Emotional Determinants of Infant-Mother Attachment

Author(s): Carroll E. Izard, O. Maurice Haynes, Gail Chisholm and Katherine Baak
Source: Child Development, Vol. 62, No. 5 (Oct., 1991), pp. 906-917
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: [Link]
Accessed: 23-01-2016 19:52 UTC

REFERENCES
Linked references are available on JSTOR for this article:
[Link]

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at [Link]
info/about/policies/[Link]

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@[Link].

Society for Research in Child Development and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
Child Development.

[Link]

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
Emotional Determinants of Infant-Mother
Attachment

Carroll E. Izard, 0. Maurice Haynes, Gail Chisholm, and

Katherine Baak
University of Delaware

IZARD,CARROLL E.; HAYNES,O. MAURICE; CHISHOLM, GAIL;and BAAK,[Link]

Determinants of Infant-MotherAttachment. CHILDDEVELOPMENT, 1991, 62, 906-917. The pres-
ent study examined the assumption that emotion-related characteristicsof mothers and infants
contribute to the development of infant-motherattachment in the first year of life. Mothers'
emotion and personality characteristicswere assessed with expressive-behavior ratingsand self-
report scales. Infant characteristicswere measured by emotion and temperament questionnaires
(mother report)and objective coding of facial expressions of emotions. Attachmentclassifications
were determined by means of the Strange Situation procedure, and a continuous-variableindex
of attachment security was derived by a discriminant function procedure. Mothers' emotion
experiences, expressive behaviors, and personality traits were significant predictors of the level
of security of the [Link]'expressive and temperamentalcharacteristics
as rated by their mothers were also significant predictors of attachment security.

Differential emotions theory (Izard, fluence the pattern or quality of infant-

1971, 1977, 1989; Izard & Malatesta, 1987), mother attachment. He discussed the
the conceptual framework for this study, possible contributions of sex differences and
holds that emotions are the keystone of ad- differences in infant activity level and excit-
aptation. They are viewed as the motiva- ability.
tional component of personality and social
Most of the empirical research on at-
relationships. Therefore, they are expected tachment has depended on the Ainsworth
to play a key role in the development of
infant-mother attachment, an outcome of ba- Strange Situation (Ainsworth, Blehar, Wa-
sic adaptational processes. ters, & Wall, 1978) as a means of studying
the behaviors and processes involved. Re-
Theoretical accounts of the determi- views of the literature reveal that research-
nants of attachment have focused mainly on ers disagree on the relative importance of
characteristics of the mother. Since Freud's mother and infant characteristics in de-
earliest writings (e.g., 1900, 1905), dynamic termining the quality of infant-mother at-
psychology has recognized the prominent tachment (for reviews, see Belsky & Isa-
role of the mother or "mother figure" in the bella, 1988; Bretherton, 1987; Bretherton &
social and emotional development of the Waters, 1985; Goldsmith & Alansky, 1987;
child. Combining concepts from psychoanal- Lamb, Thompson, Gardner, & Chernov,
ysis with ideas and observational methods 1985; Sroufe, 1985). These reviews summa-
from ethology, Bowlby (1969) brought into rize evidence from numerous studies that
sharp focus the significance of maternal be- have confirmed the Bowlby-Ainsworth posi-
havior in the development and maintenance tion that maternal sensitivity to the infant's
of the universal phenomenon of infant- signals and communications are important in
mother attachment, which he described as determining the quality of infant-mother at-
basically emotional in nature. tachment.
Bowlby (1969) did recognize that char- There is a smaller body of evidence sup-
acteristics ("biases") of the infant could in- porting the notion that infant characteristics

This work was supported by NIMH grant no. MH4205003 and NSF grant no. BND8706146.
The authors appreciate the assistance of Teri Brown Lawler, Christopher T. Hyde, Priscilla
Putnam, Deborah Libero, Geriann Krensel, Jill Matuny, Brian Simon, Beth Slomine, and Shan-
non Voirol with the collection and management of the data and the helpful comments of Larry
Cohen, Roger Kobak, and Jack Schuenemeyer on the manuscript. Address all correspondence to
Carroll E. Izard, Department of Psychology, 220 Wolf Hall, University of Delaware, Newark,
DE 19716.
[Child Development, 1991, 62, 906-917. ? 1991 by the Society for Researchin Child Development, Inc.
All rights reserved. 0009-3920/91/6205-0022$01.00]

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 907

are significant determinants, and this evi- studies, and the findings are not completely
dence comes from studies that found neona- consistent. Further, we do not know how a
tal or early infancy measures of tempera- temperamental disposition such as irritabil-
ment or reactivity (e.g., autonomic stability, ity or physiological reactivity is manifested
irritability) to be predictive of patterns of at- in observable behaviors that might influence
tachment (e.g., Belsky & Rovine, 1987; Fox, mother-infant interactions and the outcome
1985; Frodi & Thompson, 1985; Kagan, of the attachment process. One class of such
1982; Miyake, Chen, & Campos, 1985; Wa- observable behaviors might be infants' dis-
ters, Vaughn, & Egeland, 1980). Although crete emotion expressions, a view consistent
one of these studies (Waters et al., 1980) with the hypothesized relations among emo-
failed to replicate (Egeland & Farber, 1984), tions and dimensions of temperament (Gold-
meta-analysis of data from several studies smith & Campos, 1982; Rothbart, 1981). In
led to the conclusion that the temperament particular, we expected that the rate of in-
variable of proneness to distress predicted fants' negative emotion expressions in chal-
resistant behavior, a form of behavior associ- lenging or mildly stressful conditions would
ated with one type of insecure attachment predict the quality of infant-mother at-
(Goldsmith & Alansky, 1987). tachment.
All of the relevant reviews agreed that In summary, attachment theory assumes
the evidence supports the notion that attach- that attachment is essentially a relationship,
ment (as a social bond) is an important as- and that because the mother is the dominant
pect of human development, that there is figure she is the most important influence in
some relation between maternal sensitivity determining the quality of attachment (Ains-
to the infant's signals and the quality of worth et al., 1978; Bowlby, 1969, 1973).
infant-mother attachment, and that attach- Other positions (e.g., Kagan, 1982) have con-
ment as assessed in the Strange Situation is sistently argued that infant characteristics
influenced by patterns of infant-mother in- are significant determinants of attachment.
teractions. However, as Lamb, Thompson, Differential emotions theory (Izard, 1977;
Gardner, Charnov, and Estes (1984) and oth- Izard & Malatesta, 1987), which guides the
ers concluded, there is a need for further re- present study and the parent longitudinal
search on the specific maternal traits or be- project,' assumes that personality and inti-
havioral dimensions that influence the mate social relationships are rooted in emo-
quality of attachment. tions, and that emotion-related characteris-
tics of both mother and infant mediate the
Several studies have examined maternal
attachment process.
variables similar to those of the present
study. One of these showed that mothers of In keeping with the foregoing assump-
secure infants had a "better understanding tions and previous research, measures of
of their infants and their relationships with mother and infant characteristics were se-
their infants" (Egeland & Farber, 1984, p. lected for their assumed relations to emo-
759). This study found no significant differ- tions and their potential relevance to the
ences among mothers of secure and insecure development of attachment. Thus the assess-
infants on personality and anxiety scales. ment battery for mothers included measures
However, another study found that mothers of emotion experiences and emotion expres-
of securely attached infants rated them- siveness; a personality questionnaire that in-
selves as more extraverted than did mothers cludes measures of traits of affiliation, nur-
of insecurely attached infants (Bretherton, turance, and emotional security; and an
O'Connell, & Tracy, 1980). empathy inventory that measures concern
for the plight of others and the tendency to
Although some evidence indicates that become emotionally involved with others.
measures of infant irritability or reactivity
contribute to the prediction of the quality of In addition to the apparent conceptual
attachment, there are only a few supporting relations among the selected measures,

1 The parent project covers the period from birth to 51/2 years. In addition to the variables
described here, the project includes studies of the effects of positive and mildly stressful condi-
tions on expressive behavior, level of play, and problem-solving behavior. Such assessments at
ages 2-5 years provide outcome measures for evaluating the long-range predictive efficacy of
the early infancy data. The project also includes a psychophysiological component that examines
developmental changes in patterns of cardiac activity and their relations to expressive behavior
and the outcome measures.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
908 Child Development

there is some empirical evidence that shows perienced in daily life. Form IV of the DES
relations among them. Scores on the empa- was used in the present study. It is a modifi-
thy scale have been shown to interact with cation of the original DES (Izard, 1972;
other variables in predicting affective Izard, Dougherty, Bloxom, & Kotsch, 1974)
arousal and helping behavior (Stotland, Ma- that is applicable to various educational lev-
thews, Sherman, Hansson, & Richardson, els and ages. There is a three-item DES
1978), and measures of positive and negative scale for the emotions of interest, joy, sur-
affect were significantly and positively cor- prise, sadness, anger, disgust, contempt,
related with the personality traits (Diener fear, shame, shyness, guilt, and inner-
& Emmons, 1985; see also Eysenck & directed hostility (anger, disgust, and con-
Eysenck, 1985; Thayer, 1989). Finally, tempt directed toward the self). In the pres-
Bretherton et al. (1980) have shown that ent project, the reliabilities (r) for the trait
mothers of secure children rate themselves version of the DES scales administered to
as more sociable or extroverted. mothers at infant ages 2.5 and 4.5 months (N
= 66) were: interest .78, joy .77, surprise .73,
The assessment of infants included ob-
sadness .75, anger .74, disgust .72, contempt
jective measures of emotion expressions,
mothers' perceptions of their infants' emo- .70, fear .73, shame .70, shyness .69, guilt
tion expressions, and a mother-report tem- .50, inner-directed hostility .74. The DES
was administered a third time at infant age
perament measure with emotion-related di- 6 months. To maximize the stability of these
mensions. Numerous studies have shown
scales for data analyses, the DES scores
that several discrete emotion expressions are
functional in early infancy. For example, were averaged across administrations. To be
these expressions tend to elicit particular in- included, subjects had to complete the DES
at least two of the three times (N = 111).
terpretations and actions from the caregiver
(Izard, Huebner, Risser, McGinnes, & Emotion Expression Styles Ques-
Dougherty, 1980; Huebner & Izard, 1988). tionnaire.-The seven-item EESQ (Izard,
Other studies have shown that infants' be- 1981) was completed by mothers at infant
havior is clearly affected or guided by moth- age 2.5 (N = 114). This questionnaire re-
ers' emotion expressions (Campos, Hiatt, quires mothers to rate on a six-point scale
Ramsay, Henderson, & Svejda, 1978; Ter- the extent to which they typically hide or
mine & Izard, 1988). Thus we expected that express emotions in the presence of their
indexes of mothers' and infants' emotion infant. The EESQ yields expressiveness
characteristics would contribute to the pre- scores for seven emotions: interest, joy, sad-
diction of attachment. ness, anger, disgust, fear, and shame.

Method Personality traits.- Jackson's Personal-

ity Research Form E (PRF; Jackson, 1984)
Subjects was completed by mothers at infant age 2.5
One hundred fourteen white, middle- months (N = 89). The PRF consists of 22
class mother-infant dyads were recruited 16-item scales measuring specific traits of
from birth announcements in local newspa- personality.2 The scales are relatively inde-
pers. These dyads are scheduled to visit the pendent, and reliability (r) ranges from .50
Human Emotions Laboratory periodically to .91, with an average reliability of .72.
from infant age 2.5 to 63 months. Data for Moderate validity coefficients have been re-
the present study were obtained at infant ported for behavior ratings, trait ratings, and
ages 2.5, 3.0, 4.5, 9.0, and 13.0 months. self ratings (Jackson, 1984). When instructed
Eighty-one mother-infant dyads have com- to select the scales (traits) most likely to in-
pleted the procedures at 13 months, but be- fluence attachment, three judges (the first
cause of the nature of longitudinal research, two authors and Roger Kobak) unanimously
N's may vary with each set of variables. selected four traits (nurturance, affiliation,
Mother Measures autonomy, succorance), and these were used
in the analyses.
Emotion experience scales.-The Dif-
ferential Emotions Scale (DES) measures Empathy scales.-The Personal Feel-
the frequency with which emotions are ex- ing Inventory (PFI; Stotland et al., 1978) was

2 The PRF scales are

abasement, achievement, affiliation, aggression, autonomy, change,
cognitive structure, defendence, dominance, endurance, exhibition, harm avoidance, impulsiv-
ity, nurturance, order, play, sentience, social recognition, succorance, understanding, desirabil-
ity, and infrequency.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 909

completed by mothers at infant age 2.5 ment that asks the mother to rate her child
months. The PFI contains five subscales de- on 92 items relating to behavior in com-
signed to measure different aspects of empa- monly occurring situations. Only the scales
thy. These include: (1) Concern-assesses relating to negative emotions were used in
empathy in terms of role assumption, one's the present study-fear, or distress/latency
ability to imagine how it would feel to be in to approach novel stimuli (alpha = .81);
the other person's circumstances; (2) anger, or distress to limitation (alpha = .78);
Friend-measures the tendency to experi- and soothability (alpha = .82). Stability
ence a friend's emotions vicariously; (3) (9-12 months) correlations typically range
Fantasy-measures the extent of empathic from .43 to .80 (Rothbart, 1981). The IBQ
identification with characters in novels or was administered to the mother at infant age
movies; (4) Denial-determines the degree 9 months (N = 76).
to which one denies emotional involvement
or responsiveness; (5) Hostility-assesses Mother's perceptions of baby's emotion
the extent to which one reacts with anger in expressions (MPBEE).-The MPBEE
response to another's distress. (Izard, Buechler, & Huebner, 1979) asks the
mother to rate the frequency with which her
Derivation of maternal factor infant (compared with the average infant of
scores.-To obtain an acceptable subject- the same age) expresses each of 11 emotions
variable ratio, the scale or trait scores de- (interest, joy, surprise, sadness, anger, dis-
rived from each inventory of the mothers' gust, contempt, fear, shame, shyness, and
characteristics were subjected to factor anal- guilt). The mother completed the MPBEE
ysis. The factors that emerged on the vari- at infant ages 2.5, 4.5, 6, and 9 months. The
max rotation were adopted as the new mater- items reported with sufficient frequency for
nal (independent) variables to be used in the statistical analyses were interest, joy, sad-
prediction of attachment security. The re- ness, anger, crying, and demand attention.
sults for the separate factor analyses of the Reliability of single-item scales in a period
DES, EESQ, PFI scales, and the selected of rapid developmental change was, as ex-
scales of the PRF are shown in Table 1. Each pected, only moderate. For the MPBEE
inventory/measure yielded two orthogonal emotion-expression frequency scales admin-
factors. In each analysis, the factor structure istered at 4.5 and 6 months (the briefest in-
was exceptionally clear, with primary item tertest interval), the reliability coefficients
(scale) loadings ranging from the high 0.60s (N = 88) for the scales used in the present
to the 0.90s and separating widely from their study were as follows: sadness .21, anger .53,
secondary loadings. Each factor contains crying .56. For the 6- and 9-month adminis-
traits or scales that have conceptually con- trations (N = 82), the reliabilities were: sad-
gruent content and can easily be interpreted. ness .23, anger .47, crying .53. All correla-
The factor labels, shown in Table 1, were tions, except that for sadness are significant
easily derived from the item/scale content of at p < .001. The low reliability for sadness
the factors. The mothers' emotion experi- was probably due to the relative infrequency
ence (DES) and emotion expression (EESQ) of this expression. Only the negative scales
scales loaded on either a positive or negative were used in the analyses.
factor. The PFI and PRF factors can also be
viewed as generally positive or negative. Derivation of infant emotion-
More specifically, the positive PFI factor temperament factor scores from the
(empathy) represents empathic concern, two mother-report measures (IBQ,
whereas the negative PFI factor (hostility) MPBEE).-The three negative scales of the
entails projection and denial. The positive IBQ and the four negative scales of the
PRF factor (sociability) involves interper- MPBEE (which were moderately intercorre-
sonal warmth, friendliness, and extrover- lated)3 were subjected to factor analysis. The
sion, and the PRF negative factor (depen- results are in Table 2. Factor 1 is easily inter-
dence) contains items reflecting social preted as fussiness, defined as proneness to
anxiety, insecurity, and a lack of autonomy. cry and express negative emotions, espe-
cially in response to frustration. Factor 2 was
Infant Measures not readily interpretable and was excluded
Infant Behavior Questionnaire from further analyses.
(IBQ).-The IBQ (Rothbart, 1981) is a
behavior-based measure of infant tempera- Emotion expression indexes.-Emotion

3 IBQ distress to limitations correlated .48 and .36 with MPBEE crying and anger expression,
and IBQ distress to novel (fear) was correlated .22 with MPBEE crying.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
910 Child Development
TABLE 1
FACTORS
ANDITEM/SCALELOADINGS
FROMTHE FACTOR
ANALYSES
OF MATERNAL
MEASURES(Varimax Rotation)

DES-IV/EMOTION EXPERIENCE FACTORS

1. Negative 2. Positive

DES scales:
Disgust ................ .82 - .04
Sadness ................... .81 - .23
Guilt ....................... .81 -.10
Fear ........................ .77 -.01
Shame ..................... .76 -.10
Hostility, inward ..... .75 -.30
Shyness ................ .74 -.11
Anger ..................... .69 - .20
Contempt .............. .66 .08
Interest .................. - .23 .86
Joy .......................... - .34 .80
Surprise ................... .20 .74

EESQ (Emotion Expression) FACTORS

1. Negative 2. Positive

EESQ scales:
Disgust .................. .80 .08
Fear ....................... .79 - .05
Sadness .................. .70 .33
Anger .................... .68 - .09
Shame .................... .67 .15
Joy ........................... - .04 .84
Interest ................... .14 .80

PFI (Empathy) FACTORS

1. Hostility 2. Empathy

PFI scales:
Hostile .................... .79 .26
Denial ..................... .76 - .15
Friend ..................... - .72 .21
Fantasy .................. .14 .85
Concern ................. - .31 .75

PRF (Personality) FACTORS

1. Sociability 2. Dependence

PRF scales:
Nurture .................. .85 .02
Affiliation .83 .10
..............
Autonomy ................. - .06 - .78
Succorance ............. .05 .73

expression variables were derived from vid- visits at 9 months (N = 88). Two video cam-
eotaped facial behavior of infants during eras with zoom lenses provided close-up
positive and negative 2.5-3-min mother- full-face records of the infants' and mothers'
infant interactions with their mothers during faces. At 2.5 months, the conditions were
one visit at infant ages 2.5 (N = 108), 3 (N normal play, contingent play (mother
= 108), 4.5 (N = 95), 6 (N = 93), and two
responding only when her infant initiates

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 911
TABLE 2
FACTORS AND ITEM/SCALE LOADINGS FROM THE FACTOR ANALYSIS
OF THE IBQ ANDMPBEE

MPBEE/IBQ
(Temperament) FACTORS

MPBEE/IBQ SCALES 1. Fussiness 2. Mixed

Crying .................................. .81 .15

Anger ..................................... .65 -.17
Sadness ................................ .63 -.15
Distress, limits ................... .63 .41
Demand attention ........... .56 .06
Distress, novel ................. .24 .80
Soothability .................... - .27 .66

interaction), and still face (mother expres- contingent responding procedures were ad-
sionless and silent). At infant ages 3-9 ministered. The anger and sadness expres-
months, the positive interactions were nor- sion indexes for the negative emotion-
mal mother-infant face-to-face play, and in expression condition (based on mother
the negative interactions, mother expressed anger- and sadness-expression procedures)
anger or sadness facially and vocally. that were repeated at 3, 4.5, 6, and 9 months
were averaged to maximize stability and re-
Two complimentary systems were used duce the number of variables.
to code the facial behavior-a System for
Identifying Affect Expressions (Affex) Attachment classifications.-To assess
(Izard, Dougherty, & Hembree, 1983) and mother-infant attachment, the Ainsworth
the Maximally Discriminative Facial Move- Strange Situation procedure (Ainsworth et
ment Coding System (Max) (Izard, 1979). By al., 1978) was conducted at infant age 13.0
random stratified sampling, 25% of each con- months. Intercoder agreement on ratings of
dition (episode) was selected for Affex cod- the specific interactive behaviors for a sub-
ing. Between 14% and 33% of each segment sample of 16 subjects was 88%.4 The coders
was coded by independent coders using agreed on the major classifications (A,B,C)
Max. Intercoder-intersystem agreement was in 14 of the 16 cases, kappa = .77.
87%, 91%, 89%, 88%, 91%, and 90% for the The coding procedure results in the fol-
six visits, respectively. The unit variate for
lowing basic classifications: secure (B) in-
infant emotion expressions was the rate each
fants, who consistently show positive re-
emotion was expressed in the positive and
sponses to mother during reunion episodes
negative stressful conditions. The emotion and use mother as a secure base in which to
expressions observed with sufficient fre- explore the environment; insecure-avoidant
quency for statistical analyses were interest, (A) infants, who show little or no tendency to
joy, sadness, and anger. seek proximity or contact with mother even
In keeping with our focus, we reduced during reunion episodes and little to no dis-
the number of variables by retaining for the tress during separation; and insecure-
resistant (C) infants, who are characterized
analyses only the two negative expressions
(sadness and anger) in the negative by conspicuous contact and interaction re-
emotion-expression condition (which com- sisting behavior toward mother, particularly
bines the procedures of mother expressing during reunion episodes, remain angry
sadness and mother expressing anger) and throughout much of the procedure, and are
the ambiguous condition (which combines difficult to soothe.
the conditions of mother with still face or Of the 81 infants who completed the
responding only contingently). The anger Ainsworth Strange Situation procedure at
and sadness expression indexes for the am- age 13 months, 12 were classified as
biguous condition were obtained at age 2.5 insecure-avoidant (A) infants, 59 were classi-
months, the only time the still face and fied as secure (B) infants, and 10 were classi-

4 The coders were trained by an expert (Dante Cicchetti), and they also attained reliability
using video records that were precoded by another expert (Everett Waters).

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
912 Child Development

fled as insecure-resistant (C) infants. The PRF-sociability and dependence; PFI-

number of mothers completing the different hostility and empathy; DES-negative and
questionnaires and ratings, some of which positive emotion experiences. The attach-
were collected by mail, varied, as did the ment security (discriminant function) scores
number of infants who completed all the were the dependent variable. The hierarchi-
procedures at each age. cal regression analysis, presented in Table
3, yielded R2 = .26, F(8,58) = 2.55, p < .019.
Because of their relatively small N's, the
A and C subgroups were combined. Follow- To estimate the effects of missing val-
ing a scheme described by Richters, Waters, ues, typically caused by a missed visit or fail-
and Vaughn (1988), we used ratings on se- ure of the mother to return a particular in-
lected interactive behaviors (proximity seek- ventory, mothers with missing values on a
ing, contact maintaining, contact resistance, particular measure were compared on all
proximity avoidance) and frequency of cry- other factors with mothers having scores on
ing in reunion episodes 5 and 8 as indepen- the measure in question. These comparisons
dent variables in a discriminant function (t tests) revealed no significant differences."
analysis to distinguish secure and insecure Examination of the zero-order correla-
infants (as classified by the coders). We tions between mothers' emotion/personality
obtained a single discriminant function factor scores and infants' attachment security
(lambda = .393, F[9,71] = 12.17, p < .0001) scores revealed that, with one exception, the
that correctly identified 95% (56 of 59) in- relation between each of the mother charac-
fants classified as secure and 86% (19 of 22) teristics (factor scores) and the infant's secu-
of those classified as insecure, for an overall
rity of attachment was in the expected di-
hit rate of 93%.5 The infant scores or the lin- rection. Mothers of children with lower
ear discriminant function were used as a
insecurity scores (those more likely to be
continuous-variable index of attachment se- classified as securely attached) reported ex-
curity. In the distribution of security scores, periencing less negative emotion (DES fac-
the higher the score, the higher the like- tor 1) and more positive emotion (DES factor
lihood the infant was classified as inse-
2). These mothers were also more open in
cure. Hence, the discriminant-function their expression of negative emotion around
(dependent-variable) scores were termed in- their children (EESQ factor 1) and more so-
security scores. ciable, nurturant (PRF factor 1), and em-
pathic (PFI factor 2).
Results
Mothers of children with higher insecu-
Prior to examining the effectiveness of rity scores (those more likely to be classified
the mother and infant predictors, prelimi- as insecurely attached) reported experienc-
nary analyses were run to examine the char- ing more negative emotion (DES factor 1),
acteristics of their distributions. The as- but they were less open in their expression
sumptions of normality and linearity of negative emotions around their children
appeared reasonably well satisfied, and no (EESQ factor 1). These mothers had a
outliers were found. greater need for autonomy and felt more in-
Mother Characteristics and Infant-Mother secure and helpless (PRF factor 2: depen-
Attachment dence). Unexpectedly, they also reported ex-
The mothers' eight factor scores were pressing more positive emotions around
entered in a hierarchical regression with the their children in daily life (EESQ factor 2).
independent variables ordered in terms Infant Characteristics and Infant-Mother
of their hypothesized significance for Attachment
emotion-based maternal behaviors that in- A hierarchical regression with the four
fluence the quality of the infants' attach- sadness and anger expression indexes and
ment. The factors were entered in the fol- the one MPBEE/IBQ (mother-report emo-
lowing order: EESQ-negative and positive tion/temperament) negative factor as inde-
emotion expressions around the infant; pendent variables and the attachment secu-

5 The six infants who were "misclassified" by the discriminant function were recoded by
an independent coder blind to the original classifications and the statistical predictions. She
agreed with the original coder's classification five of six times and with the statistical prediction
once.
6 An analysis with missing values replaced through regression techniques yielded results
similar to those of the original analysis.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 913
TABLE 3
REGRESSION OF INSECURITY SCORES ON MEASURES OF MOTHERS'
CHARACTERISTICS

Mother, Factors R2 R2 Change P (F Change)

EESQ negative ...... .058 .058 .05

EESQ positive ......... 075 .017 .28
PRF sociability ...... .142 .068 .03
PRF dependence ...... .188 .045 .07
PFI empathy ........ .239 .051 .05
PFI hostility .......... .239 .001 .92
DES negative ............ 254 .015 .28
DES positive .............261 .007 .48
NoTE.-DES = Differential Emotions Scale, PRF = Jackson's Per-
sonality Research Form, PFI = Stotland's Personal Feeling (empathy)
Inventory, EESQ = Emotion Expression Style Questionnaire.

rity (discriminant function) scores as tions (MPBEE/IBQ factor 1). The direction
dependent variable yielded R2 = .23, was reversed for anger expression in the
F(5,51) = 3.10, p < .016 (see Table 4). En- negative emotion condition.
tering only the objectively coded infant- To test for possible interaction effects,
emotion expression indexes in the equation
the attachment scores were regressed on the
resulted in R2 = .12, p < .073.
two strongest mother variables, the two
The problem of infants with missing strongest infant variables, and their interac-
values was treated in the same way as with tions. None of the interaction terms was sig-
the mothers. There were no significant dif- nificant.
ferences between the groups with complete
and incomplete data on any of the indepen- Discussion
dent variables, and a regression with miss-
The results confirmed and extended ear-
ing values replaced yielded results very sim- lier evidence of mothers' influence on the
ilar to those from the original analysis.
infants' attachment security. More clearly
The relations between the infant emo- than previous studies, the present one
tion (independent) variables and the attach- showed that mothers' emotion-expressive
ment security (dependent) variable were behaviors are significant predictors of the
generally in the expected direction. The quality of attachment. Also unique to the
higher the infants' insecurity scores, the present study was the finding that young in-
higher were their objective (Affex-coded) fants' emotion-expressive behaviors in
and perceived (MPBEE) rate of sadness and mildly stressful interactions with their moth-
anger expressions and the more they cry, de- ers contributed to the prediction of the level
mand attention, and show distress in mildly of attachment security. In a separate study
stressful or frustrating conditions to limita- of resting-state cardiac activity as a source

TABLE 4
REGRESSION OF INSECURITY SCORES ON MEASURES OF INFANT
CHARACTERISTICS

Infant Variables R2 R2 Change P (F Change)

MPBEE/IBQ1 .......... 117 .117 .01
Sadness, NE .......... .144 .027 .20
Anger, NE ............ .196 .052 .07
Sadness, AMB ......... .201 .005 .56
Anger, AMB ........... .233 .032 .15

NOTE.-The final four variables are emotion expression indexes in

the negative emotion condition (NE) and the ambiguous condition (AMB).
MPBE/IBQ1 = the principal component (factor 1) derived from the factor
analysis of the IBQ and MPBEE scales.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
914 Child Development
of individual difference variables, heart-rate insecurity and have a strong need for social
variability was also a significant predictor of support (PRF factor 2, dependence). This
attachment, even after the strongest mother may result in maternal behaviors that reflect
variables were partialed out of the regres- insecurity and thus adversely affect the qual-
sion equation (Izard et al., in press). ity of nurturance and infant attachment. That
these mothers are less sociable (PRF factor
Mother Characteristics as Predictors of
1, sociability) and show less empathic con-
Infants' Attachment Security cern (PFI factor 2, empathy) may cause them
The results showed that mothers' char- to be perceived by their infants as less avail-
acteristic emotion experiences, emotion- able emotionally.
expressive behaviors, and other emotion-
related traits of personality (sociability, The present results showed that there
empathy) predicted the quality of attach- are apparent similarities or logical relations
ment. All the mother characteristics that ef- among the Strange Situation behaviors of in-
fectively predicted the infants' attachment fants with low or high insecurity scores and
security scores can be viewed as direct or the characteristics of their mothers. Secure
indirect measures of emotion. The results infants (or those with lower insecurity
support our assumption that the motivational scores) typically interacted with mother and
influence of discrete emotion experiences stranger in a friendly fashion. They showed
and expressions helps explain a causal rela- negative emotions in response to separation
tion between mother's personality and but greeted mother positively on return,
infant-mother attachment. were soothed easily, and were soon at play
again. Similarly, their mothers were rela-
Unexpectedly, the higher the infants' tively high on affiliation/sociability and em-
insecurity scores, the more their mothers re- pathic concern and open in their expression
ported expressing positive emotions around of negative emotions, which they experience
their children in daily life (EESQ factor 2, less frequently than mothers of children
positive expression). This kind of self report with high insecurity scores. Insecure infants
may also be a form of defense against the (or those with relatively higher insecurity
relatively frequent experiencing of negative scores) showed more negative emotions in
emotions. Interestingly, they can report ex-
early infancy and in the Strange Situation.
periencing more negative emotions and less Although their mothers experienced more
positive emotions (DES-IV), but they cannot negative emotions, they were least expres-
admit expressing negative emotions around sive of negative emotions around their chil-
their children (EESQ) consistent with their dren. The data suggest that insecure infants'
pattern of emotion experiences. This incon- behavior sometimes complements rather
sistency between their reports of emotion than mirrors that of their primary caregiver.
experiences (DES-IV) and emotion expres- Apparently, insecure infants react to stress-
sions around their children (EESQ) suggests ful events with excessive negative emotion
that their expressive behaviors may some- or behavior in order to obtain attention and
times send mixed or uninterpretable mes- nurturance from relatively unsociable and
sages to their children. emotionally unresponsive mothers.
The present results raise the question of
how the mother's emotion and personality Infant Characteristics as Predictors of
Attachment
characteristics influence her child's behav-
Infant variables derived from two mater-
iors in the Strange Situation and hence the nal reports were predictive of infant-mother
insecurity score. The study did not address attachment. The contribution to prediction
this issue directly, but a plausible explana-
of measures of Affex/Max-coded emotion ex-
tion might begin with the observations of
pressions in challenging or mildly stressful
Bowlby (1969) that the mother plays a lead- situations only approached significance. The
ing role in determining the nature or pattern diversity of these measures somewhat
of this relationship. An intimate, dyadic rela-
strengthens their support of the hypothesis
tionship requires that each member learn that individual differences in infants con-
from the other and make appropriate adapta-
tribute to the level of attachment security.
tions. In the infant-mother relationship,
mother displays a complex array of behav- We can only speculate why rate of anger
iors and characteristics, and she is fre- expressions in the negative-emotion condi-
quently in the role of model and teacher. tion was negatively correlated with insecu-
Our results suggest that mothers of insecure rity scores. Perhaps the more secure infants
children are themselves prone to feelings of perceived their mothers' negative emotion

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 915
expressions (which were continued for 2-3 characteristics predict attachment does not
min) as more atypical and disruptive than necessarily mean that they explain the rele-
did the insecure infant. Although mothers of vant developmental processes. However,
more secure infants generally express nega- each of the effective variables can be logi-
tive emotions more openly, they very proba- cally related to behaviors that are thought to
bly do not express them to their infants in be involved in the development of attach-
close-up interactions for so long a period of ment. Examples are mothers' affiliative and
time. empathic tendencies and infants' rates of
The notion that the responses of the se- negative emotion expressions in daily life.
cure infants in the negative emotion- One could argue that none of the infant
expression condition is an exception to the measures in the present study is truly inde-
more general finding of their lower negativ- pendent of the mother. The objectively
ity and irritability is supported by their coded infant emotion expressions are indi-
mothers' rating of them on the MPBEE and vidual difference measures, but they were
IBQ. The MPBEE/IBQ factor 1 (fussiness) obtained in mother-infant interactions.
scores were positively related to the insecu- However, the indexes of resting state heart-
rity scores. Thus, according to the mothers' rate variability can be considered as individ-
perceptions and ratings, the more insecure ual difference measures that are indepen-
the infants, the more frequently they cry, de- dent of the mother, and it did account for
mand attention, express sadness and anger, significant unique variance in the measure
and show distress to limitation in daily life. of attachment (Izard et al., in press). The re-
These results are consistent with data from sults for infants' emotion expressions to-
different laboratories that have shown that gether with those for heart-rate variability
individual differences in irritability or reac- provide support for our argument that indi-
tivity in the neonatal period predict the vidual differences in infants play a signifi-
broad categories of secure and insecure at- cant role in determining the level of attach-
tachment (e.g., Miyake et al., 1985). They ment security.
are also consistent with the finding from
Goldsmith and Alansky's meta-analytic References
study that proneness to distress predicts be-
havior (resistance) characteristic of one type Ainsworth, M. D., Blehar, M. D., Waters, E., &
of insecurity. Wall, S. (1978). Patterns of attachment: A
psychological study of the Strange Situation.
Studies such as those of Miyaki et al. Hillsdale, NJ: Erlbaum.
(1985), Frodi and Thompson (1985), and the Belsky, J., & Isabella, R. (1988). Maternal, infant,
present one support Kagan's (1982) notion and social-contextual determinants of attach-
that individual differences in temperament ment security. In J. Belsky & T. Nezworske
and emotionality contribute significantly to (Eds.), Clinical implications of attachment
the development of attachment. As already (pp. 41-93). Hillsdale, NJ: Erlbaum.
noted, measures of infants' resting state Belsky, J., & Rovine, M. (1987). Temperament and
heart-rate variability in early infancy have attachment security in the Strange Situation:
been found to predict attachment security An empirical rapprochement. Child Develop-
(Izard et al., in press). This finding suggests ment, 58, 787-795.
that infant predictors of attachment security Bowlby, J. (1969). Attachment and loss: Vol. 1.
can be identified and measured in the first Attachment. New York: Basic.
few months of life. Bowlby, J. (1973). Attachment and loss: Vol. 2.
On the Relative Effectiveness of Mother Separation, anxiety, and anger. New York:
Basic.
and Infant Predictors
Bretherton, I. (1987). New perspectives on attach-
Although the present data do not pro- ment relations: Security, communication, and
vide a definitive test of the relative pre-
dictive power of mother and infant charac- internal working models. In J. D. Osofsky
teristics, they do show that both members of (Ed.), Handbook of infant development (2d
ed., pp. 1061-1100). New York: Wiley.
the dyad contribute in unique ways to the
Bretherton, I., O'Connell, B., & Tracy, R. (1980,
prediction of attachment security. Each one
of a range of conceptually distinct mother March). Styles of mother-infant and
and infant variables indexed by different stranger-infant interaction. Paper presented
at the International Conference on Infant
methods accounted for unique variance in
the attachment security index. Studies, New Haven, CT.
Bretherton, I., & Waters, E. (1985). Growing
That the identified mother and infant points of attachment theory and research.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 916 Child Development
Monographs of the Society for Research in Izard, C. E. (1981). Emotion Expression Styles
Child Development, 50(1-2, Serial No. 209). Questionnaire. Unpublished manuscript.
Campos, J. J., Hiatt, S., Ramsay, D., Henderson, Izard, C. E. (1989). The structure and functions of
C., & Svejda, M. (1978). The emergence of emotions: Implications for cognition, motiva-
fear on the visual cliff. In M. Lewis & L. A. tion, and personality. In I. S. Cohen (Ed.), The
Rosenblum (Eds.), The development of af- G. Stanley Hall Lecture Series (Vol. 9, pp.
fect: Vol. 1. Genesis of behavior (pp. 149- 35-73). Washington, DC: American Psychol-
182). New York: Plenum. ogy Association.
Diener, E., & Emmons, R. A. (1985). The indepen- Izard, C. E., Buechler, S., & Huebner, R. R.
dence of positive and negative affect. Journal (1979). Mothers' perceptions of infants' emo-
of Personality and Social Psychology, 47, tion expressions. Unpublished manuscript,
1105-1117. University of Delaware, Newark.
Egeland, B., & Farber, E. A. (1984). Infant-mother Izard, C. E., Dougherty, F. E., Bloxom, B. M., &
attachment: Factors related to its develop- Kotsch, W. E. (1974). The Differential Emo-
ment and changes over time. Child Develop- tions Scale: A method of measuring the
ment, 55, 753-771. subjective experience of discrete emotions.
Eysenck, H. J., & Eysenck, M. W. (1985). Person- Unpublished manuscript, University of Dela-
ality and individual differences. New York: ware, Newark.
Plenum. Izard, C. E., Dougherty, L. M., & Hembree, E. A.
Fox, N. (1985). Behavioral and autonomic ante- (1983). A system for identifying affect expres-
cedents of attachment in high-risk infants. In sions by holistic judgments (Affex). Newark:
M. Reite & T. M. Field (Ed.), The psychobiol- Instructional Technology Center, University
ogy of attachment and separation (pp. 389- of Delaware.
414). New York: Academic Press. Izard, C. E., Huebner, R. R., Risser, D., McGin-
Freud, S. (1900). The interpretation of dreams. In nes, G. C., & Dougherty, L. M. (1980). The
J. Strachey (Ed. and Trans.). The standard young infant's ability to produce discrete
edition of the complete psychological works emotion expressions. Developmental Psy-
of Sigmund Freud (Vol. 3). London: Hogarth. chology, 16, 132-140.
Freud, S. (1905). Three essays on the theory of Izard, C. E., & Malatesta, C. Z. (1987). Perspec-
sexuality. In J. Strachey (Ed. and Trans.). The tives on emotional development: I. Differen-
standard edition of the complete psychologi- tial emotions theory of early emotional devel-
cal works of Sigmund Freud (Vol. 7). London: opment. In J. D. Osofsky (Ed.), Handbook of
Hogarth. infant development (2d ed., pp. 494-554).
Frodi, A., & Thompson, R. (1985). Infants' af- New York: Wiley-Interscience.
fective responses in the strange situation: Ef- Izard, C. E., Porges, S. W., Simons, R. F., Haynes,
fects of prematurity and of quality of attach- O. M., Parisi, M., Cohen, B., & Hyde, C. T. (in
ment. Child Development, 56, 1280-1298. press). Infant cardiac activity: Developmental
Goldsmith, H. H., & Alansky, J. A. (1987). Mater- changes and relations with attachment. De-
nal and infant temperamental predictors of at- velopmental Psychology.
tachment: A meta-analytic review. Journal of Jackson, D. N. (1984). Personality Research Form
Consulting and Clinical Psychology, 55, manual (3d ed.). London, Ont.: Research Psy-
805-816. chologists Press.
Goldsmith, H. H., & Campos, J. (1982). Toward a Kagan, J. (1982). Heart rate and heart rate variabil-
theory of infant temperament. In R. Emde & ity as signs of a temperamental dimension in
R. J. Harmon (Eds.), The development of at- infants. In C. E. Izard (Ed.), Measuring emo-
tachment and affiliative systems (pp. 161- tion in infants and children (pp. 38-66). New
193). New York: Plenum. York: Cambridge University Press.
Huebner, R. R., & Izard, C. E. (1988). Mothers' Lamb, M. E., Thompson, R. A., Gardner, W. P., &
responses to infants' facial expressions of sad- Charnov, E. L. (1985). Infant-mother attach-
ness, anger and physical distress. Motivation ment: The origins and developmental sig-
and Emotion, 12(2), 185-196. nificance of individual differences in Strange
Izard, C. E. (1971). The face of emotion. New Situation behavior. Hillsdale, NJ: Erlbaum.
York:Appleton-Century-Crofts. Lamb, M. E., Thompson, R. A., Gardner, W. P.,
Izard, C. E. (1972). Patterns of emotions. New Charnov, E. L., & Estes, D. (1984). Security
York:Academic Press. of infantile attachment as assessed in the
Izard, C. E. (1977). Human emotions. New York: "Strange Situation." Behavioral and Brain
Plenum. Sciences, 7, 127-171.
Izard, C. E. (1979). The Maximally Discriminative Miyake, K., Chen, S., & Campos, J. (1985). Infant
Facial Movement Coding System (Max). temperament, mother's mode of interactions,
Newark: Instructional Technology Center, and attachment in Japan: An interim report.
University of Delaware. In I. Bretherton & E. Waters (Eds.), Growing

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions
 Izard et al. 917

points of attachment theory and research. Hansson, R. O., & Richardson, B. Z. (1978).
Monographs of the Society for Research in Empathy,fantasy and helping. Beverly Hills,
Child Development, 50(1-2, Serial No. 209). CA: Sage.
Richters, J. E., Waters, E., & Vaughn, B. E. (1988). Termine, N. T., & Izard, C. E. (1988). Infants' re-
Empirical classification of infant-mother rela- sponses to their mothers' expressions of joy
tionships from interactive behavior and cry- and sadness. Developmental Psychology, 24,
ing during reunion. Child Development, 59, 223-229.
512-522. Thayer, R. E. (1989). The biopsychology of mood
Rothbart, M. (1981). Measurement of tempera- and arousal. New York: Oxford University
ment in infancy. Child Development, 52, Press.
569-578. Waters, E., Vaughn, B., & Egeland,.B. (1980). In-
Sroufe, L. A. (1985). Attachment classification dividual differences in infant-mother attach-
from the perspective of infant-caregiver rela- ment relationships at age one: Antecedents in
tionships and infant temperament. Child De- neonatal behavior in an urban, economically
velopment, 56, 1-14. disadvantaged sample. Child Development,
Stotland, E., Mathews, K. E., Sherman, S. E., 51, 208-216.

This content downloaded from [Link] on Sat, 23 Jan 2016 [Link] UTC
All use subject to JSTOR Terms and Conditions