ArticlePDF Available

Parental Meta-Emotion Philosophy and the Emotional Life of Families: Theoretical Models and Preliminary Data


Abstract and Figures

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. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Content may be subject to copyright.
Journal of Family Psychology
Vol. 10, No. 3, 243-268Copyright 1996 by the American Psychological Association, Inc.
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-
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.
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
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.
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,
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-
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
Meta-emotion may be a pervasive
and understudied dimension in emotion re-
An Emotion-Coaching Meta-Emotion
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-
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
tested this notion
in the
current study. Very
concerned, generally positive
warm parents
to the
and an
emotion-coaching meta-emotion philosophy
something additional that these parents bring
their roles
parents. Perhaps warmth
correlated with these meta-emotion
but we
think that they
are not the
same dimensions
In contrast,
found that
dismissing meta-
emotion philosophy
was one in
which parents
felt that
child's sadness
anger were
tentially harmful
to the
child, that
it was the
job to
change these toxic negative emo-
possible, that
realize that these negative emotions
very important,
parent's job
to the
a sense that
he or she
could ride
negative emotions without damage.
that emotion-dismissing families could
be sen-
their children's feelings
be helpful,
their approach
was to
it as
possible. Often they perceived
child's strong
as a
demand that they
and make
better. These parents hoped that
dismissing strategy would make
away quickly. They often conveyed
sense that
the child's emotion
something they
have been forced
deal with,
but it was not
in itself.
described sadness
to get
over, ride
look beyond
and not
often used distractions when their child
was sad
to move
child along,
they even used
within specified time limits,
as if
they were impatient with
negative emotion
They preferred
happy child
found these negative states
their child quite
painful. They
did not
their child's emotional experience
did not
child with problem solving.
did not see the
or as
any kind
opportunity, either
for teaching. Many dismissing families
their child's anger (without misbehavior)
enough cause
or a
being used
in its
sense. Metacommunication
about communication,
cognition about cognition. Meta-emotion
in the
narrow sense might refer
emotions about
might only
parents feel about getting angry
children (e.g., feel guilty about getting angry).
we use the
term broadly
pass feelings
thoughts about emotion.
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
their child's
their reasoning about these
sponses (i.e., what
the child when responding
to the
child's anger).
This broader construct indexes
are a
enriching part
of their lives; they believe,
in a
way, that
it is OK to
have feelings. However,
for other people, emotions
are to be
negative emotions
theoretical standpoint,
parental meta-emotion philosophy
are embedded within
a web of
measures that
parent-child interaction.
expect that
meta-emotion philosophy
is not
their parenting. Hence,
in our
include meta-emotion variables
along with parenting variables.
It is our
to be
specific about
of parenting;
heuristic purposes (within
restricted range
in our
discuss three dimensions
parenting behav-
started with everyday, mundane
negativity. Inherent
in the
is the
idea that small things
can be
quite harmful
more harmful types
of par-
enting), akin
called "natter-
Patterson, 1982). Ginott (1965) wrote
strongly about this
in his
of the im-
(a) understanding
child's emotions
and (b)
avoiding contempt
disapproval. Thus,
in our
in our
laboratory-based parent-child
included three variables: paren-
tal intrusiveness, criticism,
call this type
parenting derogatory.
In a
teaching task,
of the
in our
laboratory instructed their children, they mixed
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,
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
folding concept, see Choi, 1993; Kirchner,
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-
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
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 Physiology1
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,
Fox, 1989; Hofheimer & Law-
son, 1988; Huffman, Bryan, Pederson, &
Porges, 1988; Linnemeyer & Porges, 1986; Por-
ter, Porges, & Marshall, 1988; Richards, 1985,
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,
& 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
consistent with the support of digestive pro-
cesses (DiPietro & Porges, 1991). Premature infants
who increased vagal tone during gavage feeding had
significantly shorter hospitalizations.
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
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
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-
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,
However, in middle childhood, children
become aware of peer norms for social accep-
tance, and teasing and avoiding embarrassment
suddenly emerge (see Gottman & Parker,
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
is to start talking
about his or her own feelings as parents and
children do in an emotion-coaching interaction.
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,
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,
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-
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.
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
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,
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-
meta-emotion philosophy and the child's
achievement at age 8 years would hold over and
above preschool measures of
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
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
Child Emotion
Regulation Abilities
Figure I. Summary model for how parental meta-emotion might influence child outcomes.
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.
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
to 5-year-old child. We used a telephone version of
the Locke-Wallace marital satisfaction scale
1984) to ensure that our study included
couples with wide range of marital satisfaction levels.
The mean marital satisfaction score was 111.1 (SD =
Three years later, we recontacted 53 of the 56
families (94.6% of the original sample).
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
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
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:
Cardiac interbeat interval (IBI). This measure
was determined by measuring the time interval be-
tween successive spikes (R-waves) of the electrocar-
diogram (EKG).
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.
Finger pulse amplitude (FPA). This was an
estimate of the relative volume of blood reaching the
finger on each heart beat.
Skin conductance level (SCL). This measure
was sensitive to changes in levels of sweat in the
eccrine sweat glands located in the hand.
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-
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
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-
(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
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
the mean correlation was .96, with a range of
.86 to .99; for the KACS, the mean correlation was
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
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