others (de Waal, 2007), then genetic effects responsible for affective
dispositions implicated in empathy may be responsible for the ensuing prosocial behavior.
Similarly, the shared environment may be responsible for this relationship, because the
environmental correlates of prosocial behavior and empathy overlap substantially (e.g.,
parental warmth and sensitivity to children's needs, Hastings, Zahn-Waxler, & McShane,
2006). We examined each of these possibilities.
Development of Empathy 9
The Current Study
In this study we follow up on our earlier work (Zahn-Waxler et al., 2001) with a
substantially increased twin sample. An important change from previous work is the focus
on empathy as an individual-differences disposition, which characterizes responses to others'
distress across contexts, cognitive and affective processes, and time. This focus on overall
empathy, and the increased sample size, enabled us to separate the genetic and
environmental components common to different aspects of empathy, from those unique to
hypothesis testing and empathic concern in different contexts.
We investigated the development of empathy from infancy to early childhood. We did
this separately for cognitive and affective components of empathy, considering also the target
of children's empathy. In light of previous studies, we expected an increase in empathy with
age (Volbrecht, Lemery-Chalfant, Aksan, Zahn-Waxler, & Goldsmith, 2007; Zahn-Waxler,
Radke-Yarrow, et al., 1992, but see van der Mark, van IJzendoorn, & Bakermans-
Kranenburg, 2002), and higher empathy towards the mother than towards an examiner (van
der Mark et al., 2002; Robinson, Zahn-Waxler, et al, 2001; Young, Fox, & Zahn-
WaxlerWalker, 1999, but see Zahn-Waxler, Radke-Yarrow, et al., 1992a).
In summary, we had four main goals. First, we investigated empathy as an individual
differences variable, focusing on consistency between the affective and cognitive aspects of
empathy, between targets, and on continuity across time. Second, we investigated genetic
and environmental contributions to individual differences in empathy, both those common to
a common empathy factor and those unique to affective empathy or cognitive empathy
towards the mother and the examiner. Third, we estimated the contribution of genetic and
environmental influences to change and continuity in children's empathy. Finally, we
examined the relationship between prosocial behavior and children's empathy and estimated
the genetic and environmental contributions to this relationship.
Development of Empathy 10
Twins were recruited from monthly reports of births from the Division of Vital
Statistics of the Colorado Department of Health. Twins were selected preferentially for higher
birth weight and birth weights appropriate for gestational age. No twins with birth weights
less than 1,700 gr. or with gestational ages less than 34 weeks were selected. More than half
of parents contacted agreed to participate in the study. Over 90% of the sample is White;
other participating families are primarily Hispanic, and 1% is African-American. Participating
parents were slightly older (30 years vs. 28 years, on the average) and somewhat more
educated (14.5 years vs. 12.5 years) than the average of parents of newborns in Colorado.
This study was part of a larger longitudinal genetically informed investigation of continuity
and change in the development of cognition, emotion and temperament in twins. (See
Robinson, McGrath, & Corley, 2001, for greater detail about sampling and design).
Twin zygosity was determined by comparing twins' physical attributes with a
modified version of the Nichols and Bilbro questionnaire (1966), based on having 85%
agreement from four or more raters, and reliance on twins' sharing 11 highly informative short
tandem repeat (STR) polymorphisms (Rhea, Gross, Haberstick & Corley, 2006).
Assessments were made at 14, 20, 24 and 36 months. The original sample included
230 monozygotic (MZ) pairs and 179 dizygotic (DZ) pairs and their families, who
participated in at least one of these ages. For the present study, data were available for 391
pairs at 14 months (96% of the total sample), 353 pairs at 20 months (86%), 355 pairs at 24
months (87%), and 341 pairs at 36 months (83%), an adequate retention rate. No differences
in twins' sex or zygosity, or in their empathy and prosocial behavior, were observed between
families who dropped out of the study after 14 months (the largest drop in participation rate)
and those who participated at a later age. At each age, the vast majority of the families (80-
Development of Empathy 11
84%) participated in both the lab and home visit; a small proportion participated only in the
lab visit (2-3%), and a substantial minority (12-17%) participated only in the home visit. To
increase reliability of measures and analytic power, we combined data from the lab and home
At each measurement point, two female examiners visited the twins and their mothers
at home of the participants. One to 3 weeks later, the twin pair was brought to the laboratory
by their mother. Visits were scheduled at a time when the children were likely to be at their
best, usually after naps or meals. Each visit was completed in less than 3 hours.
Simulation procedures were used to probe for children's empathic capabilities. Four
empathy probes at each age are examined in this report. At different points during each visit,
the mother and examiner simulated distress according to specified scripts. Emphasis was
placed on inserting these events naturally in the context of ongoing activities and in situations
in which the other twin was not present. During the home visit, the examiner pretended to
close her finger in the suitcase containing testing materials. The mother was instructed to
pretend to hurt her knee as she got up from the floor. During the laboratory visit, the mother
was instructed to catch her finger in a clipboard and the examiner bumped into a chair, both
expressing pain. In all simulations, the mothers (and examiners) were instructed to
accompany the simulated injury with rubbing of the injured body part, by vocalizing pain at
low to moderate volume, and a pained facial expression over a 30-second period with a
gradual subsiding of the distress for an additional 30 seconds. Direct eye contact was always
avoided to prevent the subtle induction of a response from the child.
These empathy probes are a widely used procedure for assessing children's empathy
(e.g., Gill & Calkins, 2003; van der Mark et al., 2002; Moreno et al., in press; Volbrecht et al.,
2007; Young et al., 1999; Zahn-Waxler, Robinson, et al., 1992; Zahn-Waxler, Radke-Yarrow
Development of Empathy 12
et al., 1992). Simulations were scored for credibility. Virtually all of the examiners'
simulations were rated as credible, whereas 95% of mothers' simulations were so rated. The
high credibility rates reflected the fact that mothers and examiners had received specific
training in how to carry out these procedures.
Children's reactions to the empathy probes were videotaped, and all scoring and
reliability coding was done from these tapes. Video cameras zoomed in on the children's
upper bodies during these simulations, so that hands and faces were clearly visible for coding.
Observers used multiple passes to ascertain and clarify children's various emotional
responses. The codes for the components are based on previous research (Zahn-Waxler,
Radke-Yarrow, et al., 1992) with some extensions and adaptations. Each twin in a dyad was
assigned to a different observer. This was done to keep coders uninformed of the child's
zygosity and unbiased by the other twin's responses. Inter-observer reliability was checked
periodically, typically once every 6 months, although checks occurred more frequently when
new observers were trained. Reliabilities were estimated from independent scoring of 236
distress simulations. Observer reliabilities for empathic concern and hypothesis testing were
computed as polychoric correlations (used for ordinal scales), and were .87 and .79,
respectively. Reliability for prosocial behavior was computed as a tetrachoric correlation
(used for dichotomous scales) and was .91.
Empathic concern: Expressions of apparent concern for the victim, including facial,
vocal, or gestural–postural expressions were rated on a 4-point scale ranging from 1 = absent,
to 2 = slight (fleeting or slight change of expression that includes brow furrow), to 3 =
moderate (sustained sobering of expression that includes brow furrow), to 4 = substantial
concern (sustained sadness expressed in cooing or sympathetic vocal tones or sympathy face
in which eyebrows are drawn down and brow drawn up over the nose). Children’s responses
Development of Empathy 13
in the distress simulations at the lab and home correlated positively (across ages, for mothers,
r=.24 on average; for examiners, r=.16), and were averaged to increase reliability and reduce
the number of analyses.
Hypothesis testing: Behaviors in which the child explores and/or attempts to
comprehend distress were rated on a 4-point scale ranging from 1 = none, to 2 = simple
nonverbal (e.g., looking from injury to victim's face) or verbal (e.g., a single utterance,
“Hurt?”), to 3 = combinations of nonverbal and verbal exploration (e.g., looking at the injury
and its cause and inquiring “Owie?”), to 4 = repeated, sophisticated attempts to comprehend
the problem (e.g., asking, “Does it hurt? Did you pinch it?” or looking behind or underneath
the injury to ascertain cause). Scores from the lab and home simulations home correlated
(across ages, for mothers, r=.26 on average; for examiners, r=.41), and were averaged to
reduce the number of analyses.
Prosocial behavior: Efforts to help or comfort the mother or examiner (e.g., attempts
to comfort and distract the victim, getting band-aid, patting the victim, sharing or offering a
toy, or defending the victim by hitting the offending object, such as the clipboard) were noted
as prosocial acts. At each age, prosocial behavior was coded if it occurred at least once, either
at the lab situation or at the home situation. The scores of children participating only at home
or at the lab were based on this single visit.
Descriptive Statistics for the Empathy Scores
All frequency and mean comparisons and correlational analyses were performed with
the SPSS software, version 14.0. Because twin scores are not independent of each other, the
scores of only one twin per pair, randomly chosen, were used in the descriptive analyses and
in the within-twin correlations (similar results were obtained with data from the other twin in
each pair). Preliminary analyses showed that MZ and DZ twins did not differ in hypothesis-
Development of Empathy 14
testing or empathic concern. In addition, no significant interaction was found between
zygosity and age, sex, distress victim (mother/examiner), or type of empathy (hypothesis
testing or empathic concern) in affecting empathy levels. We therefore dropped the zygosity
variable from the mean comparison analyses.
Table 1 presents the means and standard deviations of scores for the two types of
empathy (hypothesis testing and empathic concern) toward mothers and examiners at ages 14,
20, 24, and 36 months, separately for girls and boys. Two repeated measures three-way
ANOVAs (4 age groups X 2 sexes X 2 distress victims) were conducted, one for hypothesis
testing and one for empathic concern. In the analysis model for each dependent measure, the
overall tests for the three-way interactions were not significant, F values < 1.
Age effects. There were marked increases in empathy with age, for both hypothesis
testing, F (3,290) = 143.40, p < .001, and empathic concern, F (3,290) = 41.05, p < .001. The
upper panel of Figure 1 plots these increases, using estimated marginal means and their 95%
confidence intervals. The lower panel plots empathy levels separately for boys and girls and
for mothers and examiners. The increase in empathic concern took place mainly from age 14
months to 20 months, whereas hypothesis testing increased steadily through age 36 months.
Mother vs. examiner. Hypothesis testing was more pronounced towards mothers than
examiners, F (1,290) = 71.52, p < .001. Significant interactions between age and distress
victim suggest that this difference between mother and examiner varied with age, F (3,290) =
23.37, p < .001. In post-hoc repeated contrasts, hypothesis testing was more prevalent towards
the examiner at 14 months, t (290) = -2.79, p < .01; but more prevalent for mothers at 20, t
(290) = 3.95, p < .001; 24, t (290) = 7.61, p < .001, and 36 months, t (290) = 6.17, p < .001.
Overall, empathic concern was slightly higher for the examiner at most ages, F (1,290) =
10.30, p < .001 (14 months, t (290) = -4.91 , p < .001; 24 months, t (290) = -1.52, ns; 36
months, t (290) = -2.20, p < .05), but higher empathic concern towards mothers was found at
Development of Empathy 15
20 months, t (290) = 1.69, ns (followed by a slight decrease in empathic concern towards
mothers at later ages, Figure 1b), resulting in a significant interaction between age and
distress victim, F (3,290) = 7.86, p < .001.
Sex differences. Girls scored higher than boys on empathic concern, F (1,290) = 6.37,
p < .05. This sex difference was stable and did not interact with age, F (3,870) = 1.08, ns, or
with the distress victim, F (1,290) = 0.30, ns. Although there was not a main effect of sex on
hypothesis testing, there was an interaction of sex with age, F (3,870) = 3.77, p < .05. Girls
tended to score higher in hypothesis testing at younger ages than boys (although these effects
were not significant, 14 months, t (290) = 0.84; 20 months, t (290) = 0.70; 24 months, t (290)
= 0.65, all ns), while boys scored somewhat higher than girls by 36 months, t (290) = -2.13, p
< .05 (this difference favoring boys at 36 months was not replicated when the analyses were
done with the other twin). The sex of the child also interacted with the distress victim, F
(1,290) = 7.77, p < .01. Both girls, t (141) = 8.07, p < .001, and boys, t (149) = 3.96, p < .001,
showed more hypothesis testing towards their mother than the examiner, but this difference
was stronger for girls (D = 1.36 vs. 0.65, respectively).
Evidence for Empathy as a Stable Disposition
An examination of the correlations between children's empathy towards their mother
and the examiner enabled us to investigate the stability of children's empathic tendencies.
Table 2 presents the correlations between children's empathic concern and hypothesis testing
towards the mother and the examiner. For empathic concern, the correlation between behavior
towards the mother and towards the examiner averaged .36 across the four age groups, and
was lowest at 36 months (.26). Mother-examiner hypothesis testing correlations averaged .43
across ages, and also were lowest at 36 months (.25).
The affective component (empathic concern) and the cognitive component (hypothesis
testing) of empathy also correlated positively as predicted. Correlations for empathy towards
Development of Empathy 16
the examiner averaged .32 across ages, and were lowest at 36 months (.27). Correlations for
the mother averaged .48, and were lowest at 36 months (.40). The overall picture is that of
stability in empathy towards the examiner and the mother, and across the affective and
cognitive components of empathy, and a stability that declines at 36 months. These results are
consistent with the hypothesized general latent empathy factor that is common to empathic
concern and hypothesis testing and towards mother and examiner, described below.
The longitudinal correlations between children's empathy at the different ages appear
in Table 3. All correlations between adjacent measurement points were positive and
significant, indicating some continuity in children's empathy. The average correlation from 14
to 20 months and from 20 to 24 months was .29. Correlation across the 12-month gap from 24
to 36 months averaged .21. Even correlations from 14 to 36 months tended to be positive,
averaging .15, although they were not significant for empathic concern (Table 3).
As a direct test of the idea of empathy as a dispositional variable, we ran a principal
component analysis, separately at each age. At all ages, a single factor emerged, on which all
four empathy indicators (empathic concern and hypothesis testing towards the mother and the
examiner) loaded with loadings ranging from .58 to .80. This single factor accounted for a
substantial proportion of the variance at all ages (14 months, 51%, 20 months, 55%, 24
months, 51%, 14 months, 42%).
Longitudinal stability was further tested with structural equation modeling (SEM)
(AMOS statistical package, Arbuckle, 1997). At each age, hypothesis testing and empathic
concern towards the mother and the examiner were modeled as loading on a single empathy
factor. We permitted correlated errors between empathic concern towards the mother and the
examiner, and between empathic concern and hypothesis testing towards the mother. The
analysis showed that all the indicators loaded significantly on their hypothesized age-specific
latent factors (p < .001), with standardized loadings ranging from .15 to .86. The overall fit of
Development of Empathy 17
the model with age-specific factors, allowing for the correlated errors, was good: the index of
comparative fit (CFI) was .96, and the root mean square error of approximation (RMSEA)
was .04. These results support empathy as an age-specific disposition. In addition, the
longitudinal stability of the common empathy factor was high, with the empathy factor at
each age predicting a substantial proportion of the variance at the following measurement age:
14 to 20 months, β = .76, t =8.16, p < .001, R2 = .58; 20 to 24 months, β = .75, t =5.89, p <
.001, R2 = .57; 24 to 36 months, β = .57, t =7.47, p < .001, R2 = .32.
Genetic and Environmental Influences on Empathy
To examine genetic and environmental influences on empathy at each of the 4 ages,
we first compared twin intraclass correlations obtained within MZ and DZ pairs, presented in
Table 4. At 14 months, the average correlation for MZ pairs was .35, whereas that of DZ pairs
was .32. At 20 months, the average MZ twin correlation was .30, and .33 for DZ pairs.
Positive correlations that are similar in size for DZ and MZ twin pairs suggest that twin
resemblance at these ages is mainly due to shared environment and not to genetic effects.
A different pattern was found in the later ages. At 24 and 36 months, the average MZ
twin intraclass correlations were .35 and .33, respectively. The corresponding correlations for
DZ twins were .22 and .19. This difference in the magnitude of correlations between MZ and
DZ twins suggests a genetic influence on empathy that becomes stronger with age.
As a more direct test of genetic and environmental effects, we used the Common-
Factor-Common-Pathways Multivariate model (e.g., Rijsdijk & Sham, 2002). The model
assumes that there is an underlying common factor accounting for individual differences
across the different empathy indices, and provides an estimate of the proportion of variance in
each measure of empathy associated with the common factor. In addition, the model estimates
the remaining, residual variance that is unique to each index (e.g., variance in empathic
concern towards mothers that is not accounted for by the common factor.) The benefit of this
Development of Empathy 18
model is that it can be used to disentangle genetic and environmental effects unique to
individual measures (e.g., empathic concern towards the mother) from those applying across
empathy indices. One factor affecting all four empathy indices (empathic concern and
hypothesis testing towards the mother and the examiner) is estimated. The magnitude of
genetic influences, shared environmental influences, and non-shared environmental influences
are estimated for this common empathy factor. In addition, the magnitude of variable-specific
genetic and environmental influences is estimated. A schematic representation of this model
is depicted in Figure 2. The results of this path analysis are presented in terms of components
of variance. All genetic analyses were performed using the Mx structural equation modeling
software (Neale, Boker, Xie & Maes, 1999).
In fitting the models to the data, we proceeded by dropping nonsignificant (p > .05)
path coefficients from the model. At all ages, dropping these paths (estimated at .00 in Table
5) resulted in a more parsimonious model, without worsening model fit, as indicated by the
difference in model fit between the full and the modified models at 14 months, ∆χ2(df=7) =
1.17, ns; 20 months, ∆χ2(df=7) = 1.52, ns; 24 months, ∆χ2(df=6) = 3.05, ns; 36 months,
∆χ2(df=6) = 5.98, ns. At 24 months, both the genetic and the shared environment effects on
the common empathy factor were not significant, and dropping them separately did not
significantly worsen model fit with the current sample size. However, dropping both of them
simultaneously did result in a substantially worse fit, ∆χ2(df=8) = 69.80, p < .01. Therefore,
they were both retained in the model. The models fit the data marginally well, with Akaike’s
information criterion (AIC) ranging from 24.04 to 95.43 and the root mean square error of
approximation (RMSEA) between .09 and .11.
The underlying common empathy factor. The results of analyses performed separately
at each age appear in Table 5. The four empathy indices loaded positively on the underlying
common empathy factor, with standardized loadings ranging from .36 to .76 at 14 months, .46
Development of Empathy 19
to .78 at 20 months, .41 to .77 at 24 months, and .29 to .63 at 36 months. Table 5 presents the
proportion of the variance in each empathy measure accounted for by the latent common
factor, which is equivalent to the squared standardized loadings. All indices of empathy also
had unique variability not accounted for by the common factor.
Genetic and environmental influences on the common empathy factor. The first
column of Table 5 presents the estimates and 95% confidence intervals of the variance
components accounting for individual differences in the common empathy factor. Consistent
with the MZ and DZ correlations, no genetic influences were found on the common factor at
14 and 20 months. Instead, strong shared environmental influences were observed, accounting
for most of the variance. At 24 and 36 months, genetics accounted for 34-47% of the variance
in the common empathy factor. Shared environment effects decreased from .69 at 14 months
to .00 at 36 months. Finally, the non-shared environment (and any measurement error that is
common to the four empathy measures) accounted for 31-53% of the variance across ages.
Genetic and environmental influences on the unique empathy components. Table 5
also presents, for each of the four empathy indices separately, the genetic and environmental
contributions to the variance not accounted for by the common factor. For example, in
hypothesis testing with regard to the examiner at 14 months, the common factor accounts for
40% of the variability with the remaining 60% are accounted for by an additional, unique
shared environment effect (29%) and by a unique non-shared environment effect and
measurement error (31%). Additional moderate (.16-.38) unique shared environment effects
were found for empathic concern towards the examiner at 20 months and for hypothesis
testing towards the examiner at 36 months. Genetic influences on the unique components
were found mainly for responses towards the examiner, over and above the genetic effects on
the common empathy factor. Almost no genetic influences were found with regard to
responses towards the mother (beyond those accounted for by the common factor). Finally,
Development of Empathy 20
non-shared environment effects were found for all unique components, in addition to the non-
shared environment effects on the common factor. These effects include the measurement
error unique to each of the four indices of empathy.
Change and Continuity in Genetic and Environmental Effects on Empathy over Time.
We also were interested in the contributions of genetics and the environment to
change and continuity in empathy over time. The overall empathy of twins (N=292 pairs for
which full data was available for all ages), computed as the averaged standardized scores on
empathic concern and hypothesis testing towards the mother and examiner at each age, was
analyzed using the Cholesky decomposition method, using within-twin and between-twin
multivariate variance-covariance matrices in order to decompose the variance within and
between ages into a set of genetic, shared environmental and non-shared environmental
factors. Applied to longitudinal data, the variance is decomposed so that at each age, genetic,
shared environmental, and non-shared environmental components are estimated, on which the
measures obtained in later ages can load. To the extent that scores at later and younger ages
load on the same factors, this indicates continuity. For example, if the genetic effects at 24
months and those at 36 months have substantial loadings on the same genetic factor, this
suggests the contribution of genetics to continuity. To the extent that scores at later ages do
not load on the same factors as those at younger ages, this indicates change.
Figure 3 presents the results from the Cholesky decomposition. To increase model
parsimony, paths whose coefficients were non-significant were dropped from the model. All
of these paths accounted for less than 5% of the variance, and dropping them from the model
did not significantly affect fit, χ2(df=18) = 21.07, ns. Allowing for rounding error, the
squared standardized paths shown leading to the score on empathy at each age, summed
across the A, C, and E components, account for 100% of the variance. Estimates are not
identical to those in Table 5, because the longitudinal analysis focused on the overall
Development of Empathy 21
empathy score (aggregating across examiner/mother and empathic concern/ hypothesis
testing), whereas the cross-sectional analysis in Table 5 described results for both a common
factor and for factors specific to the combinations of examiner/mother and empathic
concern/ hypothesis testing.
Genetic effects. As seen in the upper panel of Figure 3, the first genetic effects on
empathy appeared at 20 months, accounting for 9% of the variance at that age. These genetic
effects were carried on, accounting for 16% of the variance at age 24 months. At that age a
"new" genetic effect unique from the earlier genetic effect emerged. This effect accounted for
an additional 8% of the variance in empathy at 24 months (totaling a heritability of .24 at this
age). The genetic effect derived at 24 months accounted fully for the 25% heritability
estimated at 36 months. In sum, genetics accounted for both change and continuity in
empathy, but their role changed as children grew up.
Environmental effects. The analysis revealed that the shared environmental effect at
14 months was carried over but waned in importance at later ages (see also Table 5), and that
no new significant effects were found. Thus, the shared environment accounted for continuity,
but its effects became steadily weaker as children grew older. In contrast, the large non-shared
environmental effects (which include measurement error) loaded on a single factor at each
age, indicating that there was no continuity in these effects.
Prevalence of Prosocial Acts
Preliminary analyses showed no differences in the prevalence of prosocial acts by the
child's gender or the twin's zygosity. Figure 4 presents the proportion of children who
performed prosocial acts towards their mother or the examiner in at least one of the distress
simulations (home and lab), at each age. The most consistent finding is that many more
children performed prosocial acts towards their mother than towards the examiner. This
difference was significant as indicated by the McNemar test; 14 months, χ2(df=1) = 11.46, 20
Development of Empathy 22
months, χ2(df=1) = 102.15, 24 months, χ2(df=1) =95.14, 36 months, χ2(df=1) = 89.47, p <
.005 at all ages.
Prosocial behavior toward both the mother and the examiner increased with age. The
proportion of children behaving prosocially towards their mother in at least one situation
increased from 19% at 14 months to 53% at 36 months. The main increase in prosociality
towards mothers occurred from 14 to 20 months, χ2(df=1) = 24.97, p < .001. The increase
from 20 to 24 months was smaller but statistically significant, χ2(df=1) = 3.98, p < .05, and
there was further increase from 24 to 36 months, χ2(df=1) = 3.98, p < .05.
Although prosocial behavior towards the examiner also increased from 14 months
(10%) to 36 months (18%), χ2(df=1) = 6.54, p < .01, a different pattern of results was found.
A small decrease in prosociality was noted from 14 to 20 months, χ2(df=1) =7.04, p< .01,
followed by an increase from 20 to 24 months, χ2(df=1) = 12.07, p< .001, and another slight
increase from 24 to 36 months, χ2(df=1) = 3.85, p< .01. These small differences were not
fully replicated with the cotwin subsample, and should be treated with caution.
The Relationship between Empathy and Prosocial Behavior
We hypothesized that empathy towards the person in distress would be associated with
prosocial behavior towards that person. Table 6 presents the means and standard deviations of
empathic concern and hypothesis testing, comparing children who either behaved or did not
behave prosocially towards the mother or the examiner. We tested our hypothesis separately
at each age and for the mother and examiner (again, one twin was randomly chosen from each
pair for these analyses). We employed a discriminant analysis (e.g., Betz, 1987), a method of
examining the extent to which multiple predictor variables (hypothesis testing and empathic
concern) are related to a categorical criterion (performance of prosocial behavior). This
analysis, like multiple regression, provides a linear equation that maximizes differences
between children who do and do not behave prosocially. At each age, separately for examiner
Development of Empathy 23
and mother, we fitted a discriminant function for children’s hypothesis testing and empathic
concern as predictors of prosociality.
The hypothesis that empathy is related to prosociality was partially supported for the
examiner (Table 6). At 14 months, no relationship was found. At 20 and 24 months, the
overall discriminant function did not predict prosocial behavior, although a significant
contribution of hypothesis testing was found at both ages. Finally, at 36 months, both
hypothesis testing and empathic concern predicted prosocial behavior significantly but
weakly. Support for the hypothesis that empathy is positively associated with prosocial
behavior towards the mother was found at all ages. Both hypothesis testing and empathic
concern predicted children's prosocial behavior. Canonical correlations between discriminant
scores and children's prosociality towards the mother ranged from .25 to .36.
Genetic and Environmental Effects on Prosocial Behavior and its Relationship with Empathy
Next, we investigated the genetic and environmental contributions to prosocial
behavior and to its relationship with empathy at each age. To reduce the number of analyses,
we conducted this analysis on overall empathy (averaged empathic concern and hypothesis
testing) and prosocial behavior, each standardized and averaged across mother and examiner
scores at each age. We used the correlated factors model (Rijsdijk & Sham, 1999), which
specifies correlated additive genetic, shared environmental, and non-shared environmental
effects that influence prosocial behavior and empathy. The extent that the MZ cross-trait (e.g.,
prosocial behavior and empathy) twin correlation exceeds the DZ cross-trait twin correlation
indicates the degree of genetic overlap between the two traits weighted by the square roots of
heritabilities of the two traits. The model estimates the proportion of the covariance between
two variables attributed to genetic covariance between them, or bivariate heritability (Plomin
& DeFries, 1979). Bivariate shared and non-shared environmental contributions to variance
and covariance between the two variables are estimated in a similar way. Together, bivariate
Development of Empathy 24
heritability and the shared and non-shared environmental effects account for the phenotypic
correlation between empathy and prosocial behavior.
The between-twin MZ correlation for prosocial behavior (Table 7) was higher than
that for DZ twins at all ages, except for 20 months, indicating modest heritabilities (.09 to .24)
for prosocial behavior at most ages. The shared environment effect was estimated at .00 to
.16, and most of the variance was due to the non-shared environment and error (Table 7). The
genetic and environmental contributions to the phenotypic correlations between empathy and
prosocial behavior were next estimated in model-fitting analyses. At 24 months, when both
prosocial behavior and empathy showed some heritability, bivariate heritability accounted for
46% or the phenotypic correlation. In addition, at 20 months, when the shared environment
had substantial effects on both variables, a part (25%) of the phenotypic correlation was
estimated as due to bivariate shared environment effects. But the most consistent finding was
that the empathy-prosocial behavior relationship was largely due to bivariate non-shared
environment effects at all ages.
We examined the building blocks of compassion: the affective and cognitive aspects of
empathy, and prosocial behavior. This study is the largest to date of children's empathic
development in late infancy and early childhood. We found substantial developmental
changes in the prevalence of empathy, accompanied by consistency and stability across ages,
distress victims, and the affective and cognitive modalities of empathy. Viewed as a relatively
stable individual differences variable, empathy has been accounted for by variable genetic and
environmental components, with the former increasing with age and the latter decreasing with
age, and with the environment accounting for the relationship between empathy and prosocial
behavior at this young age.
Prevalence of Empathy and Prosocial behavior
Development of Empathy 25
Both empathy and prosocial behavior increased during the first years of life. This
pattern is mainly consistent with results from previous research with other longitudinal
samples (Volbrecht, et al., 2007; Zahn-Waxler & Radke-Yarrow, 1982; Zahn-Waxler, Radke-
Yarrow, et al., 1992a). The overall increase in the affective, cognitive and behavioral aspects
of compassion may reflect the development of emotion regulation, self-other differentiation
and perspective taking abilities needed for empathy, as well as language and interpersonal
skills needed in order to approach distressed others, inquire about their feelings, and provide
support (Hoffman, 2000; Underwood & Moore, 1982; Zahn-Waxler, Radke-Yarrow, et al.,
1992). While the increase in empathic concern took place mainly from 14 to 20 months,
complex modes of hypothesis testing further increase in the later ages, possibly reflecting
children's increased ability for complex, verbal forms of inquiry.
Both hypothesis testing and prosocial behavior were higher towards the mother than
towards the examiner. This is in line with the deeper involvement children have with their
mothers (Robinson, Zahn-Waxler, et al., 2001). Meaningful (e.g. mother-child) relationships,
framed in terms of reciprocated positive behaviors, are likely to result in an overall positive
attitude toward the other individual, and even increased self-relevance of the other’s suffering.
The difference favoring mothers increased with age, possibly reflecting the increasing
mother-child closeness and emotional investment in the relationship. Moreno et al. (in press)
found that aspects of the mother’s emotional availability toward her child early in the second
year of life were internalized by the child and reflected in the child’s social engagement
toward mother by the end of the second year. These internalized relationship skills accounted
for substantial variance in observed empathy at age 2 toward mothers and to a somewhat
lesser extent, toward examiners. The difference favoring the mothers was not found for
empathic concern. This contrasts with Young et al.'s (1999) finding of higher empathic
concern towards mothers at 24 months. These mixed results call for further study.
Development of Empathy 26
Consistency, Continuity, and Change in Empathy
As expected, we found support for the role of an overall empathic tendency,
generalizing across distress victims, affect and cognition, and age. At each age, the common
empathy factor accounted for 42-55% of the variance across the single empathy measures.
Across ages, using the overall empathy scores, empathy at earlier ages accounted for 32-58%
of the variance in later empathy. (Figure 3). Combined with the substantial relationship
between empathy and prosocial behavior, these findings support the existence of an altruistic,
empathic or prosocial personality (e.g., Eisenberg et al., 1999).
Although moderate support for an overall empathy disposition was found, there was
also evidence for the moderating role of social context (mother vs. examiner) and type of
empathy (hypothesis testing vs. empathic concern). Similarly, the longitudinal continuity was
significant, but substantial changes were simultaneously observed. Longitudinal correlations
were stronger for overall empathy than for any single index of empathy. The emergent picture
is that of empathy as a multifaceted construct (Singer, 2006). The different aspects of
empathy are connected through their common underlying empathy disposition, but are also
unique to some extent because they are affected by the factors of social context, modality of
response (cognitive vs. affective), and developmental stage.
Meaningful cross-situational variability concomitant with a stable disposition has been
observed with regard to children's behavior and conceptualized as part of personality by
Mischel and Shoda (1995). Similarly, sadness, anxiety/fear, and anger have been shown to
have both unique components and a shared pattern of overall negative affect (Watson &
Clark, 1992). The differences in response to the mother and examiner reflect a variation of
response to situational constraints discussed below.
There were meaningful differences between empathic concern and hypothesis testing
which show that although the two share an underlying common tendency, there are also
Development of Empathy 27
unique features... The correlations of .27 to .52 between empathic concern and hypothesis
testing indicate moderate consistency accompanied by variability, and suggest the two
constructs are partially separable. In addition, hypothesis testing rates increased slowly but
steadily with age, while empathic concern rates increased mainly from 14 to 20 months,
suggesting partially different developmental trajectories.
The differential rates of development of empathic concern and hypothesis testing are
compatible with neuro-scientific evidence. The limbic and para-limbic systems, relevant to
the more affective aspects of empathy, develop earlier than the prefrontal and temporal
cortices, which are more relevant to the cognitive aspect of empathy indexed by hypothesis
testing (Singer, 2006). This differential developmental pattern has been suggested as the basis
for expecting that the cognitive aspects of empathy develop later than the affective aspects of
empathy (Singer, 2006). Genetic influences on these maturational processes may also account
for the differential rates of development. This can be learned from the genetic contributions to
the variance in the empathy indices not accounted for by the common empathy factor (Table
5). Thus, from 20 months on, the genetic contributions to the unique variance in hypothesis
testing were larger than those of empathic concern, indicating that some of the later genetic
processes taking place were relevant to hypothesis testing but not to empathic concern.
The lowest stability was found in the latest measurement age, suggesting that the
dynamics of development continue to exert substantial influences that reduce the generality
observed in the earlier ages. The ways in which children experience and express empathy
change markedly in the life period studied in the present investigation. In this period, children
move from a basic self-differentiation that enables them to project their own concern on
others, to a more sophisticated stage in which they care about what others feel as well as
about the way their reaction would be seen by others (Hoffman, 2000; Rochat, 2002). These
changes may result not only in the increase in overall empathy levels we observed, but also in
Development of Empathy 28
the psychological antecedents of empathy at different ages (Lewis, 2000). In other words, the
changing nature of empathy may mean that the relevance of different genetic, environmental
and psychological factors changes with age.
Genetic and Environmental Influences
The dynamic changes in empathy are accompanied by meaningful shifts in the genetic
and environmental contributions to individual differences in empathy. As for other traits (e.g.,
Knafo & Plomin, 2006a; Plomin et al., 2001), the heritability of empathy tended to increase
with age. In addition, the environmental influences on the common empathy factor became
increasingly of the non-shared sort.
Because our design followed children in four consecutive time points, we were able to
identify the second half of the second year of life as the period in which the main changes in
genetic effects on the common empathy factor occur. At 14 months, no overall genetic effect
was found. By 24 months, a genetic effect accounting for about a quarter of the variance
emerged, and remained stable towards 36 months as no further new genetic effects emerged,
and heritability at 36 months loaded fully on the genetic components emerging in earlier ages
(Figure 3). This means that a substantial change occurs in this period that is mainly due to
genetic factors. This developmental period includes major transitions in self-other
differentiation, children's affective regulation, and cooperative play (e.g., Brownell &
Carriger, 1990; Eckerman, Davis & Didow, 1989; Hay, 1979; Nielsen & Dissanayake, 2004;
Zahn-Waxler et al., 2001). For example, Moreno et al. (in press) demonstrated that children’s
cognitive and language gains in the first two years as well as their social engagement skills
contributed significantly to their observed empathy at age two.
In addition to providing evidence for the importance of both heritability and the
environment to individual differences in children's overall empathy, we estimated genetic and
environment contributions to specific empathy indices. Most unique genetic effects occurred
Development of Empathy 29
with regard to the examiner. Robinson, Zahn-Waxler, et al. (2001) suggested that children's
natural inclination towards high or low empathy may be more apparent with strangers (such
as our examiners) for which no relationship-specific rules have been established with time.
This pattern demonstrates the importance of context to development, and further stresses the
need to consider heritability as the proportion of individual variance affected by genetic
factors in a specific context (e.g., age and relationship) (Plomin et al., 2001).
Research is still far from elucidating the processes through which genetic variations
affect brain processes relevant to empathy. One way for future research to proceed is to
identify the temperamental constellations predictive of empathy, and look for genetic
influences common to temperament and empathy (Zahn-Waxler, Robinson, et al., 1992). For
example, children's behavioral inhibition was negatively associated with empathy towards an
experimenter (Young et al., 1999). To the extent that behavioral inhibition is heritable (e.g.,
DiLalla, , Kagan, & Reznick, 1994), it could mediate the effects of genetics on empathy. As
another example, a temperament characterized by low affect was associated with lower
empathy towards an examiner (Young et al., 1999). Possibly, genetics indirectly influence
empathy through their effects on an overall capacity to experience emotions. An extensive
literature documents a relationship between low arousal and antisocial behavior which is
characterized by little or no empathy (Lahey, Hart, Pliszka, Applegate, & McBurnett, 1993).
Sociability is yet another trait that shows heritability and is associated with empathy
Genetic influences on variability in the functioning of brain regions relevant to empathy
and mirror neuron functioning such as the ventromedial prefrontal cortex (Amodio & Frith,
2006) and the anterior insula (Singer, Seymour, O’Doherty, Kaube, Dolan & Frith, 2004) are
likely candidates. Evidence from populations with extremely low (Viding, Blair, Moffitt, &
Development of Empathy 30
Plomin, 2005) or abnormally high (Meyer-Lindenberg et al., 2005; Zahn-Waxler, Shirtcliff, &
Marceau, 2008) levels of empathy could also be relevant.
Genetic influences on most psychological traits take the form of many genes affecting
behavior, cognition, and affect in an additive or interactive manner, and it would be unlikely
for a single gene to have strong influences on the normal variation in a trait (see Plomin et al.,
2001). Nevertheless, a careful look for specific genetic influences on empathy may prove
fruitful. Hastings et al. (2006) proposed examining the genes associated with the serotonergic
systems, as they are relevant to various affective processes (Hariri & Weinberger, 2003).
Recent evidence calls for investigating the arginine vasopressin 1a (AVPR1a) receptor gene,
which is relevant to human altruism and autism - two extreme sides of an empathy-
prosociality dimension (Knafo et al., 2008; Israel et al., 2008; Yirmiya et al., 2006). As
genetic variability in this gene has been related to behavioral and social differences within and
between other mammalian species (Hammock & Young, 2005), it may be one link in which
the cross-species empathy patterns described by de Waal (2007) may operate.
Empathy and Prosocial Behavior
Relying on the premise that empathy provides a motivational impetus for prosocial
behavior (Batson, in press), we hypothesized that empathy would relate to prosocial
behavior. The hypothesis was largely confirmed, with prosocial behavior at each age being
positively predicted by empathic responding. The results were more consistent for the
mothers, again demonstrating the role of the context in which children operate. Moreover,
children may have an established script for helping their mother, based to a great extent on
her own behavior towards them (Robinson, Zahn-Waxler, et al., 2001), whereas generalizing
behaviors modeled by the mother to the examiner's distress is likely to need more effort and
more advanced sociocognitive abilities.
Development of Empathy 31
Because children's behavioral repertoire becomes more flexible as they grow older
and encounter increasingly varied situations, their ability to come up with a prosocial act
intended to alleviate the victim's distress increases with age. The relationships between
empathy and prosocial behavior were stronger, for both the mother and the examiner, at 24
and 36 months than in the younger ages. The increase in the prosocial behavior-empathy
relationships is also compatible with Hoffman's (2000) notion that development in the
second year of life enables empathy to become less self-oriented and more other-oriented.
The increased self-regulation abilities achieved towards age 2 enable children to not only
feel for the victim, but also act upon these feelings.
Although there was evidence of genetic influences on empathy, and to a somewhat
lesser degree on prosocial behavior in a distress situation, the prosocial behavior-empathy
relationship was accounted for mainly by environmental factors. The environmental factors
likely to be relevant to both prosocial behavior and empathy and which account for the
relationship between prosocial behavior and empathy include parental warmth, inductive
discipline, and responsiveness (Hastings et al., 2006). These may be experienced similarly
(shared environment) or differentially (non-shared environment) within twin pairs, both of
which could influence overall levels of prosociality. Nevertheless, the inconsistent findings
in the literature regarding the strength of heritability of prosocial behavior in distress
situations suggest the need for further research with a wide range of observationally
measured prosocial behavior, including children’s responses to other children.
Strengths and Limitations
This is the largest genetically-informative study of young children's empathy. The
assessment of the same children at four different ages, across early childhood, is a
methodological strength of this study. This period of nearly two years between the first and
fourth measurement allowed us to examine the development of empathy in an important
Development of Empathy 32
developmental period accompanied by many social and cognitive changes. It also enabled
investigating change and stability as well as their genetic and environmental origins.
We obtained repeated assessments of the same children (four times, twice at the lab
and twice at home at each of the 4 time points) with high retention rates. This well exceeds
the number of assessments from any other studies of early empathic and prosocial
development and should have contributed to increased reliability and validity. However,
even here, genetic and environmental estimates will lack some precision. Although a sample
size of about 800 children would be considered a large one for most research purposes, it is a
modest sample for a twin study. This is evident in the wide confidence intervals obtained for
the genetic estimates, particularly for the heritability of the common empathy factor at 24
months. Generalizations therefore must be treated with some degree of caution. Finally, in
future research, it will be important to begin to measure some of the specific biological,
dispositional, and environmental (particularly parental socialization) processes implicated in
the early development of empathy and prosociality (see, e.g., Knafo & Plomin, 2006b).
This study addressed the origins and development of empathy. Empathic concern and
hypothesis testing increased with age, as did prosocial behavior. The relationships between
these three variables as well as the differences between them emphasize the importance of
viewing compassion as a complex constellation of affective, cognitive, and behavioral
components. The positive concurrent and longitudinal correlations indicated that early in
development, empathy is already a relatively stable disposition, generalizing across its
affective and cognitive components, across mother and examiner, and over time. However,
the increase in empathy and in its associated genetic influences, and the change in the
relevance of shared environmental factors, suggest the second and third years of life should
be considered a period of dynamic changes and developments in children's empathy.
Development of Empathy 33
Amodio, D.M., & Frith, C. D. (2006). Meetings of the minds: the medial frontal cortex and social
cognition. Nature Reviews Neuroscience, 7, 268-277.
Arbuckle, J. L. (1997). Amos users guide: Version 3.6. Chicago: Smallwaters.
Barnett, M. A., & Thompson, S. (1985). The role of perspective taking and empathy in children’s
Machiavellianism, prosocial behavior and motive for helping. Journal of Genetic Psychology,
Batson, C. D. (in press). Empathy-induced altruistic motivation. In: P. R. Shaver & M. Mikulincer
(Eds.), Prosocial Motives, Emotions, and Behavior. American Psychological Association
Betz, N. E. (1987). Use of discriminant analysis in counseling psychology research. Journal of
Counseling Psychology, 34, 393-403.
Brownell, C. A. & Carriger, M. S. (1990). Changes in cooperation and self-other differentiation
during the second year. Child Development, 61, 1164-1174.
Decety, J., & Jackson, P. L. (2004). The functional architecture of human empathy. Behavioral and
Cognitive Neuroscience Reviews, 3, 71−100.
DiLalla, L. F., Kagan, J., & Reznick, J. S. (1994). Genetic etiology of behavioral inhibition among
2-year-old children. Infant Behavior and Development 17, 405–412.
Eckerman, C. O., Davis, C. C., & Didow, S. M. (1989). Toddlers’ emerging ways of achieving social
coordinations with a peer. Child Development, 60, 440-453.
Eisenberg, N., & Fabes, R. (1998). Prosocial development. In W. Damon (Editor-in-Chief ) & N.
Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional, and personality
development (5th ed., pp. 701–778). New York: Wiley.
Development of Empathy 34
Eisenberg, N., Shell, R.., Pasterneck, J.., Lennon, R.., Beller, R.., & Mathy, R. (1987). Prosocial
development in middle childhood: A longitudinal study, Developmental Psychology, 23, 712-
Eisenberg, N., Guthrie, I. K., Murphy, B. C., Shepard, S. A., Cumberland, A., & Carlo, G. (1999).
Consistency and development of prosocial dispositions: A longitudinal study. Child
Development, 70, 1360–1372.
Gill, K.L. & Calkins, S.D. (2003). Do aggressive/destructive toddlers lack concern for others?
Behavioral and physiological indicators of empathic responding in 2-year-old children,
Development and Psychopathology, 15, 55-71.
Hammock, E. A., and Young, L. J. (2005) Microsatellite instability generates diversity in brain and
sociobehavioral traits. Science, 308, 1630-1634.
Hariri, A. R., & Weinberger, D. R. (2003). Functional neuroimaging of genetic variation in
serotonergic neurotransmission. Genes, Brain and Behavior, 2, 341-349.
Hastings, P.D., Zahn-Waxler, C., & McShane, K.E. (2006). We are,by nature, moral creatures:
Biological bases of concern for others. In M. Killen & J. Smetana (Eds.), Handbook of Moral
Development . NY: Lawrence Erlbaum, Inc.
Hay, D. F. (1979). Cooperative interactions and sharing between very young children and their
parents. Developmental Psychology 15, 647-653.
Hoffman, M. L. (1975). Altruistic behavior and the parent-child relationship. Journal of Personality
and Social Psychology, 31, 937–943.
Hoffman, M. L. (1988). Moral development. In M. Bornstein & M. Lamb (Eds.), Developmental
psychology: an advanced textbook (pp. 497–548). Hillsdale, NJ: Erlbaum.
Hoffman, M. L. (2000). Empathy and moral development: Implications for caring and justice. New
York: Cambridge University Press.
Development of Empathy 35
Hur, Y., & Rushton, J. P. (2007). Genetic and environmental contributions to prosocial behaviour in
2- to 9-year-old South Korean twins. Biology Letters, 3, 664-666.
Israel, S., Lerer, E., Shalev, I., Uzefovsky, F., Reibold, M., Bachner-Melman, R., Granot, R.,
Bornstein, G., Knafo, A., Yirmiya, N., & Ebstein, R.P. (2008). Molecular genetic studies of the
arginine vasopressin 1a receptor (AVPR1a) and the oxytocin receptor (OXTR) in human
behavior: From autism to altruism with some notes in between. In I.D. Neumann and R.
Landgraf (Eds.) Advances in vasopressin and oxytocin — from genes to behaviour to disease,
Progress in Brain Research, Vol. 170 (pp. 435-449), Progress in Brain Research, Amsterdam:
Knafo, A., & Plomin, R. (2006a). Prosocial behavior from early to middle childhood: Genetic and
environmental influences on stability and change. Developmental Psychology, 42, 771-786.
Knafo, A., & Plomin, R. (2006b). Parental discipline and affection, and children’s prosocial
behavior: Genetic and environmental links. Journal of Personality and Social Psychology, 90,
Knafo, A., Israel, S., Darvasi, A., Bachner-Melman, R., Uzefovsky, F., Cohen, L., et al. (2008).
Individual differences in allocation of funds in the Dictator Game and postmortem
hippocampal mRNA levels are correlated with length of the arginine vasopressin 1a receptor
(AVPR1a) RS3 promoter-region repeat. Genes, Brain and Behavior, 7, 266-275.
Lahey, B.B., Hart, E.L., Pliszka, S., Applegate, B., & McBurnett, K. (1993). Neurophysiological
correlates of conduct disorder: A rationale and review of research. Journal of Clinical Child
Psychology, 22, 141-153.
Lewis, M. (2000). Self-conscious emotions: Embarrassment, pride, shame, and guilt. In M. Lewis &
J. Haviland-Jones (Eds.), Handbook of emotions (pp. 623-636). New York: Guilford Press.
Development of Empathy 36
van der Mark, I. L.,van Ijzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2002). Development of
empathy in girls during the second year of life: Associations with parenting, attachment, and
temperament. Social Development, 11, 451–468.
Meyer-Lindenberg, A., Hariri, A. R., Munoz, K. E., Mervis, C. B., Mattay, V. S., Morris, C. A., et
al. (2005). Neural correlates of genetically abnormal social cognition in Williams syndrome.
Nature Neuroscience, 8, 991-993.
Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality:
Reconceptualizing situations, dispositions, dynamics, and invariance in personality structure.
Psychological Review, 102, 246-268.
Moreno, A. J., Klute, M. M., & Robinson, J. L. (in press). Relational and individual resources as
predictors of empathy in early childhood. Social Development.
Neale, M. C., Boker, S. M., Xie, G., & Maes, H. H. (1999). Mx: Statistical Modeling (5th ed.).
Richmond, VA: Virginia Commonwealth University, Department of Psychiatry.
Nichols, R. C., & Bilbro, W. C. (1966). The diagnosis of twin zygosity. Acta Geneticae Medicae et
Statistica, 16, 265–275.
Nielsen, M., & Dissanayake, C. (2004). Pretend play, mirror self-recognition and imitation: A
longitudinal investigation through the second year. Infant Behavior and Development, 27, 342–
Plomin, R., & DeFries, J. C. (1979). Multivariate behavioral genetic analysis of twin data on
scholastic abilities. Behavior Genetics, 9(6), 505-517.
Plomin, R., DeFries, J. C., McClearn, G. E., & McGuffin, P. (2001). Behavioral Genetics (4th ed.).
New York: Worth Publishers.
Plomin, R., Emde, R. N., Braungart, J. M., Campos, J., Corley, R., Fulker, D.W., Kagan, J., Reznick,
J.S., Robinson, J., Zahn-Waxler, C., & DeFries, J.C. (1993). Genetic change and continuity
Development of Empathy 37
from fourteen to twenty months: The MacArthur Longitudinal Twin Study. Child
Development, 64, 1354-1376.
Preston, S. D., & de Waal, F. B. M. (2002). Empathy: Its ultimate and proximate bases. Behavioral
and Brain Sciences, 25, 1–72.
Rhea, S., Gross, A. A., Haberstick, B. C., & Corley, R. P. (2006). Colorado twin registry. Twin
Research and Human Genetics, 9, 941-949.
Rijsdijk, F.V., & Sham, P.C. (2002). Analytic Approaches to Twin Data using Structural Equation
Models. Briefings in Bioinformatics, 3, 119-133.
Roberts, W., & Strayer, J. (1996). Empathy, emotional expressiveness, and prosocial behavior. Child
Development, 67, 449–470.
Robinson, J.L., McGrath, J., & Corley, R.P. (2001). The conduct of the study. In R. N. Emde & J. K.
Hewitt ( Eds.), Infancy to early childhood: Genetic and environmental influences on
developmental change (pp. 23– 31). New York: Oxford University Press.
Robinson, J. L., Zahn-Waxler, C., & Emde, R. N. (2001). Relationship context as a moderator of
sources of individual difference in empathic development. In R. N. Emde & J. K. Hewitt
(Eds.), Infancy to early childhood: Genetic and environmental influences on developmental
change (pp. 257–268). Oxford, England: Oxford University Press.
Rochat, P. (2002). Various kinds of empathy as revealed by the developing child, not the monkey’s
brain, Behavioral and Brain Sciences, 25, 45-46.
Rothbart, M. K., Ahadi, S. A., & Hershey, K. L. (1994). Temperament and social behavior in
childhood. Merrill-Palmer Quarterly, 40, 21–39.
Scourfield, J., John, B., Martin, N., & McGuffin, P. (2004). The development of prosocial behaviour
in children and adolescents: A twin study. Journal of Child Psychology and Psychiatry, 45,
Development of Empathy 38
Singer, T. (2006). The neuronal basis and ontogeny of empathy and mind reading: Review of
literature and implications for future research, Neuroscience and Biobehavioral Reviews, 30,
Singer, T., Seymour, B., O’Doherty, J., Kaube, H., Dolan, R., Frith, C. D., (2004). Empathy for pain
involves the affective but not sensory components of pain. Science, 303(5661), 1157-1162.
Underwood, B., & Moore, B. (1982). Perspective-taking and altruism. Psychological Bulletin, 91,
Valiente, C., Eisenberg, N., Fabes, R. A., Shepard, S. A., Cumberland, A., & Losoya, S. H. (2004).
Prediction of children’s empathy-related responding from their effortful control and parents’
expressivity. Developmental Psychology, 40, 911–926.
Viding, E., Blair, R. J. R., Moffitt, T. E., & Plomin, R. (2005). Evidence for substantial genetic risk
for psychopathy in 7-year-olds. Journal of Child Psychology and Psychiatry, 46, 592-597.
de Vignemont, F., & Singer, T. (2006) The empathic brain: How, when and why? Trends in
Cognitive Sciences, 10, 435-441.
Volbrecht, M. M., Lemery-Chalfant, K., Aksan, N., Zahn-Waxler, C., & Goldsmith, H. H. (2007).
Examining the familial link between positive affect and empathy development in the second
year. Journal of Genetic Psychology, 168, 105-129.
de Waal, F. B. M. (2007). Putting the altruism back into altruism: The evolution of empathy. Annual
Review of Psychology, 59, 4.1-4.22.
Watson, D., & Clark, L. A. (1992). Affects separable and inseparable: On the hierarchical
arrangement of the negative affects. Journal of Personality and Social Psychology, 62, 489-
Yirmiya, N., Rosenberg, C., Levi, S., Salomon, S., Shulman, C., Nemanov, L., et al. (2006).
Association between the arginine vasopressin 1a receptor (AVPR1a) gene and autism in a
family-based study: Mediation by socialization skills. Molecular Psychiatry, 11, 488-494.
Development of Empathy 39
Young, S. K., Fox, N. A., & Zahn-Waxler, C. (1999). The relations between temperament and
empathy in 2-year-olds. Developmental Psychology, 35, 1189–1197.
Zahn-Waxler, C. & Radke-Yarrow, M. (1982). The development of altruism: Alternative research
strategies. In N. Eisenberg-Berg (Ed.) The development of prosocial behavior. New York:
Academic Press, 109-137.
Zahn-Waxler, C. & Radke-Yarrow, M. (1990). Research on children of affectively ill parents: Some
considerations for theory and research on normal development. Development and
Psychopathology, 2, 349-366.
Zahn-Waxler, C., Radke-Yarrow, M., Wagner, E., & Chapman, M. (1992). Development of concern
for others. Developmental Psychology, 28, 126–136.
Zahn-Waxler, C., Robinson, J., & Emde, R. N. (1992). The development of empathy in twins.
Developmental Psychology, 28, 1038-1047.
Zahn-Waxler, C., Schiro, K., Robinson, J. L., Emde, R. N., & Schmitz, S. (2001). Empathy and
prosocial patterns in young MZ and DZ twins: Development and genetic and environmental
influences. In: R. N. Emde & J. K. Hewitt, (Eds.), Infancy to early childhood: Genetic and
environmental influences on developmental change (pp. 141-162). New York: Oxford
Zahn-Waxler, C., Shirtcliff, E.A., & Marceau, K. (2008). Disorders of childhood and adolescence:
Gender and psychopathology. Annual Review of Clinical Psychology. 4, 11.1-11.29.
Development of Empathy 40
The study was supported by grants from the Fetzer Institute and the John D. & Katharine T.
McArthur Foundation to the second author. The work of the first author was supported by a
grant from the Israel Science Foundation. We thank the families participating in this study
and the research assistants who collected and coded the data.
Development of Empathy 41
Empathy of Children towards the Mother or the Examiner
Empathic concern Hypothesis-testing
Examiner Mother Examiner Mother
2.26 2.10 2.27 2.15
0.49 0.56 0.49 0.47
2.48 2.52 2.34 2.40
0.55 0.67 0.51 0.63
2.51 2.41 2.56 2.79
0.50 0.58 0.56 0.68
2.61 2.52 2.74 3.00
0.51 0.62 0.66 0.75
2.38 2.18 2.27 2.24
0.54 0.66 0.47 0.49
2.56 2.64 2.32 2.50
0.52 0.66 0.50 0.58
2.59 2.58 2.53 2.90
0.51 0.57 0.52 0.63
2.62 2.53 2.56 2.92
0.50 0.60 0.61 0.69
Development of Empathy 42
Correlations between Empathic Concern and Hypothesis Testing towards the Mother and the Examiner
14 months Empathic concern - mother
Hypothesis testing - examiner
Hypothesis testing - mother
.17** .49** .51**
20 months Empathic concern - mother
Hypothesis testing - examiner
Hypothesis testing - mother
.24** .50** .54**
24 months Empathic concern - mother
Hypothesis testing - examiner
Hypothesis testing - mother
.21** .52** .40**
36 months Empathic concern - mother
Hypothesis testing - examiner
Hypothesis testing - mother
.10 .40** .25**
** p< .01.
Development of Empathy 43
Longitudinal correlations in children's empathy
(months) 14 20 24 14 20 24
Examiner 20 .20** .44**
24 .10 .19** .46** .50**
36 .05 .15** .16** .31** .21** .41**
Mother 20 .16**
24 .14** .22** .21** .24**
36 .07 .12* .15** .17** .18** .11*
Note: Ns for these analyses ranged from 309 to 343 children.
* p < .05; ** p< .01.
Development of Empathy 44
Twin Intraclass Correlations in Empathy
Age Aspect of empathy
14 months Empathic concern Examiner .25** .12
Mother .18** .10
Hypothesis testing Examiner .56** .65**
Mother .40** .39**
20 months Empathic concern Examiner .26** .24**
Mother .12* .17*
Hypothesis testing Examiner .57** .43**
Mother .23** .48**
24 months Empathic concern Examiner .22** .02
Mother .23** .18*
Hypothesis testing Examiner .53** .40**
Mother .41** .30**
36 months Empathic concern Examiner .19** .04
Mother .18** .13
Hypothesis testing Examiner .50** .30**
Mother .33** .28**
* p < .05; ** p< .01.
Estimates of Variance Components (and 95% Confidence Intervals) accounting for Individual Differences in a Common Empathy Factor and in Specific Factors.
Empathic concern Hypothesis testing
Variance accounted for by
Examiner Mother Examiner Mother
14 .13 (.08 - .19) .26 (.19 - .34) .40 (.30 - .49) .58 (.49 - .66)
.00 (.00 - .00) .18 (.07 - .28) .00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00)
Shared environment .69 (.58 - .78) .00 (.00 - .00) .00 (.00 - .00) .29 (.21 - .37) .00 (.00 - .00)
Non- shared environment .31 (.22 - .42) .69 (.58 - .81) .74 (.66 - .81) .31 (.25 - .39) .42 (.34 - .51)
20 Variance accounted for by
.21 (.14 - .29) .32 (.25 - .40) .42 (.33 - .51) .62 (.52 - .70)
.00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00) .26 (.17 - .35) .00 (.00 - .00)
Shared environment .56 (.44 - .66) .16 (.07 - .24) .00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00)
Non- shared environment .44 (.34 - .56) .63 (.53 - .74) .68 (.60 - .75) .32 (.25 - .41) .38 (.30 - .48)
24 Variance accounted for by
.17 (.10 - .24) .43 (.35 - .51) .25 (.17 - .33) .60 (.51 - .69)
.34 (.00 - .74) .13 (.01 - .23) .00 (.00 - .00) .37 (.27 - .46) .00 (.00 - .00)
Shared environment .30 (.00 - .63) .00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00)
Non- shared environment .36 (.24 - .51) .71 (.59 - .83) .57 (.49 - .65) .39 (.31 - .48) .40 (.31 - .49)
36 Variance accounted for by
.13 (.07 - .21) .40 (.26 - .59) .08 (.02 - .16) .31 (.20 - .46)
.47 (.26 - .66) .12 (.00 - .24) .00 (.00 - .00) .00 (.00 - .00) .24 (.12 - .35)
Shared environment .00 (.00 - .00) .00 (.00 - .00) .00 (.00 - .00) .38 (.28 - .46) .00 (.00 - .00)
Non- shared environment .53 (.34 - .74) .75 (.62 - .88) .60 (.41 - .74) .54 (.46 - .64) .45 (.33 - .57)
Development of Empathy
Results of Discriminant Analysis of Children's Empathy as Predictor of Prosocial Behavior towards the Mother or the Examiner
Note: Absence=children who did not behave prosocially towards the mother or examiner; Presence: children who behaved prosocially at least in one
distress simulation towards the mother or examiner; SDC= Standardized discriminant coefficient; WL=Wilks's Lambda.
* p < >05; ** p < .001
WL χ2 (df=2)
2.30 (.53) 2.42 (.45) 1.00 1.94 (1,388) 1.01
2.25 (.48) 2.21 (.42) 1.00 0.30 (1,388) -0.65
.99 3.07 .09
2.50 (.56) 2.76 (.56) .99 3.83 (1,347) 0.60
2.32 (.51) 2.55 (.50) .99 3.90* (1,347) 0.61
.98 5.57 .13
2.54 (.51) 2.60 (.51) 1.00 0.61 (1,352) 0.07
2.53 (.55) 2.73 (.54) .99 5.19* (1,352) 0.98
.99 5.16 .12
2.55 (.52) 2.78 (.46) .97 9.40** (1,338) 0.62
2.59 (.63) 2.88 (.64) .97 9.87** (1,338) 0.65
.96 15.05** .21
2.07 (.61) 2.47 (.48) .93 27.27** (1,385) 0.93
2.15 (.47) 2.34 (.50) .98 8.98** (1,385) 0.13
.93 26.61** .26
2.44 (.61) 2.75 (.66) .85 20.52** (1,346) 0.73
2.34 (.55) 2.58 (.63) .92 13.75** (1,346)0. 42
.94 23.05** .25
2.27 (.57) 2.66 (.51) .89 43.35** (1,347) 0.75
2.62 (.68) 2.98 (.59) .96 27.55** (1,347) 0.39
.88 45.76** .35
2.29 (.57) 2.72 (.57) .85 48.31** (1,335) 0.95
2.81 (.76) 3.07 (.70) .97 10.87** (1,335) 0.13
.87 45.58** .36
Development of Empathy
Table 7. Twin correlations and Variance Component Estimates for Prosocial Behavior, Phenotypic Correlations between Prosocial Behavior and
Empathy, and Proportion of Correlation Mediated by Bivariate Genetic and Environmental Factors
* p < .05; ** p < .01
Prosocial behavior Prosocial behavior and empathy
Variance component estimates and 95% confidence intervals
MZDZ Heritability Shared environment Non- shared
environment and error
14 .20* -.01 .17 (.05 - .29) .00(.00- .00) .83(.71 - .95).12* 0% 0% 100%
20 .09 .22* .00 (.00- .00) .15(.05- .25).85 (.75 - .95) .27* 0% 25% 75%
24 .25** .18* .09 (.004 - .37) .16(.00 - .28) .76(.63 - .87).28* 46% 0% 54%
36 .33**.27** .24 (.00- .44).09 (.00 - .35) .66(.55 - .79).29* 0% 0% 100%
Girls - MotherGirls - Examiner
Boys - Mother Boys - Examiner
3624 20 14
36 2420 14
Girls - Mother Girls - Examiner
Boys - MotherBoys - Examiner
36 24 20 14
36 2420 14
Means (and 95% Confidence Intervals) of Empathic Concern (Broken Line) and Hypothesis Testing
(Full Line) Scores across Ages.
a. Empathic Concern and Hypothesis Testing across children's gender and mothers/examiners
b. Results presented separately for girls and boys, mothers and examiners
Development of Empathy
Figure 2. Common-Factor-Common-Pathways Multivariate Model of Genetic and Environmental Effects on Empathy.
Note. Rectangles indicate observed scores on empathy. Circles indicate the common empathy factor and the variance components estimates. A =
Heritability; C = Shared environment; E = Non-shared environment (and error) contributions to the common factor. For each observed score unique
variance components are also estimated, for which a = Heritability; c = Shared environment; e = Non-shared environment (and error) contributions to the
unique variance of each observed score.
Development of Empathy
Longitudinal Cholesky Decomposition of Variance Components of Empathy
a. Genetic components
b. Shared environment components
c. Non-shared environment components
Note. The three panels represent a single analysis, and appear separately to simplify presentation. Circles indicate
variance components estimates, and rectangles indicate observed scores on empathic behavior. A = Heritability;
C = Shared environment; E = Non-shared environment (and error). The number in each circle represents the age
to which the variance component is attributed. Numbers in parentheses are 95% confidence intervals.
(.06 - .27)
(.03 - .16)
(.03 - .19)
(.14 - .36)
(.35 - .51)
(.25 - .42)
(.04 - .15)
(.16 - .33)
(.43 - .61)
(.49 - .65)
(.56 - .67)
E14 E20 E24 E36
(.49 - .65)