Journal of Family Psychology
1996,
Vol. 10, No. 3, 243-268
Copyright 1996 by the American Psychological Association, Inc.
O893-32O0/9OT3.OO
Parental Meta-Emotion Philosophy and the Emotional Life
of Families: Theoretical Models and Preliminary Data
John M. Gottman, Lynn Fainsilber Katz, and Carole Hooven
University of Washington
This article introduces the concepts of parental meta-emotion, which refers to parents'
emotions about their own and their children's emotions, and meta-emotion philosophy,
which refers to an organized set of thoughts and metaphors, a philosophy, and an
approach to one's own emotions and to one's children's emotions. In the context of a
longitudinal study beginning when the children were 5 years old and ending when they
were 8 years old, a theoretical model and path analytic models are presented that relate
parental meta-emotion philosophy to parenting, to child regulatory physiology, to
emotion regulation abilities in the child, and to child outcomes in middle childhood.
The importance of parenting practices for
children's long-term psychological adjustment
has been a central tenet in developmental and
family psychology. In this article, we introduce
a new concept of parenting that we call parental
meta-emotion philosophy, which refers to an
organized set of feelings and thoughts about
one's own emotions and one's children's emo-
tions.
We use the term meta-emotion broadly to
encompass both feelings and thoughts about
emotion, rather than in the more narrow sense
of one's feelings about feelings (e.g., feeling
guilty about being angry). The notion we have
in mind parallels metacognition, which refers to
the executive functions of cognition (Allen &
Armour, 1993; Bvinelli, 1993; Flavell, 1979;
Fodor, 1992; Olson & Astington, 1993). In an
analogous manner, meta-emotion philosophy
John M. Gottman, Lynn Fainsilber Katz, and Car-
ole Hooven, Department of Psychology, University
of Washington.
The research of this article was supported by Na-
tional Institute of Mental Health (NIMH) Research
Grants MH42484 and MH35997, by an NIMH Merit
Award to extend research in time, and by Research
Scientist Award K2MH00257.
This research received a great deal of support from
Michael Guralnick, director of the Center for Human
Developmental Disabilities (CHDD), and CHDD's
core facilities, particularly the Instrument Develop-
ment Laboratory at the University of Washington.
Correspondence concerning this article should be
addressed to John M. Gottman, Department of Psy-
chology, Box 351525, University of Washington,
Seattle, Washington 98115.
refers to executive functions of emotion. In this
article, we discuss the evolution of the meta-
emotion construct and describe its relationship
to various aspects of family and child function-
ing. We present a parsimonious theoretical
model of the role of parental meta-emotions in
children's emotional development, operational-
ize this model, and present some path analytic
tests of the model. In this model, we argue (a)
that parental meta-emotion philosophy is re-
lated to both the inhibition of parental negative
affect and the facilitation of positive parenting;
(b) that it directly affects children's regulatory
physiology; and (c) that this, in turn, affects
children's ability to regulate their emotions—
hence, parental meta-emotion philosophy has an
impact on a variety of child outcomes.
Background
The Concept of Meta-Emotion
Research in developmental psychology on the
effects of parenting has focused on parental
affect and discipline, selecting variables such as
warmth, control, authoritarian or authoritative
styles,
and responsiveness (see Ainsworth, Bell,
& Stayton, 1971; Baumrind, 1967, 1971;
Becker, 1964; Cohn, Cowan, Cowan, & Pear-
son, 1992; C. P. Cowan & P. A. Cowan, 1992;
Maccoby & Martin, 1983; Patterson, 1982; and
Schaefer, 1959). Little attention has been placed
on examining the parents' feelings and cogni-
tions about their own affect or their feelings and
cognitions about their child's affect.
243
244 GOTTMAN, KATZ, AND HOOVEN
Our review of popular parenting guides also
revealed that the overwhelming majority of
these parenting guides are based on obtaining
and maintaining child discipline. However, one
genre of parenting guides focuses on children's
emotions and on how to make immediate and
everyday emotional connections with a child
that are not critical and contemptuous, but ac-
cepting. These kinds of parenting guides can be
traced to the seminal influence of one child
psychologist: Haim Ginott (Ginott, 1956, 1971,
1975).
Although many psychological systems
of thought (e.g., attachment theory, psychoanal-
ysis) have written about the importance of the
child's affect, Redl (1966) and Ginott both em-
phasized intervening with a child's strong neg-
ative emotions while the child is having the
emotions. They also emphasized intervening di-
rectly, dealing with the child's conscious
thoughts and actions. This difference was so
important, in our view, that it amounted to a
revolution in "how one deals with children," to
use Redl's (1966) words.
Our initial interest was in this concept of
parents' awareness of their children's emotional
lives and their attempts to make emotional con-
nections with their children. This interest led to
the development of our meta-emotion interview
(Katz & Gottman, 1986). All the parents were
separately interviewed about their own experi-
ence of sadness and anger, their philosophy
of emotional expression and control, and their
attitudes and behavior about their children's
anger and sadness. In our pilot work, we dis-
covered a great variety in the emotions, experi-
ences,
philosophies, and attitudes that parents
had about their own emotions and the emotions
of their children. For example, one pair of par-
ents said that they viewed anger as "from the
devil," and that they would not permit them-
selves or their children to express anger. Some
parents were accepting of sadness and anger but
did not engage in problem solving with their
child. Other parents were not disapproving of
anger but, instead, in laissez-fairre fashion, ig-
nored anger in their children. Still other parents
encouraged the expression and exploration of
anger. There was similar variation with respect
to sadness. Some parents minimized sadness in
themselves and in their children, saying such
things as, "I can't afford to be sad," or "What
does a child have to be sad about?" Other par-
ents thought that emotions like sadness in their
children were opportunities for intimacy, that
sadness was important information that some-
thing was missing in one's life.
This area of meta-emotion is probably char-
acterized by great variability even in laboratory
experiments that elicit emotions. Researchers
have reported large variability in results from
laboratory experiments designed to elicit emo-
tion, such as the startle response. Ekman,
Friesen, and Simons (1985) reported a consis-
tent set of responses across participants to being
startled, but there were huge individual differ-
ences in the emotional response to having been
startled, that is, in people's meta-emotions to
the startle; Levenson and Sutton (personal com-
munication, June 15, 1994) reported a similar
set of
results.
Meta-emotion may be a pervasive
and understudied dimension in emotion re-
search.
An Emotion-Coaching Meta-Emotion
Philosophy
In our pilot work, we noticed that there are
some parents who are aware of the emotions in
their lives (particularly the negative emotions),
who can talk about those emotions in a differ-
entiated manner, who are aware of these emo-
tions in their children, and who assist their
children with their emotions of anger and sad-
ness,
acting like an emotion coach. This is a
parental meta-emotion philosophy we call an
emotion-coaching philosophy. We found that an
emotion-coaching meta-emotion philosophy
had five components: parents (a) said that they
were aware of low intensity emotions in them-
selves and in their children; (b) viewed the
child's negative emotion as an opportunity for
intimacy or teaching; (c) validated their child's
emotion; (d) assisted the child in verbally label-
ing the child's emotions; and (e) problem solved
with the child, setting behavioral limits, and
discussing goals and strategies for dealing with
the situation that led to the negative emotion.
We hypothesized that these parents have a
greater ability than other parents to maneuver in
the world of emotions, that they are more com-
fortable with the world of emotions, and that
they are better able to regulate emotions. We
expected them to be more affectionate with their
children and less autocratic than other parents.
However, it was our observation that an
emotion-coaching meta-emotion philosophy
was different from parental warmth, and we
META-EMOTION
245
tested this notion
in the
current study. Very
concerned, generally positive
and
warm parents
can
be
oblivious
to the
world
of
emotion,
and an
emotion-coaching meta-emotion philosophy
is
something additional that these parents bring
to
their roles
as
parents. Perhaps warmth
and
limit
setting
are
correlated with these meta-emotion
variables,
but we
think that they
are not the
same dimensions
of
parenting.
In contrast,
we
found that
a
dismissing meta-
emotion philosophy
was one in
which parents
felt that
the
child's sadness
or
anger were
po-
tentially harmful
to the
child, that
it was the
parents'
job to
change these toxic negative emo-
tions
as
quickly
as
possible, that
the
child
needed
to
realize that these negative emotions
would
not
last
and
were
not
very important,
and
that
it
was
the
parent's job
to
convey
to the
child
a sense that
he or she
could ride
out
these
negative emotions without damage.
We
found
that emotion-dismissing families could
be sen-
sitive
to
their children's feelings
and
wanted
to
be helpful,
but
their approach
to
sadness,
for
example,
was to
ignore
or
deny
it as
much
as
possible. Often they perceived
a
child's strong
emotion
as a
demand that they
fix
everything
and make
it
better. These parents hoped that
the
dismissing strategy would make
the
emotion
go
away quickly. They often conveyed
a
sense that
the child's emotion
was
something they
may
have been forced
to
deal with,
but it was not
interesting
or
worthy
of
attention
in itself.
They
described sadness
as
something
to get
over, ride
out,
but
look beyond
and not
dwell
on.
They
often used distractions when their child
was sad
to move
the
child along,
and
they even used
comfort,
but
within specified time limits,
as if
they were impatient with
the
negative emotion
itself.
They preferred
a
happy child
and
often
found these negative states
in
their child quite
painful. They
did not
present
an
insightful
de-
scription
of
their child's emotional experience
and
did not
help
the
child with problem solving.
They
did not see the
emotion
as
beneficial
or as
any kind
of
opportunity, either
for
intimacy
or
for teaching. Many dismissing families
saw
their child's anger (without misbehavior)
as
enough cause
for
punishment
or a
Time
Out.
It
is
important
to
point
out
that
the
term
meta-emotion
is
being used
in its
broadest
sense. Metacommunication
is
communication
about communication,
and
metacognition
is
cognition about cognition. Meta-emotion
in the
narrow sense might refer
to
emotions about
emotion;
for
example,
we
might only
be
study-
ing
how
parents feel about getting angry
at
their
children (e.g., feel guilty about getting angry).
However,
we use the
term broadly
to
encom-
pass feelings
and
thoughts about emotion.
As
the examples just provided suggest,
the con-
struct being tapped involves parents' feelings
and thoughts about their
own and
their chil-
dren's emotions, their responses
to
their child's
emotions,
and
their reasoning about these
re-
sponses (i.e., what
the
parent
is
trying
to
teach
the child when responding
to the
child's anger).
This broader construct indexes
a
fundamental
attitude
or
approach
to
emotion.
For
some
peo-
ple,
emotions
are a
welcome
and
enriching part
of their lives; they believe,
in a
fundamental
way, that
it is OK to
have feelings. However,
for other people, emotions
are to be
avoided
and
minimized;
the
world
of
negative emotions
is
seen
as
dangerous
(see
Appendix).
Meta-Emotion
and
Parenting
From
a
theoretical standpoint,
we
think
our
measures
of
parental meta-emotion philosophy
are embedded within
a web of
measures that
tap
parent-child interaction.
We
expect that
par-
ents'
meta-emotion philosophy
is not
indepen-
dent
of
their parenting. Hence,
in our
theoretical
model,
we
include meta-emotion variables
along with parenting variables.
It is our
view
that
we
need
to be
specific about
our
description
of parenting;
for
heuristic purposes (within
the
restricted range
of
families
in our
samples),
we
discuss three dimensions
of
parenting behav-
iors.
First,
we
started with everyday, mundane
negativity. Inherent
in the
literature
on
parent-
ing
is the
idea that small things
in
everyday
parenting
can be
quite harmful
for
children
(or
serve
as
indexes
of
more harmful types
of par-
enting), akin
to
what
J.
Reid
has
called "natter-
ing"
(see
Patterson, 1982). Ginott (1965) wrote
strongly about this
in his
discussion
of the im-
portance
of
(a) understanding
and
validating
the
child's emotions
and (b)
avoiding contempt
and
disapproval. Thus,
in our
measurement
of
this
negativity
in our
laboratory-based parent-child
interactions,
we
included three variables: paren-
tal intrusiveness, criticism,
and
mockery.
We
call this type
of
parenting derogatory.
In a
teaching task,
as
some
of the
parents
in our
laboratory instructed their children, they mixed
246
GOTTMAN, KATZ, AND HOOVEN
in a blend of frustration, taking over for the
child as soon as the child had trouble with the
task (which we call intrusiveness), using criti-
cism and derisive humor (mockery, humiliation,
belittlement of the child). We think that this
dimension of parenting represents the microso-
cial processes characteristic of parental rejec-
tion (e.g., Whitbeck, Hoyt, Simons, & Conger,
1992).
Next, we also wished to measure two kinds of
positive parenting. The first is the kind of
warmth that Baumrind (1967, 1971) and others
described: We refer to this dimension of posi-
tive parenting as warmth. Following C. P.
Cowan and P. A. Cowan (1992), we include in
this dimension of warmth co-warmth, which
includes warmth between parents while inter-
acting with the child. Our second dimension of
positive parenting involves a positive structur-
ing, responsive, enthusiastic, engaged, and af-
fectionate parenting during the teaching task in
our laboratory. This type of positive parental
response goes beyond warmth. It includes the
responsive style that attachment theorists have
identified, but it is more complex than that. We
call it scaffolding-praising (on the general
scaf-
folding concept, see Choi, 1993; Kirchner,
1991;
Pratt, Kerig, Cowan, & Cowan, 1988; and
Vygotsky, 1987). From watching and coding
the videotapes, we noticed that this is a dimen-
sion quite different from Baumrind's authorita-
tive parenting. Parents high on the scaffolding-
praising dimension provided structure for the
task, stating the goals and procedures of the
game simply, in a relaxed manner, and with low
information density; they then waited for their
child to act and commented primarily when the
child did something right, acting like a cheering
section at a football game, giving praise and
approval. Parents low on this scaffolding-prais-
ing dimension either provided little structure for
the learning situation for their children, or they
gave information rapidly, with high density, and
enthusiastically, appearing to excite and con-
fuse the child; such parents then waited to com-
ment until their child had made a mistake.
These parents were then usually critical of the
child's performance.
Are the concepts of meta-emotion and emo-
tion coaching simply subdimensions of positive
parenting? We think not; we think that they add
to current concepts in the parenting literature
and are more general. Emotion coaching is one
reason why the parenting advice literature is, in
our view, far richer than the parenting research
literature. For example, what would we predict
that an authoritative parent would do (or rec-
ommend that he or she should do) when his or
her child has just had a nightmare? Being warm
and structuring provides no real guidelines.
Emotion coaching does provide these guide-
lines.
What is the expected relationship between
meta-emotion and the derogation, warmth, and
scaffolding-praising dimensions? When we be-
gan this study, we had two working hypotheses.
The first hypothesis was that a coaching meta-
emotion philosophy might be nested within a
web of positive parenting. We proposed that an
emotion-coaching meta-emotion philosophy en-
tails parenting that goes a step beyond the idea
of warmth; that is, we suggested that it entails
scaffolding-praising parenting. The second hy-
pothesis was that meta-emotion performs its
major function by inhibiting parental deroga-
tion; in particular, as Ginott (1965) noted, we
proposed that understanding and validating the
child's emotions serves to avoid criticism, con-
tempt, and disapproval of the child. Most of the
examples from Ginott's books had to do with
the importance of emotion coaching in avoiding
escalating negativity, frustration, disapproval,
and emotional distance between parents and
children. It appears to have been first suggested
foremost as a mechanism for obtaining exten-
sive relief from spiraling negativity: Perhaps
validating the child's affect serves as an oppo-
nent process to derogation. Hence, it was en-
tirely reasonable to hypothesize that the major
effect of a coaching meta-emotion philosophy
would be inhibiting parental negativity and that
it might have no effect on positive parenting.
Meta-Emotion and the Development of
Emotion Regulation Abilities
Precisely how do we think that meta-
emotions affect the functioning of families and
act to affect child outcomes? What do we pro-
pose as the mechanism? We are particularly
drawn to theories that attempt to integrate be-
havior and physiology, and our theorizing is
oriented toward approaches that have empha-
sized the importance of (a) the development of
children's abilities in the regulation of emotion
(Garber & Dodge, 1991) and (b) the develop-
ment of children's abilities to self-soothe strong
META-EMOTION
247
and potentially disruptive emotional states
(Dunn, 1977), focus attention, and organize
themselves for coordinated action in the service
of some goal. We think that these general sets of
abilities underlie the development of other com-
petencies. We are especially interested in chil-
dren's peer social skills, particularly because of
their predictive validity (Parker & Asher, 1987).
Central peer social competencies include the
ability to resolve conflict, to find a sustained
common ground play activity, to compromise in
play, and to empathize with a peer in distress
(e.g., see Asher & Coie, 1990; Gottman, 1983;
Gottman & Parker, 1986).
We suggest that fundamental to these abilities
is the ability to soothe one's self physiologically
and to focus attention. These abilities underlie
being able to listen to what one's playmate is
saying, being able to take another's role and
empathize, and being able to engage in social
problem solving. They involve the child know-
ing something about the world of emotion, both
his or her own and others'. We propose that this
knowledge arises only out of emotional connec-
tion being important in the home. In the follow-
ing section, we briefly review our reasons for
measuring child physiology as related to the
construct of emotion regulation.
Regulatory Physiology
1
We used Porges' (1984) suggestion that there
may be a physiological basis for the ability to
regulate emotion. To explain his notions, we
discuss two concepts related to the child's para-
sympathetic nervous system (PNS) physiology.
The major nerve of the PNS is called the vagus
nerve. The vagus nerve (so called because it is
the vagabond nerve that travels throughout the
body, innervating the viscera) is the X-th cranial
nerve. The tonic firing of the vagus nerve slows
down many physiological processes, such as the
heart rate. Research by Porges and his col-
leagues on the PNS has indicated a strong as-
sociation between high vagal tone and good
attentional abilities, and there is speculation that
these processes are related to emotion regula-
tion abilities. Porges (1992) reviewed evidence
that suggests that a child's baseline vagal tone is
related to the child's capacity to react to envi-
ronmental stimuli. There is a substantial body of
literature showing that basal vagal tone is re-
lated to both greater behavioral reactivity and
greater soothability; it is also related to greater
ability to focus attention and greater ability to
self-soothe and explore novel stimuli (DiPietro
& Porges,
1991;
Fox, 1989; Hofheimer & Law-
son, 1988; Huffman, Bryan, Pederson, &
Porges, 1988; Linnemeyer & Porges, 1986; Por-
ter, Porges, & Marshall, 1988; Richards, 1985,
1987;
Stifter & Fox, 1990; Stifter, Fox, &
Porges, 1989).
There is also another dimension of vagal tone
that needs to be considered, namely, the ability
to suppress vagal tone. In general, vagal tone is
suppressed during states that require focused or
sustained attention, mental effort, attention to
relevant information, emotional interaction, and
organized responses to stress. Thus, the child's
ability to perform a transitory suppression of
vagal tone in response to environmental (and
particularly emotional) demands is another in-
dex that needs to be added to the child regula-
tory physiology construct.
2
It relates to the like-
lihood of approach rather than withdrawal;
some infants with a high vagal tone who were
unable to suppress vagal tone in attention-
demanding tasks exhibited other regulatory dis-
orders (e.g., sleep disorders; Huffman, Bryan,
Pederson, & Porges, 1992; Porges, Walter,
Korb,
& Sprague, 1975). Porges, Doussard-
Roosevelt, and Portales (1992) found that
9-month-old infants who had lower baseline
vagal tone and less vagal tone suppression dur-
ing the Bailey examination had the greatest
behavioral problems at 3 years of age, as mea-
sured by the Achenbach and Edelbrock (1986)
Child Behavior Checklist. Measures of infant
1
We did not include the sympathetic portion of the
child's physiology in our modeling. However, we did
find that the child's concentration of adrenaline in the
24-hr urine sample at age 5 correlated (r = 0.39, p <
.001) with the child's illness at age 8.
2
We hypothesize that basal vagal tone is related to
the child's ability to sustain attention, whereas the
ability to suppress vagal tone is related to the child's
ability to shift attention when that is called for.
Porges and Doussard-Roosevelt (in press) pointed
out that one must be cautious about expecting the
suppression of vagal tone to always be the appropri-
ate vagal response to external demands. In the neo-
natal intensive care unit, the appropriate response to
gavage feeding turned out to be increases in vagal
tone,
consistent with the support of digestive pro-
cesses (DiPietro & Porges, 1991). Premature infants
who increased vagal tone during gavage feeding had
significantly shorter hospitalizations.
248
GOTTMAN, KATZ, AND HOOVEN
temperament derived from maternal reports
(Bates, 1980) were not related to the 3-year
outcome measures.
3
The Theoretical Challenge in Predicting
Peer Social Competence in Middle
Childhood From Emotion-Coaching
Interactions
A major goal of our research was to predict
peer social relations in middle childhood from
variables descriptive of the family's emotional
life in preschool. It is now well known that the
ability to interact successfully with peers and to
form lasting peer relationships are important
developmental tasks. Children who fail at these
tasks,
especially in the making of friends, are at
risk for a number of later problems (Parker &
Asher, 1987). The peer context presents new
opportunities and formidable challenges to chil-
dren. Interacting with peers provides opportuni-
ties to learn about more egalitarian relation-
ships,
to form friendships with agemates, to
negotiate conflicts, to engage in cooperative and
competitive activities, and to learn appropriate
limits for aggressive impulses. On the other
hand, children are typically less supportive than
caregivers when their peers fail at these tasks. In
our research, we have found that the quality of
the child's peer relationships forms an impor-
tant class of child outcome measures.
We should explain what the theoretical chal-
lenge is in predicting peer social relations across
these two major developmental periods, from
preschool to middle childhood. Major changes
occur in peer relations in middle childhood.
Children become aware of a much wider social
network than the dyad. In preschool, children
are rarely capable of sustaining play with more
than one other child (e.g., see Corsaro, 1979,
1981).
However, in middle childhood, children
become aware of peer norms for social accep-
tance, and teasing and avoiding embarrassment
suddenly emerge (see Gottman & Parker,
1986).
Children become aware of clique struc-
tures and of influence patterns as well as social
acceptance. The correlates of peer acceptance
and rejection change dramatically, particularly
with respect to the expression of emotion. One
of the most interesting changes is that the so-
cially competent response to a number of salient
social situations, such as peer entry and teasing,
is to be a good observer who is somewhat wary,
"cool," and emotionally unflappable (see Gott-
man & Parker, 1986). It is well known that the
worst thing a middle-childhood child can do
when entering a group of
peers
is to start talking
about his or her own feelings as parents and
children do in an emotion-coaching interaction.
Thus,
the basic elements and skills a child
learns through emotion coaching (labeling, ex-
pressing one's feelings, talking about one's
feelings, and drawing attention to one's self)
become liabilities in the peer social world in
middle childhood. Therefore, the basic model
linking emotion coaching in preschool to peer
relations in middle childhood cannot be a sim-
ple isomorphic transfer of social skills model.
Instead, it becomes necessary to identify a
mechanism that makes it possible for the child
to learn something in the preschool period that
underlies the development of appropriate social
skills across this major developmental shift in
what constitutes social competence with peers,
the development in the child of what Salovey &
Mayer (1990) and Goleman (1995) called
"emotional intelligence," a kind of "social
moxie." A number of researchers are addressing
concepts related to this idea of the child's de-
veloping social intelligence, such as the child's
developing ability to cope with stress (Saarni,
1993),
the child's emotional competence (Den-
ham, Renwick, & Hewes, 1994), the child's
developing empathy (Eisenberg & Fabes, 1990;
Eisenberg & Strayer, 1987), the child's devel-
oping social understanding (Denham, Zoller, &
Couchoud, 1994; Dunn & Brown, 1994), the
child's developing social and emotional compe-
tence and regulation as well as the child's de-
veloping theory of social mind (Casey & Fuller,
1994;
Fox, 1994; Harris & Kavanaugh, 1993;
Thompson, 1991), and the child's ability to
recognize emotional expressions (Cassidy,
Parke, & Butkovsky, 1992; Walden, 1988).
3
Time 1 and Time 2 down regulation scales cor-
related
0.41,
p = .003. The Time 1 down regulation
scale was unrelated to awareness and coaching, but it
correlated .31 (p = .020) with derogation, .44 (p =
.004) with Time 2 child negative affect, and 0.59
(p < .001) with Time 2 teacher ratings of negative
peer relations. It was unrelated to scaffolding-
praising parenting, to basal vagal tone, or to suppres-
sion of vagal tone.
META-EMOTION
249
The Theoretical Model: Hypotheses About
Meta-Emotion, Parenting, Regulatory
Physiology, and Child Outcomes
In building our theoretical model, we sought
to explain how meta-emotion philosophy might
be related to a variety of child outcomes. The
theoretical model is depicted in Figure 1. Given
the hypothesized effects of parental meta-
emotion philosophy on parenting skills, we ex-
pected that some effects would occur through
parenting practices. To be specific, we hypoth-
esized that the emotion-coaching meta-emotion
philosophy would be related to scaffolding-
praising and to the inhibition of parental dero-
gation. We also expected meta-emotion philos-
ophy, high scaffolding-praising, and low
derogation to be related to superior emotion
regulation abilities (as indexed by PNS func-
tioning and parental report). We also asked
whether effects between child physiology and
emotion coaching were bidirectional and
whether our effects varied with child tempera-
ment. One concern with these results was
whether parents were coaching their children
differentially as a function of the children's
temperament.
We also proposed that there would be a rela-
tionship between the parents' meta-emotion
coaching philosophy and a variety of child out-
comes. We hypothesized that a parental meta-
emotion coaching philosophy would be related
to the child's developing social competence
with other children. We expected that the
child's peer social competence would hold in
the inhibition of negative affect (Guralnick,
1981),
particularly negativity such as aggres-
sion, whining, oppositional behavior, fighting
requiring parental intervention, sadness, and
anxiety with peers. We also expected that an
emotion-coaching meta-emotion philosophy
would predict the development of superior cog-
nitive skills of the child (through superior vagal
tone and greater ability to focus attention). We
predicted that the relationship between the par-
ents'
meta-emotion philosophy and the child's
achievement at age 8 years would hold over and
above preschool measures of
intelligence.
Thus,
we predicted that two preschool children of
equal intelligence would differ, in part, in their
ultimate achievement in school as a function of
the parents' meta-emotion philosophy. Finally,
we examined the child's physical health as an
outcome variable. Because the vagus innervates
the thymus gland (Bulloch & Moore, 1981;
Bulloch & Pomerantz, 1984; Magni, Bruschi, &
Kasti, 1987; Nance, Hopkins, & Bieger, 1987),
a central part of the immune system that is
involved in the production of T-cells, we also
expected that basal vagal tone would be related
to better child physical health. The path models
also tested direct effects of meta-emotion,
which are theoretically unexplained by the
model.
We recognize that our correlational studies
could not provide us with a causal understand-
ing of the theoretical pathways we proposed.
However, we expected the correlational data to
yield results consistent with or disconfirming
these causal models, thereby suggesting some
directions for future research.
Direct Pathway
Child Outcomes
A
Child Emotion
Regulation Abilities
Figure I. Summary model for how parental meta-emotion might influence child outcomes.
250
GOTTMAN, KATZ, AND HOOVEN
Method
An abbreviated set of procedures is presented in
this article in the interests of conserving space. See
Gottman and Katz (1989) and Katz and Gottman
(1993) for more detail.
Participants
Fifty-six normal families were recruited from the
Champaign-Urbana, Illinois, community for this
study; 24 of the families had a male and 32 a female
4-
to 5-year-old child. We used a telephone version of
the Locke-Wallace marital satisfaction scale
(Krokoff,
1984) to ensure that our study included
couples with wide range of marital satisfaction levels.
The mean marital satisfaction score was 111.1 (SD =
29.6).
Three years later, we recontacted 53 of the 56
families (94.6% of the original sample).
Procedure
The procedure involved laboratory sessions and
home interviews for both parents and children. We
used a combination of naturalistic interaction, highly
structured tasks, and semistructured interviews.
Home and laboratory visits consisted of two home
visits-one with the marital couple and one with the
child—and three laboratory visits—one with the cou-
ple only, one with the couple and their 4 to 5-year-old
child, and one with the child alone. Families were
seen at Time 1, when children were 4-5 years old,
and again three years later at Time 2, when children
were on average 8 years old.
Time 1
Assessment
Meta-emotion interview. All parents were sepa-
rately interviewed about their own experience of sad-
ness and anger; their philosophy of emotional expres-
sion and control; and their feelings, attitudes, and
behavior about their children's anger and sadness
(Katz & Gottman, 1986). Their behavior during this
interview was audiotaped. A script for this semistruc-
tured interview is available from John M. Gottman or
Lynn Fainsilber Katz.
Parent- child interaction. The parent- child inter-
action session consisted of a modification of two
procedures used by P. A. Cowan and C. P. Cowan
(1987).
In the first task, parents were asked to obtain
information from their child. The parents were in-
formed that the child had heard a story, and they were
asked to find out what the story was. The story that
the children heard did not follow normal story gram-
mar and was read in a monotone voice; thus, the story
was only mildly interesting for the children and hard
to recall. The second task involved teaching the child
how to play an Atari game that the parents had
learned to play while the child was hearing the story.
The interaction lasted 10 min.
Children's film viewing. During a second visit to
the laboratory, children were shown segments of
emotion-eliciting films. Our main interest was to
obtain indexes of physiological activity during emo-
tional and nonemotional events. Each film clip was
preceded by a neutral story and an emotion induction
film clip of an actress who acted out the emotions of
the protagonist in the upcoming story. The function
of the emotion induction was to direct the child to
identify with the protagonist and experience the spe-
cific emotion in question. Although each film clip
was designed to elicit a specific emotion, the emotion
elicitation was not very successful; instead, we ob-
tained a range of facial expressions of emotion during
in each film. Hence, we refer to the films by their
titles instead of by the emotion they were intended to
induce. The child viewed clips from six films: (a) Fly
fishing; (b) Wizard ofOz (Leroy & Flemming, 1939),
flying monkey scene; (c) Charlotte's Web (Barbara,
Nicolas, & Takamoto, 1973), Charlotte dies; (d)
Meaning of Life (Goldstone & Gilliam, 1983), res-
taurant scene; (e) Wizard of Oz, taking Toto away;
and (f) Daisy (Alcroft & Mitton, 1984).
Child's physiological functioning. We assessed
the following physiological variables from the child
under baseline conditions, during parent-child inter-
action, and during film-viewing:
1.
Cardiac interbeat interval (IBI). This measure
was determined by measuring the time interval be-
tween successive spikes (R-waves) of the electrocar-
diogram (EKG).
2.
Pulse transmission time to the finger (PTT-F).
This was a measure of the elapsed time between the
R-wave of the EKG and the arrival of the pulse wave
at the finger.
3.
Finger pulse amplitude (FPA). This was an
estimate of the relative volume of blood reaching the
finger on each heart beat.
4.
Skin conductance level (SCL). This measure
was sensitive to changes in levels of sweat in the
eccrine sweat glands located in the hand.
5.
General somatic activity (ACT). To measure
somatic activity, the participant's chair was mounted
on a platform that was coupled to a rigid base in such
a way as to allow an imperceptible amount of flexing.
Child intelligence. The Wechsler Preschool
Scales of Intelligence (WPPSI; Wechsler, 1974)
Block Design, Picture Completion, and Information
subscales were administered to each child.
Measure and Coding
Meta-emotion coding system. The audiotapes of
the meta-emotion interview were coded using a spe-
META-EMOTION
251
cific checklist rating system that codes for parents'
awareness of their own anger and sadness, their own
regulation of anger and sadness, and their acceptance
and assistance (coaching) with their child's anger and
sadness (Hooven, 1994). For each dimension, the
coding manual was quite detailed and specific. The
Awareness score was a sum of 12 subscales: experi-
encing the emotion; being able to distinguish the
emotion from others; having various experiences
with the emotion; being descriptive of the experience
of the emotion; being descriptive of the physical
sensations connected with this emotion; being de-
scriptive of the cognitive processes connected with
this emotion; providing a descriptive anecdote;
knowing the causes of the emotion; being aware of
remediation processes; answering questions about the
emotion easily, without hesitation or confusion; talk-
ing at length about this emotion; and showing interest
or excitement about this emotion. Coaching was a
sum of 11 scales: showing respect for the child's
experience of the emotion, talking about the situation
and the emotion when the child is upset, intervening
in situations that give rise to the emotion, comforting
the child, teaching the child rules for appropriate
expression of the emotion, educating the child about
the nature of this emotion, teaching the child strate-
gies to soothe the child's own emotion, being in-
volved in the child's experience of the emotion, feel-
ing confident about how to deal with this emotion,
having given thought and energy to the emotion and
what one wants the child to know about this emotion
(goals), and using strategies that are age and situation
appropriate. Because coders used rating scales, the
appropriate statistics to use for computing interrater
reliability were correlations between independent ob-
servers for each scale, rather than Cohen's kappa,
which is appropriate for categorical data. The range
of interobserver reliabilities for the awareness and
coaching scales was 0.73 to 0.86.
Observational coding of parent-child interaction.
Parenting was coded using the Cowans' Observa-
tional System, the Kahen Engagement Coding Sys-
tem (KECS), and the Kahen Affect Coding Systems
(KACS;
Gottman, in press). The KECS consists of
seven parental engagement codes, including three
positive, three negative, and one neutral code. The
three Kahen positive engagement codes are as fol-
lows:
(a) Engaged, which consisted of parental atten-
tion toward the child; (b) Positive Directiveness, in
which parents issued a directive statement that began
in a positive way (e.g., "move to your right"); and (c)
Responds to Child's Needs, in which parents re-
sponded to a child's question or complaint. The three
negative engagement codes are as follows: (a) Dis-
engaged, in which parents were not attending to the
child; (b) Negative Directiveness, in which parents
issued a directive statement that began in a negative
way (e.g., "Don't move around so much"); and (c)
Intrusiveness, which involved physical interference
with the child's actions (e.g., grabbing the joy stick).
The KACS also consists of seven parental affect
codes.
The three positive affect codes are as follows:
(a) Affection, which consisted of praise and physical
affection; (b) Enthusiasm, which was coded as prais-
ing and excitement at the child's performance; and
(c) Humor, which involved parental laughter or jok-
ing. The three negative affect codes are as follows:
(a) Criticism, which involved direct disparaging
comments or put-downs of the child's behavior or
performance; (b) Anger, in which parents were visi-
bly frustrated by the child's actions or demonstrated
disappointment, annoyance, or irritation toward the
child; (c) Derisive Humor, in which parents used
humor at the child's expense (e.g., through sarcasm
or by making fun of the child).
Parent-child interaction was coded continuously in
real time with coding synchronized to the original
parent-child interaction. The total number of times
each variable occurred in the 10-min parent-child
interaction session was recorded and totals across
time were calculated for each of the 14 parent-child
interaction variables. This index was therefore an
estimate of the frequency of the parenting behavior
within a 10-min period. Independent observers coded
mothers and fathers. Engagement and affect dimen-
sions were also coded by independent observers.
Reliability was calculated across coders using a cor-
relation coefficient. Because total number of seconds
within each parent code was the variable computed
and used in all data analyses, the appropriate reliabil-
ity statistic was a correlation coefficient, rather than
Cohen's kappa or percentage agreement. For the
KECS,
the mean correlation was .96, with a range of
.86 to .99; for the KACS, the mean correlation was
.93,
with a range of .84 to .97. We computed the sum
of derisive humor, intrusiveness, and criticism for
both parents to form our derogation variable. The
Kahen systems were also used to measure the
scaffolding-praising dimension, which consisted of
parental affection, engagement, positive structuring,
responsiveness, and enthusiasm; we computed the
sum of these variables across parents. Although it is
certainly reasonable to expect to find differential
effects of mothers and fathers on children and we
have evidence that maternal and paternal parenting
were uncorrelated (the correlation between mother
and father derogation was .21 and was .12 between
mother and father scaffolding-praising), we summed
across parents* scores for the sake of economy. The
scaffolding-praising dimension differs from Baum-
rind's authoritative parenting in that it includes a
responsive and enthusiastic parenting style; in the
teaching task, this was reflected in parents' (a) effec-
tively structuring and scaffolding the child's learning
and (b) generally waiting until the child did some-
thing right and men praising enthusiastically, rather
252
GOTTMAN, KATZ, AND HOOVEN
than waiting until the child made a mistake and then
being critical.
The parents' behavior during the parent-child in-
teraction was also coded using P. A. Cowan and C. P.
Cowan's (1987) coding system. This coding system
codes parents behavior on dimensions of warmth-
coldness, presence or lack of structure and limit set-
ting, whether or not parents back down when their
child is noncompliant, anger and displeasure, unre-
sponsiveness or responsiveness, and whether or not
parents make maturity demands of their child. The
behavior of parents toward each other during their
interactions with their child (their coparenting) was
also coded on dimensions of warmth, cooperation or
competition, anger, disagreement, responsiveness,
pleasure in coparenting, clarity of communication,
and amount of interaction. For the purposes of this
study, only the warmth dimension (parenting and
coparenting) was of interest. For the parenting di-
mension, coders rated the overall degree of warmth
and the highest level of warmth and coldness exhib-
ited by each parent. For the coparenting dimension,
coders rated the overall degree of warmth and the
highest level of warmth and coldness exhibited by the
couple toward each other. Warmth was defined as the
sum of all the warmth variables minus the sum of all
the coldness variables. Interreliability for the warmth
variable was .64.
Child regulatory physiology. An estimate of the
child's baseline vagal tone was taken when the child
was listening to the introduction to an interesting
story taken from an animated cartoon film (clip from
Charlotte's Web), a variable we called basal vagal.
The child's ability to withdraw vagal tone was esti-
mated as a difference between this estimate of basal
vagal tone and the child's vagal tone during an ex-
citing film clip designed to elicit strong emotion (the
scene from The Wizard of Oz when the flying mon-
keys kidnap Dorothy). We expected vagal tone to be
withdrawn and heart rate to increase when the child
was emotionally engaged with the fearful stimuli in
this second film clip. We called this second variable
delta vagal. This second variable indexed the child's
ability to suppress vagal tone when engaging with a
strong emotional stimulus that included an environ-
mental demand for changing attentional focus or reg-
ulating emotion. In this case, the engagement with
the environment involved the demands for an emo-
tional response being elicited by the emotional film
as well as the demands to focus attention on the Atari
videogame the child played immediately after each
film clip. The index of vagal tone was computed as
the amount of variance in the cardiac interbeat inter-
val spectrum that was within the child's respiratory
range using spectral time-series analysis. This index
of vagal tone measures respiratory sinus arrhythmia,
a measure of PNS tonus, which has been found to
index attentional processes and emotion regulation
abilities (Porges, 1984). Mean levels of interbeat
interval at baseline, interbeat interval variability (a
measure of vagal tone; Izard, Porges, Simons, &
Haynes, 1991), reactivity of heart rate variability
from baseline to the mean of the parent-child inter-
action (used as an index of the child's ability to
modulate vagal tone by DiPietro, Porges, & Uhly,
1992),
and mean skin conductance level during base-
line (first visit to the lab) were also computed as
indexes of the amount of the child's chronic physio-
logical arousal and PNS functioning.
Time 2
Assessment
Overview. Time 2 assessment consisted of
teacher ratings of child outcomes and couple's re-
ports of considerations of marital dissolution. Fami-
lies were recontacted 3 years later for follow-up
assessments of child and marital outcomes. Children
were, on average, 8 years old (M = 96.9 months;
range = 82-110 months). Ninety-five percent (53 out
of 56) of the families in the initial sample and 86%
(48 out of 56) of the children's teachers at follow-up
agreed to participate in the Time 2 assessments.
Ratings of children's behavior problems.
Teachers completed the Child Adaptive Behavior
Inventory (CABI; P. A. Cowan & C. P. Cowan,
1990).
We used the CABI as a measure of child
outcomes for two reasons. First, the CABI was de-
veloped on a normal sample and contains subscales
that are less pathological in nature than the Child
Behavior Checklist (Achenbach & Edelbrock, 1986).
Second, the CABI also controls for teacher rating
bias by having teachers complete the scale on all
same-sexed children in the classroom and deriving z
scores for the target child. The CABI has good inter-
nal consistency (average alpha =
.81;
range = .66 to
.90) and predictive validity (P. A. Cowan & Cowan,
1990).
The factors and subscales that comprise the
CABI include the following: (a) the Antisocial Fac-
tor, which consists of the Hyperactivity, Antisocial
Behavior, Negative Engagement with Peers, Hostil-
ity, Fairness-Responsibility (keyed negatively),
Calm Response to Challenge (keyed negatively), and
Kindness-Empathy (keyed negatively) subscales;
and (b) the Internalizing Factor, which consists of the
Introversion, Depression, Victim-Rejected, Tension,
and Extroversion (keyed negatively) subscales.
Differential Emotions Scale. We used the Differ-
ential Emotions Scale (Izard, 1982) as a measure of
temperament, given Goldsmith and Campos' (1982)
definition of temperament in terms of affect expres-
sion (for a review and critique, see Bates, 1987). The
Differential Emotions Scale is a 36-item question-
naire that mothers complete with regard to the fre-
quency with which they have observed their children
display specific emotions in the past week. Each item
META-EMOTION
253
is rated on a
5-point
scale ranging from
1
(rarely
or
never)
to 5
(very often).
We
computed
the
total
number of emotions and the total number of positive
and negative emotions for the week.
Teacher ratings
of
peer
aggression.
Teachers
also completed
the
Dodge Peer Aggression Scale
(Dodge, 1986). The Dodge Peer Aggression scale
consists of items that measure the degree
to
which the
child uses overt aggression with peers.
Child academic
achievement.
Children were
in-
dividually administered
the
Peabody Individual
Achievement Test—Revised (PIAT-R) as a measure
of academic achievement (e.g., see Costenbader &
Adams, 1991). They were administered the mathe-
matics, reading recognition, reading comprehension,
and general information tests.
Emotion regulation
questionnaire.
Mothers filled
out
a
newly developed 45-item questionnaire about
the degree
to
which their child requires external
regulation of emotion (Katz & Gottman, 1986). This
questionnaire includes items that reflect instances
when the parent needs to down regulate the child to
reduce
the
child's level
of
activity, inappropriate
behavior, and misconduct. The alpha coefficient for
the scale was .74.
Child physical
health.
Child illness was assessed
by parental report using
a
version of the Rand Cor-
poration Health Insurance Study measures (available
from John M. Gottman or Lynn Fainsilber Katz; see
Gottman
&
Katz, 1989). The following Likert
or
true-false items were summed: "In general, would
you say that this child's health is excellent, good, fair
or poor?," "The child's health
is
excellent," "The
child seems to resist illness very well," "When some-
thing
is
going around this child usually catches it,"
"This child has had a nosebleed in the past 30 days"
(alpha
=
0.82).
Results
We begin
by
discussing
the
selection
and
validity
of
variables used for building the theo-
retical model. The reduced set of variables used
to index meta-emotion philosophy
are pre-
sented, and we address the construct validity of
the derogatory parenting
and
scaffolding-
praising variables by indicating that derogatory
parenting
is not
anger
and
that scaffolding-
praising parenting
is not
warmth.
We
then
present the results
of
our path-analytic model-
ing, looking separately
at
models related to pa-
rental derogation
and to
parental scaffolding-
praising.
We
also test
the
temperament
hypothesis
to see
whether parents select their
parenting style to be consistent with the child's
behavior. Finally,
we
review
the
result that
meta-emotion predicts child academic achieve-
ment
at
Time
2,
even controlling
for
Time
1
child intelligence.
Selection and Validity
of
Key
Theoretical Variables
Reduced Set of Meta-Emotion Variables
In the interest of parsimony, we needed to cut
down the choice
of
variables for the modeling,
and it was thus necessary to limit the number of
meta-emotion variables.
We
started with
12
variables (awareness
of
own emotion, aware-
ness
of
the child's emotion, and coaching,
for
father and mother and
for
sadness and anger).
Two variables from this
set of 12
were con-
structed, one
of
which was the sum
of
parental
awareness of the parents' own emotions and the
sum
of
the parents' awareness
of
the child's
emotions and the other of which was the sum of
the parental coaching
of
the child's emotions.
Table 1 gives
a
summary
of
the correlations of
these
two
summary meta-emotion variables
(i.e.,
awareness and coaching) with the original
12 meta-emotion variables that were obtained.
Therefore,
it
is a presentation of item-total cor-
relations that demonstrate the internal construct
validity
of
the summary codes. These correla-
tions show that the two variables we constructed
are related
to all
the individual awareness and
coaching variables.
Table
1
Correlations
of
the Two Summary Meta-
Emotion Variables With Original 12 Variables
Original meta-
emotion variables
Father sadness
Awareness own
Awareness child
Coaching
Father anger
Awareness own
Awareness child
Coaching
Mother sadness
Awareness own
Awareness child
Coaching
Mother anger
Awareness own
Awareness child
Coaching
*p<.05. **p<.0\
Awareness
.80***
.68***
.26*
.75***
.69***
44***
.56***
.66***
.48***
.57***
.64***
.37**
. ***
p
<
Coaching
.55***
.37**
.63***
.33**
49***
j
4
***
.32**
.63***
.72***
.29*
.36**
.66***
.001.
254 GOTTMAN, KATZ, AND HOOVEN
Validity of Child Regulatory Physiology
To examine the validity of the vagal tone
variables, we correlated baseline vagal tone and
suppression of vagal tone with other indexes of
autonomic arousal relating to parasympathetic
activation and general nervous system arousal.
As Table 2 shows, baseline vagal tone and
suppression of vagal tone were related to the
child's mean heart rate during the parent-child
interaction, to the child's lower resting heart rate
during the first visit to the laboratory when it was
a mildly stressful and novel situation, to lower
baseline skin conductance during the second
visit to the laboratory after initial adaptation, to
higher heart rate variability, and to lower heart rate
reactivity in the parent-child interaction.
Validity of Parenting Variables
Derogatory parenting is not anger. In de-
velopmental psychology today, there is a gen-
eral equation of anger and negativity (see Cum-
mings & Davies, 1994), which we consider
unfortunate. When the construct of negative af-
fect is broken into specific negative emotions,
there is evidence that specific negative emotions
serve different functions. In marital interaction,
Table 2
Validity of the Two Child Physiology
Variables Selected for Model Building
Variable
Mean heart rate
during parent-
child
interaction
Mean heart rate
during first
time in the
laboratory
Heart rate
reactivity in
parent-child
interaction
Heart rate
variability
Skin conductance
level after
adaptation to
laboratory
Base vagal
-.26*
-.28*
-.24*
.25*
-.37**
Delta vagal
-.40**
-.33**
-.30*
.39**
-.22
Note. Basal vagal = child's baseline vagal tone;
delta vagal = child's vagal tone when emotionally
engaged.
* p < .05.
**
p < .01.
anger is not predictive of marital dissolution,
whereas contempt is (Gottman, 1993, 1994;
Gottman &
Krokoff,
1989; Gottman & Leven-
son, 1992). Ginott (1965) distinguished anger
versus emotional interactions with the child that
communicated contempt for the child, global
versus specific criticism, and criticism versus
blaming or communications suggesting that the
child was incompetent. He suggested that the
latter communications are harmful to a child,
whereas anger is not and may even be healthy.
To deal with a tendency to overgeneralize and
conclude that anger is harmful for children, we
selected the parental derogation codes from the
Kahen Coding Systems, which were designed to
measure Ginott's cluster of negative parenting;
derogation is not the same as anger parents
express toward children. We tested this
specif-
ically by looking at the relationships between
parental anger and derogation. Parental anger
was uncorrelated with our measure of parental
derogation (for father anger, r = —.11; for
mother anger, r =
—
.06),
uncorrelated with the
meta-emotion codes, and uncorrelated with neg-
ative child outcomes.
Scaffolding-praising is not warmth. It is
important to emphasize that scaffolding-
praising is not merely a dimension of global
positivity, such as is tapped by the Cowan pa-
rental warmth variable. Correlations computed
between the warmth codes from the Cowan
system and the scaffolding-praising code indi-
cated that only maternal warmth was related to
our scaffolding-praising variable (r = .32, p <
.05).
Paternal warmth, maternal and paternal
coldness, and co-warmth all showed no relation
to scaffolding-praising.
Summary of the Seven Dimensions of the
Theoretical Model
Aside from the outcome variables, there are
seven dimensions of the theoretical model. The
two meta-emotion variables were derived from
the meta-emotion coding system used to code
the meta-emotion interview: coaching (a sum of
11 scales) and awareness (a sum of 12 sub-
scales).
The two parenting dimensions were ex-
tracted from the Kahen observational scales,
scaffolding-praising (sum of three scales of the
KECS—engagement, positive directiveness,
and responsiveness to the child's needs—and
two scales of positive affect of the KACS—
META-EMOTION
255
affection-praise and enthusiasm) and deroga-
tion (the intrusiveness code of negative engage-
ment on the KECS and criticism and derisive
humor of the KACS negative affect codes). The
regulatory physiology dimensions were basal
vagal tone (when the child was listening to the
introduction to an interesting story taken from
an animated cartoon film, Charlotte's Web) and
the child's ability to suppress vagal tone, which
was estimated as a difference between this es-
timate of basal vagal tone and the child's vagal
tone during an exciting film clip designed to
elicit strong emotion (the scene from the Wizard
of Oz when the flying monkeys kidnap Dor-
othy).
The emotion regulation dimension was
taken from a questionnaire mothers completed
(when the child was 8 years old) about the
degree to which their child required external
down regulation of emotion.
Building the Theoretical Model
The basic template for the theoretical model
is presented in Figure 2. There are eight con-
ceptual pathways consistent with our theoretical
formulation and a direct pathway from meta-
emotion to child outcome. First, we predicted
that there would be statistically significant path
coefficients from the meta-emotion variables to
the parenting variable (Path 1). This was a va-
lidity test to see if the variables derived from
our interview related to actual parenting behav-
ior. Second, we tested the significance of the
paths from meta-emotion and parenting to the
physiological variables (Paths 2 and 3). We
hypothesized that meta-emotion, operating (in
part) through parenting, would significantly af-
fect these physiological variables. That is, we
fundamentally believed that these physiological
variables are not entirely engraved in stone,
even if they are biological; rather, we believed
that these variables are malleable and shaped in
part by parents through their emotional interac-
tions with the child. These conceptual pathways
enable us to assess whether meta-emotion phi-
losophy and parenting in some way are related
to the child's regulatory physiology, or, con-
versely, if the child's regulatory physiology is
related to meta-emotion philosophy or parent-
ing. We also predicted that there would be sta-
tistically significant path coefficients connect-
ing the child physiology variables to the child
outcome variables (Path 4), suggesting that the
child physiology at age 5 years predicts child
outcome at age 8 years (Path 7). We also pre-
dicted the physiological variables would predict
the emotion regulation variable, and that the
regulation and the parenting variables would
relate to child outcome (Paths 5 and 8). We
predicted that parenting would have a direct
effect on child regulation (Path 6). There may
be direct effects between the meta-emotion vari-
ables and the child outcome variables, but, to the
extent that this is true, we have not completely
succeeded in our theory, because we will not have
had a mechanism to explain these effects.
In the interest of data reduction, we created
the following four child outcome variables:
1.
Child achievement was the sum of the
mathematics and reading comprehension
scores.
2.
Child's emotion regulation abilities was
assessed with the Down-Regulation scale of the
Katz-Gottman Emotional Regulation Question-
Delta Vagal
Physiology
Basal Vagal
Child Outcomes
Regulation
Figure 2. Outline of the general expected path structure of the theoretical model.
256 GOTTMAN, KATZ, AND HOOVEN
naire, assessed when the children were 8 years
old (called down regulation). Because this is
technically an outcome variable and not a pro-
cess variable (because it was measured when
the children were 8 years old), the emotion
regulation variable appears in every child out-
come model.
3.
Child's peer relations was the sum of three
teacher rating scales: the CABI Negative Peer
scale, the CABI Antisocial Scale, and the
Dodge Peer Aggression scale.
4.
Child illness was our health measure at
age 8.
We built two sets of theoretical models.
Given the hypotheses that meta-emotion philos-
ophy might either be related to the inhibition of
parental derogation or the facilitation of
scaffolding-praising parenting, we constructed
separate models to examine hypotheses related
to derogation and scaffolding-praising. All
modeling computations were performed using
Bentler's (1992) computer program, EQS.
Modeling With Derogatory Parenting
The results of our model building are pre-
sented in Figures 3, 4, and 5. The following are
our goodness-of-fit statistics: for the academic
achievement outcome variable, ^(14, N =
56) = 13.68, p = .474, Bentler-Bonett normed
fit index (BBN) = .981; for the peer relations
outcome variable, ^(13, N = 56) = 17.95,/? =
.159,
BBN = .986; for child illness, ^(12, N =
56) = 13.04, p = .366, BBN = .987. Hence, all
models fit the data. Multiple Rs for the outcome
variables varied: for child academic achieve-
ment, R = 0.41; for negative peer teacher rat-
ings,
R = 0.62; for child physical illness, R =
0.60. The figures present the path coefficients,
with z scores for each coefficient in parentheses.
As can be seen from these figures, the model
building using our theory was generally suc-
cessful. We were able to find linkages for the
major pathways we proposed between meta-
emotion and parenting, between meta-emotion
and the physiological variables, between parent-
ing and the physiological variables for child
peer relations, between physiology and emotion
regulation, between emotion regulation and
child outcome, and between parenting and child
outcome. For child illness, we suggested that
basal vagal tone should relate to less physical
illness, and it turned out that the significant
paths to child illness were parental emotion
coaching and basal vagal tone.
In all models, either awareness or coaching of
the child's emotions was negatively related to
the negative parenting variable, supporting the
hypothesis that awareness or coaching is an
inhibitor of parental derogation of the child. It
was interesting that in all models the child's
ability to suppress vagal tone at age 5 was a
significant predictor of the child's emotion reg-
ulation at age 8. The greater the child's ability
to suppress vagal tone at age 5, the less the
parents had to down regulate the child's nega-
tive affects, inappropriate behavior, and overex-
citement at age 8.
Modeling With Scaffolding-Praising
Parenting
We tested whether the same equations we had
used for negative parenting would fit when the
, Awareness
.73 (6.89)
Figure 3. Path model for child academic achievement with parental derogation. Numbers in
parentheses are z scores.
META-EMOTION 257
(Awareness)
.73 (6.68) ( Derogation
-.47
(-3.72)
.45 (3.65)
-.47
(-3.65)
.48 (4.20)
Child Peer
Relations
A
.31 (2.48)
>( Down Reguate )
v
.61 (5.53)
Figure 4. Path model for child peer relations (teacher ratings) with parental derogation. Numbers
in parentheses are z scores.
parental derogation variable was replaced with the
scaffolding-praising parenting variable. We ex-
pected that some of the path coefficients would
change, but we first tested whether or not a similar
model would fit the data. For the academic
achievement outcome variable, a very similar
model fit the data, ^(14, AT = 56) = 13.12,
p = .517, BBN = .986; for the peer relations
outcome variable, ^(15, N = 56) = 24.14,
p = .063, BBN = .978; for child illness, ^(15,
N = 56) =
18.82,
p = .222, BBN =
.981.
These
results are presented as Figures 6, 7, and 8.
Was scaffolding-praising related to the vari-
ables of the model? For child academic achieve-
ment, the scaffolding-praising variable was sig-
nificantly related to the outcome and to
coaching. For teacher ratings of peer interac-
tion, the scaffolding-praising variable was not
directly related to the outcome, but it was sig-
nificantly related to coaching. For child illness,
the scaffolding-praising variable was unrelated
to the outcome, but it was significantly related
to awareness.
Direction of Effects Between Emotion
Coaching and Child Regulatory
Physiology: Testing the Temperament
Hypothesis
Although our path models present data sup-
porting the notion that emotion coaching can
-.43 (-3.17)
Figure 5. Path model for child illness with parental derogation. Numbers in parentheses are z
scores.
258
GOTTMAN, KATZ, AND HOOVEN
Figure 6. Path model for child academic achievement with positive parenting in the model.
Numbers in parentheses are z scores. SCAF-PR = scaffolding-praising.
affect a child's vagal tone, it is possible that the
direction of effects may be reversed. That is, it
is quite possible that either (a) child physiology
is a part of child temperament, and parents may
select parenting style to be consistent with in-
dividual differences in child behavior or (b)
emotion coaching changes a child's vagal tone.
Perhaps parents are more likely to do emotion
coaching with a child higher in vagal tone, or
perhaps emotion coaching can affect a child's
basal vagal tone. Although we could not answer
causal questions in our path modeling, we did
conduct additional analyses to see if results
either were consistent with or disconfirmed the
temperament hypothesis.
We tested the hypothesis that child vagal tone
might affect emotion coaching by reversing the
arrow between these two variables. The models
fit just as well with the direction of effects
reversed. First, we considered the models with
derogatory parenting. For the child outcome of
negative peer relations,
;^(13,
N = 56) =
19.32,
p = .113, BBN = .985, and the path
coefficient from basal vagal tone to emotion
coaching was .37, z = 3.62. For the child out-
come of child achievement,
)^(14,
N = 56) =
11.28,
p = .664, BBN = .989, and the path
coefficient from basal vagal tone to emotion
coaching was .35, z = 3.74. For the child out-
come of child illness, ^(12, N = 56) = 13.36,
p = .344, BBN = .986, and the path coefficient
from basal vagal tone to emotion coaching was
.33,
z = 3.05. Next, we considered the models
with scaffolding-praising. For the child out-
come of academic achievement, ;^(14, N =
56) = 12.77, p = .545, BBN = .986, and the
path coefficient from basal vagal tone to emo-
tion coaching was .34, z = 3.35. For the child
42 (-3.35)
Delta
Vagal
) >( Down Reguate
Figure 7. Path model for child peer relations (teacher ratings) with positive parenting in the
model. Numbers in parentheses are z scores. SCAF-PR = scaffolding-praising.
META-EMOTION
259
A
.02 (.13)
-.53 H.02)
Child Illness
.43(3.18)
"A
Figure 8. Path model for child illness with positive parenting in the model. Numbers in paren-
theses are z scores. SCAF-PR = scaffolding-praising.
outcome of negative peer relations, ^(15, N =
56) = 19.82, /> = .179, BBN = .982, and the
path coefficient from basal vagal tone to emo-
tion coaching was .40, z = 3.69. For the child
outcome of child illness, ^(14, N = 56) =
19.77,
p = .138, BBN = .982, and the path
coefficient from basal vagal tone to emotion
coaching was .37, z = 3.64. Thus, our modeling
can rule out neither a direction for effects, nor
the possibility that the effects are bidirectional.
Given recent theorizing that vagal tone is a
temperament dimension (Porges, Doussard-
Roosevelt, Portales, & Suers, 1994), one con-
cern with these results concerns the direction of
effects; the question is whether parents are
coaching their children differentially as a func-
tion of the children's temperament. To further
test this hypothesis, we correlated coaching
with our temperament measures from the Dif-
ferential Emotions Scale. Coaching was uncor-
related with the amount of child negative affect
(r = .02, ns), the amount of child positive affect
(r = .20, ns), and the amount of child total
affect (r = .16, ns). We were also concerned
that the direct effects of coaching on child out-
comes might be qualified as a function of the
child's basal vagal tone. Coaching was signifi-
cantly directly correlated with only one of the
three outcomes, the child's Time 2 physical
illness (r = -.55, p < .001). The child's basal
vagal tone was also significantly correlated with
this outcome (r = .30, p < .05); however, the
partial correlation between coaching and the
child's physical illness, controlling basal vagal
tone,
remained significant (r = -.56, p <
.001).
Hence, it appears that the direct benefits
of coaching are unaffected by the child's basal
vagal tone.
Meta-Emotion Predicts Child Academic
Achievement at Time 2, Controlling for
Time 1 Child Intelligence
If, as we hypothesized, the meta-emotion
variables affect school achievement through
emotion regulation, we should expect that the
relationships between the meta-emotion vari-
ables and the child's achievement at age 8 will
hold, even controlling the child's Time 1 IQ (at
age 5). To test this hypothesis, we performed a
regression analysis forcing in the three IQ scales
(WPPSI Block Design, Picture Completion, and
Information Scale scores) before entering the
mother's awareness of her own sadness in pre-
dicting the child's math scores and, in a second
analysis, before entering the father's coaching
of the child's anger. The F tests for change were
computed as well as the partial correlations. For
the prediction of the child's mathematics scores
from the mother's awareness of her own sad-
ness,
F(4, 48) = 6.12, p < .05 (partial correla-
tion = 0.34). For the prediction of the child's
reading comprehension scores from the father's
coaching of the child's anger, F(4, 44) = 9.41,
p < .01 (partial correlation = 0.37). For the
prediction of the child's total achievement score
from both the mother's awareness of her own
sadness and the father's coaching of the child's
anger, the two variables were summed for the
260
GOTTMAN, KATZ, AND HOOVEN
analysis, F(4, 45) = 4.13, p < .05 (partial
correlation = 0.29).
Summary
Let us summarize the results of our modeling
in Figure 9 with respect to the eight conceptual
paths numbered in Figure 2. We recognize that
our models are probably not independent of one
another.
Conceptual Paths
Path 1: Meta-emotion and parenting. Meta-
emotion was related to both the inhibition of
parental derogation of the child and to the fa-
cilitation of scaffolding-praising parenting.
Path 2: Meta-emotion and child physiology.
For models including the derogation parenting
variable, coaching was directly and signifi-
cantly related in all the models (achievement,
ratings of child peer relations, and child health)
to the child's physiology. This linkage was
also evident in the models that included the
scaffolding-praising parenting variable for the
achievement and peer relations models (margin-
ally for the child health model, z =
1.80).
This
suggests the intriguing hypothesis that perhaps
parents can influence a child's physiology by
emotion coaching; however, we cannot differ-
entiate directionality of effects without an ex-
periment. As we posited, the child's physiology
was significantly related to parental emotion
regulation ratings, which, in turn, were related
to child outcomes in the models for derogatory
parenting and scaffolding-praising and the
child's negative child peer relations. There were
direct links between child physiology and child
outcome only for child illness. We note, in
passing, that in path analysis, a path coefficient
is a partial correlation; thus, for example, the
pathway between delta vagal and down regulate
partials out basal vagal tone. For this reason, we
speak of these pathways in terms of the regula-
tory physiology variables instead of in terms of
single variables.
Path 3: Parenting and child physiology.
This hypothesized pathway was not supported
by any of the models.
Path 4: Parenting and emotional down reg-
ulation. This pathway was not supported by
any of the models.
Path 5: Parenting and child outcome. This
pathway was supported in three of the six mod-
els:
for derogatory parenting, there was a direct
link with child academic achievement and child
peer relations, whereas for scaffolding-praising
parenting, there was a direct link only with
academic achievement.
Path 6: Child physiology and child outcome.
This linkage was supported only for child ill-
ness and only in the derogation model.
Path 7: Child physiology and later emotional
down regulation. This pathway, which was
from the 5-year-old suppression of vagal tone
variable to the 8-year-old emotional down reg-
ulation variable, was supported in all of the
models.
Path 8: Emotional down regulation and child
outcome. This linkage was supported in two of
the six models, for models predicting peer rela-
tions.
Delta Vagal
"— -
Physiology
,-
Basal Vagal
Figure
9.
Revision
of the
path model derived from
our
theory.
META-EMOTION
261
Unexpected Direct Pathway
In two models for derogatory parenting
(teacher ratings of peer relations and child ill-
ness) and in one model for scaffolding-praising
(child illness), there was a direct pathway from
either awareness or coaching of
the
child's emo-
tions and the child outcome. For one of these
models (derogatory parenting: child peer rela-
tions),
more awareness predicted more negative
ratings, opposite to what we might have pre-
dicted, whereas for child illness, more coaching
of the child's emotions was related to less child
illness, which we predicted. One possible ex-
planation for the unexpected linkage is that at
times parental awareness of emotion is not such
a good thing for children. Perhaps being attuned
to a child's negative emotion fosters its expres-
sion, and even teachers pick up on this when the
child is 8 years old and rate it negatively. As
part of this explanation, perhaps the arrow of
causation should be reversed, meaning that par-
ents have heightened awareness of their child's
negative emotions because the child is highly
expressive of them. With this possibility in
mind, we attempted to fit path models with the
arrow reversed for the three models for which
awareness was positively related to a negative
child outcome. The results of these post hoc
analyses were as follows. For the derogatory
parenting models, with the child peer relations
outcome, the new model did not fit the data,
^(13,
N = 56) = 48.60, p < .001. Hence, this
hypothesis is not consistent with our data.
4
The
revised figure (Figure 9) places the Time 1
meta-emotion variables as the exogenous, or
driving, variables in the model, having their
purported causal effects on parenting, which
predicts 8-year-old child outcome, and on child
physiology, which has its major effects on out-
come through the child's emotion regulation
abilities at age 8. These results are consistent
with the theory we proposed.
Discussion
The preliminary results of our investigation
of meta-emotion are encouraging. The data sug-
gest that meta-emotion is related to parenting
behavior and to child regulatory physiology;
that child regulatory physiology at age 5 pre-
dicts emotion down regulation ability at age 8;
and that parenting, meta-emotion, and the
child's regulatory physiology and behavior are
related to child outcomes. In general, aside from
two direct connections (from awareness to peer
relations and from coaching to child illness),
there are two pathways through the models that
predict child outcomes: one from the meta-
emotion variables through child physiology and
emotion regulation and one from the meta-
emotion variables through parenting.
The pathway from meta-emotion to the
child's regulatory physiology suggests the hy-
pothesis that coaching the child's emotions has
a soothing effect on the child that may change
some key aspects of the child's PNS. If this
were true, then the child's ability to self-soothe,
to regulate negative emotion, and to focus at-
tention may be a result of temperament and may
also be affected by the parents' meta-emotion
philosophy. However, the data also support an
interpretation that suggests that parents may be
4
Because the arrow for awareness has coaching
coming into it and, as a variable, a line from aware-
ness represents a partial correlation of awareness,
controlling for coaching, we wondered if the problem
arose because some families are high in awareness
but do not coach. We split the families who were
above the median in awareness into two groups:
Group 2 included those above the median (N = 18),
and Group 1 included those below the median (N =
8) in coaching. Thus, Group 1 represented 8 families
high in awareness but low in coaching, whereas
Group 2 represented 18 families high in both aware-
ness and coaching. Group 1 had children whose play
quality at age 5 was significantly lower than Group 2,
F(l,
24) =
4.29,
p < .05, Group
1
M =
14.13,
Group
2 M = 16.00. There were also marginal effects at age
5 in the observed negative peer interaction with a best
friend; in noncompliance, F(l, 24) = 3.05, p = .093,
Group 1 = 2.75, Group 2 = .83; and in crying, F(l,
24) = 3.74, p = .065, Group
1
=
.63,
Group 2 = .00.
Also,
when the children were 8 years old, teachers
used the Dodge Rating Scale and rated the children
on three scales of peer relations. The teachers rated
the children in Group 1 significantly lower in the
Dodge Social Skills scale, F(l, 25) =
7.51,
p < .05,
Group 1 = -3.80, Group 2 =
-0.21;
higher in the
Dodge Aggression Scale, F(l, 25) = 4.56, p < .05,
Group 1 = -3.29, Group 2 = -0.40; and lower in
the Dodge scale of Overall Peer Relations, F(l,
21) = 5.83, p < .05, Group 1 =
-2.21,
Group 2 =
0.21 than Group 2. These post hoc results support the
hypothesis that the negative linkage between aware-
ness and negative child outcomes may be attribut-
able,
in part, to those families who are aware of their
child's emotion and perhaps are even accepting of it,
but who do not emotion coach their children.
262 GOTTMAN, KATZ, AND HOOVEN
selecting emotion coaching as a good strategy
of parenting with children who have higher basal
vagal tone. However, parents are not differentially
coaching their children as a function of our mea-
sures of their temperaments. The truth probably
lies in a combination of these two directions of
influence, or perhaps the influence is bidirectional;
an experimental test of the effects of training par-
ents in emotion coaching on basal vagal tone
would help sort out these alternatives. We have
now developed and pilot tested a parent training
intervention in emotion coaching.
The social skills that are related to peer social
competence in middle childhood are not the
same as the skills that emotion-coaching parents
are building in their children. It has been well
established that calling attention to oneself and
one's feelings are precisely the opposite of what
socially competent children do in a variety of
critical peer situations in middle childhood (for
data on peer entry, see Putallaz & Gottman,
1981,
and Asher & Gottman, 1981; for data on
teasing, see Gottman & Parker, 1986, and Asher
& Coie, 1990). They do not express their emo-
tions.
On the contrary, they act cool and
unruf-
fled, as if they have had an "emotion-ectomy."
The point here is that children whose parents
were emotion coaches at age 5 are acting in a
way that teachers describe as socially competent
at age 8. Hence, they are skillful enough to
know what to do to be competent with peers at
age 8, and these social skills are not at all the
same as what they learned from their parents at
age 5. Apparently, emotion-coached children
have learned to be savvy about peer social sit-
uations, and they can do what is called for. They
may have a heightened sense of awareness of
their own emotions, a better ability to
self-
regulate their own upset, and a greater ability to
attend to the salient aspects of any challenging
peer situation. This sense of heightened aware-
ness may lead them to be more likely to know
what is called for and to do it. This interpreta-
tion of our results suggests that what children
learn from emotion coaching is not at all
modeling-specific social skills. It is far more
likely that what they have acquired are the tools
to learn how to learn in emotionally challenging
situations, even if that calls for inhibiting emo-
tional responding.
From a theoretic standpoint, it is interesting to
consider why and how parental coaching would
affect the child's regulatory physiology. On the
basis of the seminal work of Davidson and of
Fox (Davidson, 1984; Davidson, Ekman, Saron,
Senulis, & Friesen, 1990; Davidson & Fox,
1982;
Davidson, Schaffer, & Saron, 1985; Dav-
idson & Tomarken, 1990; Fox & Davidson,
1987,
1988), we can reason that talking about
negative emotions while having the emotions
might entail changing a right frontal-limbic-
autonomic experience of emotion into a left
frontal dominant experience in which the left
frontal lobe assumes control over the right fron-
tal lobe and its limbic and autonomic connec-
tions.
The positive affects, language, and anger
are lateralized in the left frontal lobe, and Dav-
idson called these left-lateralized emotions ap-
proach emotions. Anger is considered an ap-
proach emotion because it generally tends to
engage people with the world rather than lead-
ing to disengagement and withdrawal. This may
be,
in part, the reason that many clinical inter-
ventions suggest a healing process in which
depression changes into anger. The other nega-
tive affects, such as fear, disgust, and sadness,
are lateralized in the left frontal lobe; Davidson
called these withdrawal affects.
We posit that talking about negative with-
drawal emotions while having them entails a
greater sense of approach rather than with-
drawal, a greater sense of control of the nega-
tive emotions, and greater parasympathetic con-
trol of autonomic reactions. These hypotheses
suggest to us a series of studies using concom-
itant electroencephalographic (EEG) and auto-
nomic measurement while children are either
experiencing a negative emotion, being dis-
tracted from an emotional response they are
having (similar to what dismissing parents do),
or talking about the emotions while having
them (similar to what coaching parents do). The
conditions in these experiments would involve a
systematic dismantling of the emotion-coaching
parents' behaviors; this group would be com-
pared to appropriate controls for rival hypothe-
ses and to a dismantling of the counterpart of
the coaching parent, reflecting a meta-emotion
philosophy we call dismissing. These studies
are currently underway in our laboratory, but
we do not hesitate to point out that these hy-
potheses are highly speculative.
The results of this first study of meta-emotion
are encouraging, but replication and extension
to a more diverse sample of families is clearly
needed. We are currently conducting a replica-
tion and extension study. Our initial sample was
not racially or ethnically diverse, and it was
META-EMOTION
263
limited to families with two married parents.
Subsequent investigation needs to assess
whether these results generalize to various kinds
of single-parent families and to blended fami-
lies.
Replication is also important because of the
relatively small sample size of the present study
and its possible effects on structural equations
modeling (Loehlin, 1992). Naturalistic samples
of parents interacting with their children during
emotional moments are also needed to see how
emotion-coaching families actually talk to their
children during times of heightened emotion.
These samples should be obtained across vari-
ous developmental levels of both the children
and the parents. We also need to assess the
stability of parental meta-emotion philosophy
over time. Changes to the meta-emotion inter-
view are also needed, including an expansion to
include other emotions (particularly fear, guilt,
and shame) and the development of a measure
of child meta-emotion philosophy.
The second general pathway through our
models was from the meta-emotion variables
through parenting. What this adds to the parent-
ing literature is that perhaps parents' own basic
feelings and thoughts about their emotions are
strongly related to or underlie the way they
parent. The results of our modeling with longi-
tudinal correlational data suggest a series of
experiments that would further test the model.
Each arrow in the model suggests an interven-
tion for altering the variable at the head of the
arrow by manipulating the variable at the foot of
the arrow. Thus, parent training in meta-
emotion should affect two pathways: It should
alter both parenting and the child's regulatory
physiology, and it may also have a direct effect
on child outcome. We are now planning to
conduct this research.
References
Achenbach,
T. M, &
Edelbrock,
C.
(1986). Manual
for
the
Teacher's Report Form
and
Teacher Ver-
sion of the Child Behavior
Profile.
Burlington, VT:
University
of
Vermont, Department
of
Psychiatry.
Ainsworth,
M. D. S.,
Bell,
S. M., &
Stayton,
D. J.
(1971).
Individual differences
in the
strange situa-
tion behavior
of
one-year-olds.
In H. R.
Schaffer
(Ed.),
The
origins
of
human social relations
(pp.
17-57).
London: Academic Press.
Alcroft,
B.
(Producer),
&
Mitton,
D.
(Director).
(1984).
Daisy [film]. Strand Home Video.
Allen,
B. A., &
Armour,
T. E.
(1993). Construct
validation
of
metacognition. Journal
of
Psychol-
ogy, 127,
203-211.
Asher,
S. R., &
Coie,
J. D.
(Eds.). (1990). Peer
rejection in
childhood.
New York: Cambridge Uni-
versity Press.
Asher,
S. R., &
Gottman,
J. M.
(1981). The develop-
ment
of
children's friendships.
New
York:
Cam-
bridge University Press.
Barbara, H. (Producer), Nicolas, C.
A.,
& Takamoto,
I. (Directors). (1973). Charlotte's
web
[film].
Paramount.
Bates,
J. E.
(1980).
The
concept
of
difficult temper-
ament. Merrill-Palmer Quarterly,
26,
299-319.
Bates,
J. E.
(1987). Temperament
in
infancy.
In
J.
D.
Osofsky (Ed.), Handbook
of
infant development
(pp.
1101-1149).
New
York: Wiley.
Baumrind, D. (1967). Child care practices anteceding
3 patterns
of
preschool behavior. Genetic Psychol-
ogy Monographs, 75, 43-88.
Baumrind,
D.
(1971). Current patterns
of
parental
authority. Developmental Psychology Monograph,
4(1,
Pt. 2),
1-103.
Becker,
W. C.
(1964). Consequences
of
different
kinds of parental discipline. In M. L. Hoffman &
L.
W. Hoffman (Eds.), Review
of
child development
research (Vol.
1, pp.
169-208).
New
York:
Rus-
sell Sage Foundation.
Bentler,
P. M.
(1992).
EQS:
Structural equations
program manual.
Los
Angeles: BMDP Statistical
Software.
Bulloch,
K., &
Moore,
R. Y.
(1981). Innervation
of
the thymus gland
by
brain stem
and
spinal cord
in
mouse and rat. American Journal of Anatomy,
162,
157-66.
Bulloch
K., &
Pomerantz,
W.
(1984). Autonomic
nervous system innervation
of
thymic-related lym-
phoid tissue
in
wildtype and nude mice. Journal of
Comparative Neurology, 228, 57-68.
Bvinelli,
D. J.
(1993). Reconstructing
the
evolution
of mind. American Psychologist,
48,
493-509.
Casey,
R. J., &
Fuller,
L. J.
(1994). Maternal regu-
lation
of
children's emotions. Journal
of
Nonver-
bal Behavior,
18,
57-89.
Cassidy,
J.,
Parke,
R. D., &
Butkovsky,
C.
(1992).
Family correlates
of
childhood peer relations.
Child Development,
63,
603-618.
Choi,
S.
(1993). Mothers'
and
older siblings'
scaf-
folding
of
preschool children's pretend play.
Dis-
sertation Abstracts International, 5i(7-A), 2234-
2235.
Cohn, D. A., Cowan, P. A., Cowan, C. P., & Pearson,
J. (1992). Mothers' and fathers' working models
of
childhood attachment relationships, parenting
styles,
and child behavior. Development & Psycho-
pathology,
4,
417-431.
Corsaro, W. (1979). We're friends, right?: Children's
use
of
access rituals
in a
nursery school. Language
in Society,
8,
315-336.
Corsaro, W. (1981). Friendship
in
the nursery school:
264
GOTTMAN, KATZ, AND HOOVEN
Social organization in the peer environment. In S.
Asher & J. M. Gottman (Eds.), The development of
children's friendships. New York: Cambridge
University.
Costenbader, V. K., & Adams, J. W. (1991). A
review of the psychometric and administrative
features of the PIAT-R: Implications for the
practitioner. Journal of School Psychology, 29,
219-228.
Cowan, C. P., & Cowan, P. A. (1992). When parents
become partners. New York: Basic Books.
Cowan, P. A., & Cowan, C. P. (1987, April). Cou-
ple's relationships, parenting styles and the child's
development at three. Paper presented at the Soci-
ety for Research in Child Development, Baltimore,
MD.
Cowan, P. A., & Cowan, C. P. (1990). Becoming a
family: Research and intervention. In I. Sigel & A.
Brody (Eds.), Methods of family research (pp.
1-51).
Hillsdale, NJ: Erlbaum.
Cummings, E. M., & Davies, P. (1994). Children and
marital conflict: The impact of family dispute res-
olution. New York: Guilford Press.
Davidson, R. J. (1994). Temperament, affective style,
and frontal lobe asymmetry. In G. Dawson & K.
W. Fischer (Eds.), Human behavior and the devel-
oping brain (pp. 518-536). New York: Guilford
Press.
Davidson, R. J., Ekman, P., Saron, C. D., Senulis, J.
A., & Friesen, W. (1990). Approach-withdrawal
and cerebral asymmetry: Emotional expression and
brain physiology. Journal of Personality and So-
cial Psychology, 58,
330-341.
Davidson, R. J., & Fox, N. A. (1982). Asymmetrical
brain activity discriminates between positive and
negative affective stimuli in human infants. Sci-
ence, 218, 1235-1237.
Davidson, R. J., Schaffer, C. E., & Saron, C. (1985).
Effects of lateralized presentations of faces on
self-
reports of emotion and EEG asymmetry in de-
pressed and non-depressed subjects. Psychophysi-
ology, 22, 353-364.
Davidson, R. J., & Tomarken, A. J. (1990). Laterality
and emotion: An electrophysiological approach. In
F.
Boiler & J. Grafman (Eds.), Handbook of neu-
ropsychology (Vol 3, pp. 419-441). Amsterdam:
Elsevier.
Denham, S. A., Renwick, D. S., & Hewes, S. (1994).
Emotional communication between mothers and
preschoolers: Relations with emotional compe-
tence. Merrill Palmer Quarterly, 40, 488-508.
Denham, S. A., Zoller, D., & Couchoud, E. A.
(1994).
Socialization of preschoolers' emotional
and social competence. Developmental Psychol-
ogy, 30, 928-936.
DiPietro, J. A., & Porges, S. W. (1991). Relations
between neonatal states and
8-month
developmen-
tal outcome in preterm infants. Infant Behavior &
Development, 14, 441-450.
DiPietro, J. A., Porges, S. W., & Uhly, B. (1992).
Reactivity and developmental competence in pre-
term and full-term infants. Developmental Psy-
chology, 28,
831-841.
Dodge, K. A. (1986). A social information process-
ing model of social competence in children. In M.
Perlmutter (Ed.), Minnesota Symposium on Child
Psychology (Vol. 18, pp. 77-125). Hillsdale, NJ:
Erlbaum.
Dunn, J. (1977). Distress and comfort. Cambridge,
MA: Harvard University Press.
Dunn, J., & Brown, J. (1994). Affect expression in
the family, children's understanding of emotions,
and their interactions with others. Merrill Palmer
Quarterly, 40, 120-137.
Eisenberg, N. A., & Fabes, R. A. (1990). Empathy:
Conceptualization, measurement, and relation to
prosocial behavior. Motivation and Emotion, 14,
131-149.
Eisenberg, N. A., & Strayer, J. (Eds.). (1987). Em-
pathy and its development. New York: Cambridge
University Press.
Ekman, P., Friesen, W. V., & Simons, R. C. (1985).
Is the startle reaction an emotion? Journal of Per-
sonality and Social Psychology, 49, 1416-1426.
Flavell, J. H. (1979). Metacognition and cognitive
monitoring: A new area of cognitive-developmen-
tal inquiry. American Psychologist, 34,
906-911.
Fodor, J. A. (1992). A theory of the child's theory of
mind. Cognition, 44, 283-296.
Fox, N. A. (1989). The psychophysiological corre-
lates of emotional reactivity during the first year of
life.
Developmental Psychology, 25, 364-372.
Fox, N. A. (Ed.). (1994). The development of emo-
tion regulation. Monographs of
the
Society for Re-
search in Child Development,
59(2-3,
Serial No.
240).
Fox, N. A., & Davidson, R. J. (1987). Electroenceph-
alogram asymmetry in response to the approach of
a stranger and maternal separation in 10-month-
old infants. Developmental Psychology, 23, 233-
240.
Fox, N. A., & Davidson, R. J. (1988). Patterns of
brain electrical activity during facial signs of emo-
tion in 10-month-old infants. Developmental Psy-
chology, 24, 230-236.
Garber, J., & Dodge, K. A. (Eds.). (1991). The de-
velopment of emotion regulation and dysregula-
tion.
New York: Cambridge University Press.
Ginott, H. G. (1965). Between parent and
child.
New
York: Avon Books.
Ginott, H. G. (1971). Between parent and teenager.
New York: Avon Books.
Ginott, H. G. (1975). Teacher and
child.
New York:
Avon Books.
Goldsmith, H. H., & Campos, J. J. (1982). Toward a
theory of infant temperament. In R. N. Emde & R.
Harmon (Eds.), The development of attachment
META-EMOTION
265
and affiliative systems (pp. 161-193). New York:
Plenum.
Goldstone, J. (Producer), & Gillian, T. (Director).
(1983).
Meaning of
life
[film]. Universal.
Goleman, D. (1995). Emotional intelligence. New
York: Bantam Books.
Gottman, J. M. (1983). How children become friends.
Monographs of the Society for Research in Child
Development, 48(2, Serial No. 201).
Gottman, J. M. (1993). The roles of conflict engage-
ment, escalation, or avoidance in marital interac-
tion: A longitudinal view of five types of couples.
Journal of Consulting and Clinical Psychology, 61,
6-15.
Gottman, J. M. (1994). What predicts divorce?
Hillsdale, NJ: Erlbaum.
Gottman, J. M. (in press). Meta-emotion: The mea-
sures. Hillsdale, NJ: Erlbaum.
Gottman, J. M, & Katz, L. F. (1989). The effects of
marital discord on young children's peer interac-
tion and health. Developmental Psychology, 25,
373-381.
Gottman, J. M., &
Krokoff,
L. (1989). Marital inter-
action and satisfaction: A longitudinal view. Jour-
nal of Consulting and Clinical Psychology, 57,
47-52.
Gottman, J. M., & Levenson, R. W. (1992). Marital
processes predictive of later dissolution: Behavior,
physiology and health. Journal of Personality and
Social Psychology, 63, 221-233.
Gottman, J. M., & Parker, J. (Eds.). (1986). Conver-
sations of friends. New York: Cambridge Univer-
sity Press.
Guralnick, M. J. (1981). Peer influences on the de-
velopment of communciative competence. In P.
Strain (Ed.), The utilization of classroom peers as
behavior change agents (pp. 31-68). New York:
Plenum.
Harris, P. L., & Kavanaugh, R. D. (1993). Young
children's understanding of pretense. Monographs
of the Society for Research in Child Development,
58(1,
Serial No. 231).
Hofheimer, J. A., & Lawson, (1988). Neurophysio-
logical correlates of interactive behavior in preterm
newborns. Infant Behavior and Development, 11,
143.
Hooven, C. (1994). The meta-emotion coding system.
Unpublished manuscript, University of Washing-
ton, Seattle.
Huffman, L. C, Bryan, Y. E., Pederson, F. A., &
Porges, S. W. (1988). Infant temperament: Rela-
tionships with heart rate variablility. Unpublished
manuscript, National Institute of Mental Health,
Rockville, MD.
Huffman, L. C, Bryan, Y. E., Pederson, F. A., &
Porges, S. W. (1992). Autonomic correlates of
re-
activity and self-regulation at twelve weeks of age.
Unpublished manuscript, National Institute of
Mental Health, Rockville, MD.
Izard, C. (1982). Measuring emotions in human de-
velopment. In C. Izard (Ed.), Measuring emotions
in infants and children (pp. 3-20). New York:
Cambridge University Press.
Izard, C. E., Porges, S. W., Simons, R. F., & Haynes,
O. M. (1991). Infant cardiac activity: Developmen-
tal changes and relations with attachment. Devel-
opmental
P)tychology,
27, 432-439.
Katz, L. F., &\Gottman, J. M. (1986). The meta-
emotion interview. Unpublished manual, Univer-
sity of Washington, Department of Psychology.
Katz, L. F., & Gottman, J. M. (1993). Patterns of
marital conflict predict children's internalizing and
externalizing behaviors. Developmental Psychol-
ogy, 29, 940-950.
Kirchner, D. M. (1991). Using verbal scaffolding to
facilitate conversational participation and language
acquisition in children with pervasive developmen-
tal disorders. Journal of Childhood Communica-
tion Disorders, 14, 81-98.
Krokoff,
L. (1984). Anatomy of negative affect in
working class marriages. Dissertation Abstracts
International, 45, 7A. (University Microfilms No.
84-22109)
Leroy, M. (Producer), & Flemming, V. (Director).
(1939).
Wizard of Oz [film]. MGM.
Linnemeyer, S. A., & Porges, S. W. (1986). Recog-
nition memory and cardiac vagal tone in 6-month-
old infants. Infant Behavior and Development, 9,
43-56.
Loehlin, J. C. (1992). Latent variable models: An
introduction to factor, path, and structural analy-
sis.
Hillsdale, NJ: Erlbaum.
Maccoby, E. E., & Martin, J. A. (1983). Socialization
in the context of the family: Parent-child interac-
tion. In E. M. Hetherington (Ed.), Handbook of
child psychology, Vol. 4, Socialization, personality
(pp.
1-102).
New York: Wiley.
Magni, F., Bruschi, F., & Kasti, M. (1987). The
afferent innervation of the thymus gland in the rat.
Brain Research, 424, 379-385.
Nance, D. M., Hopkins, D. A., & Bieger, D. (1987).
Re-investigation of the innervation of the thymus
gland in mice and rats. Brain, Behavior, and Im-
munity, 1,
\JiA-41.
Olson, D. R., & Astington, J. W. (1993). Thinking
about thinking: Learning how to take statements
and hold beliefs. Educational Psychologist, 28,
7-23.
Parker, J. G., & Asher, S. R. (1987). Peer relations
and later personal adjustment: Are low-accepted
children at risk? Psychological Bulletin, 102, 357-
389.
Patterson, G. R. (1982). Coercive family process.
Eugene, OR: Castalia.
Porges, S. W. (1984). Heart rate oscillation: An index
of neural mediation. In M. G. H. Coles, J. R.
Jennings, & J. A. Stern (Eds.), Psychophysiologi-
cal perspectives: Festschrift for Beatrice and John
266
GOTTMAN, KATZ, AND HOOVEN
Lacey (pp. 229-241). New York: Van Nostrand
Reinhold.
Porges, S. W. (1992). Autonomic regulation and at-
tention. In B. A. Campbell, H. Hayne, & R. Rich-
ardson (Eds.), Attention and information process-
ing in infants and adults (pp. 201-223). Hillsdale,
NJ: Erlbaum.
Porges, S. W., & Doussard-Roosevelt, J. A. (in
press).
The psychophysiology of temperament. In
J. D. Noshpitz (Ed.), Handbook of child and ado-
lescent psychology. New York: Wiley Press.
Porges, S. W., Doussard-Roosevelt, J. A., & Portales,
L. A. (1992). Difficultness and cardiac vagal tone:
Stable indices of temperament in infants and tod-
dlers. Unpublished manuscript.
Porges, S. W., Doussard-Roosevelt, J. A., Portales, L.
A., & Suers, P. E. (1994). Cardiac vagal tone:
Stability and relation to difficulties in infants and 3
year olds. Developmental Psychobiology, 27, 289-
300.
Porges, S. W., Walter, G. F., Korb, R. J., & Sprague,
R. L. (1975). The influence of methylphenidate on
heart rate and behavioral measures of attention in
hyperactive children. Child Development, 46, 727-
733.
Porter, F. L., Porges, S. W., & Marshall, R. E. (1988).
Newborn pain cries and vagal tone: Parallel
changes in response to circumcision. Child Devel-
opment, 59, 495-505.
Pratt M. W., Kerig, P., Cowan, P. A., & Cowan, C. P.
(1988).
Mothers and fathers teaching 3-year-olds:
Authoritative parenting and adult scaffolding of
young children's learning. Developmental Psy-
chology, 24, 832-839.
Putallaz, M., & Gottman, J. M. (1981). An interac-
tional model of children's entry into peer groups.
Child Development, 52, 986-994.
Redl, F. (1966). When we deal with children. New
York: Free Press.
Richards, J. E. (1985). The development of sustained
visual attention in infants from 14 to 26 weeks of
age.
Psychophysiology, 22, 409-416.
Richards, J. E. (1987). Infant visual sustained atten-
tion and respiratory sinus arrhythmia. Child Devel-
opment, 58, 488-496.
Saarni, C. (1993). The socialization of
emotion.
In M.
Lewis & J. M. Haviland (Eds.), Handbook of
emo-
tions (pp. 435-446). New York: Guilford Press.
Salovey, P., & Mayer, J. D. (1990). Emotional intel-
ligence. Imagination, Cognition and Personality,
9,
185-211.
Schaefer, E. S. (1959). A circumplex model for ma-
ternal behavior. Journal of Abnormal and Social
Psychology, 59, 226-235.
Stifter, C. A., & Fox, N. A. (1990). Infant reactivity:
Physiological correlates of newborn and
5-month
temperament. Developmental Psychology, 26,
582-588.
Stifter, C. A., Fox, N. A., & Porges, S. W. (1989).
Facial expressivity and vagal tone in 5- and 10-
month-old infants. Infant Behavior and Develop-
ment, 12, 127-137.
Thompson, R. A. (1991). Emotional regulation and
emotional behavior. Educational Psychology Re-
view, 3, 269-307.
Vygotsky, L. S. (1987). The collected works of
L.
S.
Vygotsky (R. W. Rieber & A. S. Corton, Eds.).
New York: Plenum.
Walden, T. A. (1988). A developmental approach to
social behavior. Journal of Social Behavior and
Personality, 3, 67-84.
Wechsler, D. (1974). Selected papers of David
Wechsler. New York: Academic Press.
Whitbeck, L. B., Hoyt, D. R., Simons, R. L., &
Conger, R. D. (1992). Intergenerational continuity
of parental rejection and depressed affect. Journal
of Personality and Social Psychology, 63, 1036-
1045.
META-EMOTION 267
Appendix
Coaching and Awareness Variables
We created two variables from our coding of the
meta-emotion interview: coaching and awareness. To
give readers a qualitative sense of the emotional
nature of these two variables, we present a few ex-
amples of quotes from parents who were high and
low on these two variables. Parents high on coaching
said things like the following: "I feel close to my
child when he is sad," "When Jason is sad it makes
me feel like a real Dad, now my heart just goes out to
him," "When my child is sad I let her know that I
understand," "When Markie is sad I want to know
what he is thinking," "It makes me want to hold him
close," "I feel affectionate," "I feel love even when
he's upset and angry at me."
Some examples of emotion metaphors and con-
cepts parents expressed were as follows: "Anger
gives me energy and drive," " I think that sadness can
be good and even productive," "Sadness tells you to
slow down," "When I am sad I know something is
missing from my life," "Anger is just like clearing
your throat, just clear it and go on," "In my view,
attending to sadness is cleansing," "I want her to be
sad like in movies. It means she can feel and empa-
thize," "I often have the good cry, and I think she
does that too," "Getting angry can be a
relief,
like a
storm that finally happens."
Parents low on coaching included parents who
were either disapproving of the child's emotion or
dismissing of the child's emotion. Examples of com-
ments they made include the following: "Seeing my
child sad makes me uncomfortable," "I think that
sadness is OK as long as it's under control," "A
child's anger deserves a time out," "I get annoyed
when my child acts sad," "Children often act sad just
to get their own way," "She looks cute and silly when
she's angry, like a little midget," "I warn him about
not developing a bad character," "Her shouting scares
me,"
"Molly gets into these black moods."
One example of an emotion metaphor expressed
by a parent low on coaching was as follows: "When
people get angry they are just relieving themselves on
others." Other disapproving examples had to do with
loss of control; humiliation in public; metaphors of
fire,
pressure, heat, and other explosion and violence
metaphors for anger; and generally defeat, hopeless-
ness,
and pathology concerns about sadness.
For the awareness variable, we found that only
people who are aware of emotion and can differen-
tially talk about nuances of emotion and emotion
intensity find emotional expression to be acceptable.
If the low awareness reaction to negative emotions
were to be characterized as a specific emotion paired
to anger and sadness, it would be fear. People low in
awareness and coaching saw these emotions as toxic
and dangerous.
People high in awareness tended to believe that
one should not stifle one's emotions; that it was good,
healthy, and positive to pay attention to emotion; and
that emotions are always there, a part of life, and it is
best to be aware of them. They said things like it is
best to get anger or sadness "out of your system" by
becoming aware of it and then being able to deal with
it. They believed that it was important to recognize
smaller and less intense expressions of emotion to
prevent them from escalating. They could speak in a
differentiated manner about each emotion and the
bodily sensations of each; for example, some parents
talked about the "delicious" aspects of sadness in
some romantic movies but the awful grief that ac-
companies an important loss. These parents often
described the physical sensations that accompanied
an emotion, for example, "Sometimes I get so mad
that my stomach is in knots."
People low in awareness said that these negative
effects of anger and sadness were often so aversive
for them that they tended to prefer to minimize their
importance or not to notice them at all so they
wouldn't have to deal with them. That is what we
tapped with this variable and is probably the essence
of being low on the awareness variable. For example,
one parent said, "When he gets on my nerves like
that,
I
just tune him out," and, "He's not sad much. It
hurts me to see him sad though. I have to go out for
a run." Many parents who expressed discomfort with
their child's negative affect tended to view it as toxic
and believed that it was their role to get the child out
of the negative mood as quickly as possible. Many of
these parents also said that they and their children
rarely showed the emotion, or (to prove that they
could survive the negative mood) that they can "roll
with the punches." They seemed to be at a loss when
asked to describe how they could tell when they or
their children had the emotion, and they seemed
unaware of what might make a child feel sad or angry
and what might be done about it. They often ex-
pressed the philosophy that the way to cope with
negativity is to emphasize the positive aspects of life
and to substitute a positive emotion for the negative
emotion. They said that negative emotion must sim-
ply be endured; that the passage of time alone will
solve emotional problems; that to get over a negative
emotion one should just get on with life's routines,
ignore the emotion and just go on; that anger or
sadness meant loss of control; that feelings are pri-
vate,
not public; and that it is embarrassing to be sad
or angry and better not to be aware of it. These
(Appendix continues on next page)
268 GOTTMAN, KATZ, AND HOOVEN
Appendix (continued)
parents were often most aware of the demand com- and sadness, even if their names, awareness and
ponent of a child's emotion (that they fix the world coaching, do not sound very emotional,
and make it perfect so that their child will not have
this awful emotion). Received January 22, 1995
To summarize, these two variables represent Revision received December 14, 1995
highly emotional reactions and metaphors to anger Accepted December 15, 1995 •
New Editors Appointed, 1998-2003
The Publications and Communications Board of the American Psychological Association
announces the appointment of five new editors for
6-year
terms beginning in 1998.
As of January
1,1997,
manuscripts should be directed as follows:
• For the Journal of Experimental Psychology: Animal Behavior Processes,
submit manuscripts to Mark E. Bouton, PhD, Department of Psychology,
University of Vermont, Burlington, VT 05405-0134.
• For the Journal of Family
Psychology,
submit manuscripts to Ross
D.
Parke,
PhD,
Department of Psychology and Center for Family Studies-075,1419
Life Sciences, University of California, Riverside, CA 92521-0426.
• For the Personality Processes and Individual Differences section of the
Journal of Personality and Social Psychology, submit manuscripts to Ed
Diener, PhD, Department of Psychology, University of Illinois, 603 East
Daniel, Champaign, IL 61820.
• For Psychological Assessment, submit manuscripts to Stephen N. Haynes,
PhD,
Department of Psychology, University of Hawaii, 2430 Campus Road,
Honolulu, HI 96822.
• For Psychology and Aging, submit manuscripts
to
Leah
L.
Light,
PhD,
Pitzer
College, 1050 North Mills Avenue, Claremont, CA 91711-6110.
Manuscript submission patterns make the precise date of completion of the 1997 volumes
uncertain. Current editors, Stewart H. Hulse, PhD; Ronald F. Levant, EdD; Russell G.
Geen, PhD; James N. Butcher, PhD; and Timothy A. Salthouse, PhD, respectively, will
receive and consider manuscripts until December 31, 1996. Should 1997 volumes be
completed before that date, manuscripts will be redirected to the new editors for consider-
ation in 1998 volumes.
