7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 1/11

See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/276853075

Mediating and moderating processesunderlying the association between maternalcognition and infant attachment

ARTICLE in JOURNAL OF APPLIED DEVELOPMENTAL PS YCHOLOGY · AUGUST 2015Impact Factor: 1.85 · DOI: 10.1016/j.appdev.2015.04.001

READS

54

4 AUTHORS:

Erinn Hawkins

Bond University

5 PUBLICATIONS 20 CITATIONS

SEE PROFILE

Sheri Madigan

The University of Calgary

32 PUBLICATIONS 569 CITATIONS

SEE PROFILE

Greg Moran

The University of Western Ontario

116 PUBLICATIONS 1,720 CITATIONS

SEE PROFILE

David Pederson

The University of Western Ontario

91 PUBLICATIONS 1,749 CITATIONS

SEE PROFILE

Available from: Sheri MadiganRetrieved on: 28 October 2015

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 2/11

Mediating and moderating processes underlying the associationbetweenmaternal cognition and infant attachment

Erinn Hawkins a , , Sheri Madigan b ,c, Greg Moran a , David R. Pederson a

a Department of Psychology, University of Western Ontario, London, Canadab Applied Psychology and Human Development, University of Toronto, Toronto, Canadac Suspected Child Abuse and Neglect Program (SCAN), Hospital for Sick Children, Toronto, Canada

a b s t r a c ta r t i c l e i n f o

Article history:Received 25 September 2012Received in revised form 6 February 2015Accepted 1 April 2015Available online xxxx

Keywords:AttachmentSecure-base scriptsMaternal sensitivityMaternal insight

The goal of this longitudinal study was to elucidate aspects of maternal cognition associated with maternalsensitivity and the development of infant attachment. Seventy-seven mothers were administered a number of measures over the course of the infant ’s rst year, including the Adult Attachment Interview (AAI), secure basescripts, and a measure of maternal insight about her infant ’s behavior. Maternal sensitivity was assessed usingthe maternal sensitivity Q-Sort and infant attachment using the Strange Situation Procedure. Associations werefound between AAI coherence, secure base scripts, and maternal sensitivity; however, no direct association wasfound between maternal insight and sensitivity. A mediator-moderator model suggested that: 1) AAI coherencemediated the associationbetween secure base scripts and maternal insight;and2) maternal insight andsensitivityinteractedto predictsecureattachment.Findings suggest theneed to further investigatethe roleof cognitive factorsimplicated in the intergenerational transmission of attachment.

© 2015 Elsevier Inc. All rights reserved.

Attachment theory stipulates that cognitive representations of early

relationships have an enduring in uence on the development andmaintenance of subsequent relationships throughout the life span(Bowlby, 1969, 1973, 1980 ). In line with this theory, maternal attach-ment representations, measured via the Adult Attachment Interview(AAI; Main, Goldwyn, & Hesse, 2002 ), have been found to be a strongpredictor of the quality of infant attachment ( Madigan et al., 2006a;van IJzendoorn, 1995 ), although the speci c mechanisms of this trans-mission process remain elusive ( Atkinson et al., 2005 ). Central to devel-opmental theorizing is the proposition that maternal sensitivity, such asprompt andappropriate responding to infant needsand cues, is a prima-ry mechanism through which maternal attachment representationsshape the quality of infant attachment ( Cassidy, 1994; Main, Kaplan, &Cassidy, 1985 ). However, van IJzendoorn ’s (1995) in uential meta-analysis describing what he termed the transmission gap has called thisfundamental hypothesis into question. A more recent meta-analysis fur-ther concluded that evidence for the mediational role of maternal sensi-tivity is ephemeral, appearing and disappearing as a function of theparticular measures used and the nature of the population under study(Atkinson et al., 2005 ). These ndings suggest a need to explore othermechanisms underlying the transmission of attachment betweenmother and child ( Atkinson et al., 2000; Bernier & Meins, 2008; De

Wolff & van IJzendoorn, 1997; Meins, Fernyhough, Fradley, & Tucky,

2001; Pederson, Gleason, Moran, & Bento, 1998; Raval et al., 2001 ).Extant theory and research suggests that other maternal cognitivefactors, such as interpersonal scripts related to attachment – termed se-cure base scripts – aswell as a mother ’s immediate insight or attributionsabout the nature of speci c infant behavior may have an important im-pact on maternal sensitivity, and in turn, infant attachment (e.g., Bernier& Meins, 2008; Fonagy & Target, 2005; Meins et al., 2001; Oppenheim &Koren-Karie, 2002; Slade, Grienenberger, Bernbach, Levy, & Locker,2005; Tini, Corcoran, Rodrigues-Doolabh, & Waters, 2003 ). The primarypurpose of the present study was to examine the interrelations betweendifferent levels of maternal cognitions (i.e., attachment representationson theAAI,securebase scripts, and maternal insightinto infantbehavior),their associations with maternal sensitivity, and infant attachment. Theinterrelations between these variables have been relatively unexaminedin a single study to date.

Attachment cognition

Based on the premise that information contained in the social worldis much too complex for the brain to process in its entirety, Bowlby(1969) reasoned that the brain must contain cognitive structures,which he termed internal working models , that direct attention toward,represent, and organize relevant features of the social environment.He surmised that these mental models of relationships are derived fromrepeated interactions with attachment gures in infancy. Advances incognitive and social psychology have led to a number of attempts to

Journal of Applied Developmental Psychology 39 (2015) 24 – 33

Correspondingauthor at:Psychology Clinic, BondUniversity, UniversityDrive, Robina,QLD, Australia, 4229.

E-mail address: ehawkins@bond.edu.au (E. Hawkins).

http://dx.doi.org/10.1016/j.appdev.2015.04.001

0193-3973/© 2015 Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

Journal of Applied Developmental Psychology

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 3/11

clarify the role of internal working models in attachment beyond in-fancy by further describing their content, organization, and function incaregiver-child and other close relationships (c.f., Baldwin, 1992;Bretherton, 1991; Bretherton & Munholland, 1999; Main et al., 1985;Mikulincer et al., 2001; Waters, Rodrigues, & Ridgeway, 1998 ). Manyof these authors converge in their conjecture that internal workingmodels are composed of two major components: (1) complementarydeclarative knowledge about the self and other in various interactions,

and the affective and behavioural patterns involved in these interac-tions (i.e., attachment representations); and (2) procedural knowledgecontaining rules for processingsocial information, suchas propositionalstatements in the form of if-then statements, and interpersonal scripts.

Declarative knowledge speci es the properties that characterize ob- jects or people, and consists of episodic memory (memory of speci cpastevents) and semanticmemory (abstract or generalized informationextracted from repeated experiences of a similar kind). Accordingly,through repeated experiences in interactions with attachment gures,individuals encode episodic and semantic memories of the self, otherand their relationship ( Baldwin, 1992 ). A variety of terms have beenused to describe the cognitive representations containing declarativeknowledge of relationships, including state of mind about attachment(Main et al., 2002 ), secure base schemas (e.g., Mikulincer et al., 2001 ),and relational schemas (e.g., Baldwin, 1992 ). We selected the termattachment representations as a more general term to convey the no-tion that these cognitive structures are organized around key featuresof the attachment relationship (i.e., secure base/safe haven function of the relationship). The Adult Attachment Interview ( Main et al., 2002 )provides a measure of the coherence of an adult ’s declarative knowl-edge, or attachment representations, regarding their relationship withtheir owncaregivers. It is the degreeof coherencewith which one artic-ulates episodic and semantic memories of their relationship with par-ents that is of prime importance in the classi cation of attachmentsecurity in adulthood using this measure.

In addition to declarative knowledge, internal working models arealso proposed to contain procedural knowledge. According to Mainet al. (1985) , the procedural aspects of attachment representations pro-vide rules for the direction and organization of attention and memoryas it relates to social interaction in signi cant or intimate relationships.These procedural rules are thought to further control an individual ’saccess to speci c forms of declarative knowledge regarding self, other,and the relationship between self and other. Procedural knowledge isthought to function largely outside of conscious awareness, and as a re-sult, it tends to be triggered automaticallyby relevant information withinan individual ’s environment. These implicit procedural rules are thoughtto be encoded in the form of if-then propositions and interpersonalscripts ( Baldwin, 1992; Bretherton, 1991; Waters et al., 1998; Waters &Waters, 2006 ).

Secure base scripts

Scripts have been de ned as stereotyped sequences of action

containing systematic rules about common or frequently occurringsituations ( Schank & Abelson, 1977 ). Waters and colleagues ( Waters &Rodrigues, 2004; Waters & Waters, 2006; Waters et al., 1998 ) proposedthat interpersonal scripts related to attachment provide a summary of an individual ’s history of secure base support. By abstracting variousfeatures of recurrent interactions with signi cant caregivers, individ-uals construct a script-like representation of the attachment relation-ship, which over time is elaborated and generalized to other closerelationships.

In keeping with the primary tenets of attachment theory, Watersand Waters (2006) maintained that attachment representations areorganized around a generalized secure base script that is described bythe following sequence of events: (1) an individual and secure base(e.g., mother/partner) engage in constructive social interaction; (2) the

individual encounters an obstacle or threat, which evokes distress;

(3) the individual seeks comfort and assistance from the secure base;(4) assistance and comfort from the secure base is offered; (5)assistanceandcomfortfrom thesecurebase isaccepted andis effective; and(6)con-structive social interaction is re-established between members of thedyad. Consistent with Bowlby ’s (1969) conceptualization of internalworking models, the formation of a complete, well consolidated, andreadily accessible secure base script is dependent on a history of securebase support that is consistent and coherent. Conversely, a history of

secure base support that is inconsistent, incomplete or ineffective shouldresult in the formation of a secure base script that is less well con gured,less consolidated and less accessible. Once established, the secure basescript is thought to provide individuals with generalized expectations of behavioral and emotional outcomes across all secure base interactions,which aids in the planning and organization of their own behavioraland emotional responses, including those with their own children.

Following this logic, the quality of maternal secure base scriptsshould provide a cognitive framework for secure base interactions intheir relationships with their own children. In a study of 31 mother-infant dyads, Coppola, Vaughn, Cassibba, and Costantini (2006) foundthat mothers ’ access to a rich and detailed secure base script was astrong predictor of their sensitivity in responding to their infants ’ signalsand cues. Moreover, the extent to which mothers ’ secure base scriptswere well elaborated accounted for 43% of the variance in their interac-tive behavior with their infants, substantially more than that predictedby maternal representations of their own attachment history ( vanIJzendoorn, 1995 ). Maternal secure base scripts were also found to corre-late with coherence scores on the AAI ( Coppola et al., 2006; Dykas,Woodhouse, Cassidy, & Waters, 2006 ), secure infant-mother attachmentrelationships, as assessedby the Strange Situation Procedure ( Ainsworth,Blehar, Waters, & Wall, 1978; Tini et al., 2003 ) and the AttachmentBehavior Q-Set (AQS; Bost et al., 2006; Vaughn et al., 2006a; Waters,1995; Wong et al., 2011 ).

Maternal insight into infant behavior

Ainsworth (1969 ; Ainsworth, Bell, & Stayton, 1971 ) emphasized theimportance of certaincognitive capacitiesthatserve to facilitate maternalsensitivity, and in turn, a secure attachment relationship. These includedan awareness of infants ’ signals and cues, an accurate understanding of infant behavior and the circumstances leading to such behavior, and acognitive capacity to see things from the infant ’s point of view. Con-versely, she observed that mothers of infants with insecure attachmentappeared to be less sensitive in their responding because of a dif cultyin reading infant behavior, misattributing the intentionsof their infants ’

behavior, and/or an inability to see past their own perspective to adoptthat of the infant. Several terms have been used to describe the cognitivephenomenon alluded to in Ainsworth ’s early observations, includingre ective function ( Fonagy, 2002; Fonagy & Target, 1997 ), mind-mindedness ( Meins et al., 2001 ), and insightfulness ( Oppenheim &Koren-Karie, 2002 ). For the purpose of the current study, the termmaternal insight was selected to describe a mother ’s proclivity for un-

derstanding her infant ’s behavior in terms of the underlying emotionalor psychological states motivating that behavior.

Drawingon Ainsworth ’s early observations, several authorsexaminedthepropositionthat maternal insightabout their infant ’s behavioris asso-ciated with maternal sensitivity and the quality of the attachment rela-tionship ( Demers, Bernier, Tarabulsy, & Provost, 2010; Fonagy, 2002;Fonagy & Target, 1997; Laranjo, Bernier, & Meins, 2008; Meins et al.,2001; Oppenheim & Koren-Karie, 2002; Slade, 2005 ). In a sample of 129 mother-infant dyads, Koren-Karie, Oppenheim, Dolev, Sher, andEtzion-Carasso (2002) found that mothers who were classi ed as posi-tively insightful displayed signi cantly more sensitive behavior, mea-sured concurrently, than mothers who did not demonstrate a proclivityfor positive insight. Further, maternal insightfulness made a unique con-tribution to the prediction of secure attachment relationships, beyond

that of maternal sensitivity. Similarly, in a sample of 71 mother-infant

25E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 4/11

dyads, Meins et al. (2001) found that appropriate mind-minded com-ments were moderately correlated with concurrent measures of mater-nal sensitivity and they also successfully differentiated between secureand insecure attachment dyads. Laranjo et al. (2008) also found thatthe number of appropriate comments mothers made about their infant ’smental state was moderately related to maternal sensitivity. Togetherthese ndings suggest that maternal insight about infant behavior mayprovide a basis for a broader range of caregiving behavior, not immedi-

ately encapsulated by the concept of maternal sensitivity, that alsoserve to promote infant attachment security ( Koren-Karie et al., 2002 ).

A network of cognitive structures: Mediator-moderator model

Given the different array of cognitive structures shown to be associat-ed with maternal sensitivity and infant attachment security, there is aneed to clarify the conceptual relations between these variables withinan intergenerational transmission model. Bretherton (1991) proposedthat cognitive structures are best conceptualized as “ multiple-levelschema-hierarchies ” (p.18) with information related to speci c interac-tions within close relationships forming the lowest level of the hierarchy,and increasingly more abstract and generic constructs making up thehigher levels. Accordingly, the different levels of maternal cognition arenot orthogonal, but rather, overlap in a network of cognitive structurescontaining complementary declarative knowledge of self and signi cantothers, as well as procedural knowledge providing rules for behaviourand information processing in particular relational contexts ( Baldwin,1992 ). Whereas declarative knowledge about the self and other in inter-action is thought to be important in theinterpretation of socialbehaviour,procedural knowledge about which information is attendedto is thoughtto be important in theorganizationof incoming social information, in thegenerationof interpersonal expectations, andin theplanning of appropri-ate behavioural responses.

In line with Bretherton (1991) , we propose thatmaternal cognitionsare connected in a network along a continuum of more abstract andgeneric to more speci c cognitive structures (see Fig. 1). Moving alongthis continuum, secure base scripts are the most generic cognitive struc-tures and contain basic expectanciesof secure base interactions and corebeliefs about self and other. This is consistent with Waters and Waters(2006) conjecture that secure base scripts are the underpinnings of more generalized representations of attachment and close relationships.Nested below the level of the secure base script are more speci c repre-sentations of close relationships. These representations store primarilysemantic or declarative knowledge of self and others in relationships of a particular domain (e.g., representations of child-parent relationships,as measured by the AAI). It has been suggested that secure base scriptrepresentations in uence AAI coherence in a variety of ways, such as un-derstanding and recognizing the salience of the AAI questions, retrieving

and organizing memories of early experiences with caregivers, and eval-uating the relevance and completeness of one ’s responses to the AAI(Waters, Brockmeyer, & Crowell, 2013 ). Finally, the most speci c levelof the network is the immediate cognitive appraisal of behavior, whichinfers meaning frombehavior in the form of attributions, or psychologicalinsight about the behavior.

Based on this line of reasoning, Fig. 1 illustrates an alternative inter-generational transmission model whereby maternal attachment repre-

sentations of early childhood are thought to mediate the associationsbetween secure base scripts and maternal insight into infant behavior.In other words, a well-elaborated secure base script should lead to amore coherent conceptualization of one ’s own attachment history,which should lead to increased cognitive insight regarding infant be-havior during parent-child interactions. A direct link between maternalsecure base scriptsand maternal sensitivity is also suggested due to thestrong procedural component of secure base scripts. Previous researchhas shown that increased maternal insight into infant behavior is bothassociated with maternal sensitivity, and uniquely predictive of secureattachment relationships ( Koren-Karie et al., 2002; Meins et al., 2001 ).This suggests that maternal insight regarding infant behavior mightinteract with maternal sensitivity to produce relational outcomes.Assuming these two aspects of sensitive caregiving (i.e., cognitive andbehavioral) are moderately, but not perfectly correlated, we wouldexpectmaternal insight to addto theprediction of infantattachment se-curity differentially. That is, mothers who are both highly sensitive andinsightful would be most likely to promote infant attachment security,while mothers low on both sensitivity and insightfulness would beleast likely to promote infant attachment security. Mothers high oneither sensitivity or insightfulness should also promote infant attach-ment security,but to a lesserextent than mothershigh on both sensitiv-ity and insightfulness.

Two main objectives were pursued in the current study. First, to rep-licate and extend previous associations between maternal sensitivity,measured using the shortened version of the maternal sensitivity Q-sort(Pederson, Moran, & Bento, 1999 ) and a) attachment representations,as measured by the AAI ( Main et al., 2002 ); b) secure base scripts(Waters & Rodrigues, 2004 ); and c) insight using an adapted version of Koren-Karieet al. ’s (2002) insightfulness measure. We expectto replicatethe previously established associationsamongst these measures. Second,in an original lineof research, weseek to explore the extent to which var-ious cognitive structures work together in a mediated-moderator modelto shape the quality of attachment relationship. Our predictions here aretwofold. First, maternal attachment representations are expected to me-diate the associations between secure base script and maternal insightabout infant behavior. Second, maternal insight about infant behavior isexpected to interact with maternal sensitivity to predict attachmentrelationships.

Fig. 1. Mediated-moderator model of attachment transmission.

26 E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 5/11

Method

Participants

Participants were primiparous adult mothers and their infants whowere recruited from a registry of mothers expressing interest in partici-pating in child development studies during their postpartum hospitalstay. Inclusion criteria were as follows: mother was 1) at least 20 years

of age at the time of infant’

s birth; 2)

uent in English; and infant was3) rstborn; 4) born physically healthy at full term gestation; and5) less than 3 months of age upon request for participation. Of the 138mother-infant dyads eligible for participation, 77 (56%) agreed to partic-ipate, 52 (38%) declined, and 8 (6%) could not be reached. Mothers werenot asked to give speci c reasons for declining; however, of those whovolunteered reasons, time constraints, travel plans, or intending tomove from the local area were provided. Mothers declining participationdiffered fromthose agreeing to participate in terms of household income,with those declining participation reporting a lower average annualhousehold income ($40 000 – 49 000 CAD) than those agreeing to partic-ipate ($50 000 – 59 000 CAD).

Thecurrent study included the77 participating mother-infantdyads(36 female; 41 male) who took part in the rst home visit at infant age3 months. Retention rates were strong, with 70 (91%) mother-infantdyads continuing in the study at the 4 month home visit, and 66(86%) remaining in the study at both the 10 month home visit and13 month laboratory visit. All mothers gave informed consent for theirparticipation as approved by the institutional Research Ethics Board.

Demographic information was collected at the outset of the study,when the infants were three months of age. Average maternal age andeducation was 30.08 years (range = 20.20 – 44.85 years), and 14.5 years(range = 11 – 18 years), respectively. The majority of mothers weremarried (75%) or common-law (17%). Participating families were pre-dominantly Caucasian (96%) and middle-class with an average annualhousehold income within the $50 000 – $59 000 range. Eighty-eight per-cent of mothers were on maternity leave, 7% worked part-time and 5%worked full-time.

Procedure

The study included three home visits, when infants were 3, 4, and10 months of age, and one laboratory visit, when infants were13monthsof age.Eachhomevisit was between 1 ½ to 2 hours inlength.At the 3-month visit, basic demographic information was obtained, andthe AAI was administered. At the 4-month home visit, mothers and in-fants participated in a modi ed version of the still face procedure(Tronick, Als, & Brazelton,1977 ), which was later used to assess mater-nal insightinto infant behavior. Thesequenceof the still-face procedureoccurred in three 60 second increments: 1) interact with the infantwithout touch; 2) mother stops all verbal and nonverbal interactionwith the infant and as best as possible, maintains a neutral facial expres-sion; and 3) mother reengages in face-to-face interaction with the infant

without using touch. Following this procedure, mothers were askedto engage their infants in play on the oor as they normally wouldfor 3 – 5 minutes (i.e., post still face procedure).

The still face and post still face procedures were videorecorded andupon their completion a home visitor reviewed the videorecordingwith mothers via a laptop computer. While reviewing the video,mothers were asked to provide commentary about their own andtheir infants ’ internal experience during the following speci c segmentsof recorded interaction: the nal thirty seconds of the rst still faceperiod (playful interaction), both thirty second segments of the secondstill face period(actualstill face), the rst thirty seconds of the third stillface period (return to playful interaction), and the rst thirty seconds of the post still face procedure. Mothers were instructed to place them-selves back into the same frame of mind as they were while interacting

with their infants, and describe for the visitor their own and their

infants ’ thoughts and feelings during the various segments (see Sanders& Dadds, 1992 ).

At the 10-month home visit mother-infant interactions werevideorecorded in several different contexts, including: 1) a brief separa-tion where the mother was instructed to leave the infant alone in theroom with the visitor for approximately 2 minutes; 2) unstructuredplay, in which mothers were asked to play with their infants withoutthe use of toys; 3) structured play, in which mothers were asked to

engage the child in new toys brought by the visitors; and

nally, 4) afeeding session. Following these tasks, mothers were asked to generatestories using the word prompt methodology developed by Waters andcolleagues ( Waters & Rodrigues, 2004; Waters et al., 1998 ). Home visi-tors kept running notes describing maternal sensitivity as well as inter-actions between the dyad and subsequent to the visit, completed the25- item version of the MBQS. Mother-infant dyads were invited totheuniversitylaboratory to participatein theStrangeSituation Procedure(SSP; Ainsworth et al., 1971; Ainsworth et al., 1978 ) when infants were13 months of age.

Every effort was made to ensure that visitors and coders were blindto other information about the dyads. For instance, new home visitorswere assigned to different dyads across time periods, and all transcriptsand videorecordings were renumbered with identifying informationremoved to increase anonymity of dyads across coding schemes.

Measures

Attachment representations: the Adult Attachment Interview (AAI; Mainet al., 2002 ). The AAI is a semi-structured interview designed to assessadults ’ current conceptualizations of their own attachment history,and consists of a series of questions asking participants to re ect onpast experiences with their primary caregivers. Interviews wereaudiotaped, transcribed verbatim, and scored along a series of several9-point summary scales, including coherence of mind, which are thenused to classify each transcript into one of the four categories: autono-mous, dismissing, preoccupied, unresolved/disoriented.Transcripts char-acterized by a coherent state of mind, and included descriptions of childhood relationships with parents that were convincingly supportedby episodic memories, were classi edas autonomous . Transcripts charac-terizedby positiveor idealized depictionsof attachment gures thatwereeither notsuf cientlysupported, or werecontradicted by episodic mem-ories were classi ed as dismissing . Transcripts characterized by currentanger with an attachment gure, or vague unfocused discussion aboutchildhood relationshipswithparents were classi edas preoccupied . Tran-scripts whereby the mother demonstrated lapses in monitoring of rea-soning or discourse during discussions of loss or abuse were classi edas unresolved/disoriented . Transcripts were coded by trained and reliablecoders. A total of 16 transcripts (20.8%) were coded for the purpose of reliability, and agreement between coders for all four classi cationswas 100%. Reliability on the coherence of mind scale was robust(ICC = .90).

Secure base scripts ( Waters & Rodrigues, 2004 ). Secure base scripts wereassessed using the word prompt methodology developed by Watersand colleagues ( Waters & Rodrigues, 2004; Waters et al., 1998 ) admin-istered at the 10-month home visit. During this task mothers were pre-sented with sixsets of promptword outlines that suggesteda particularst ory line. Four story outlines presented attachment-relevant wordprompts, including two mother-child scenarios (e.g., Baby ’s Morning:mother, baby, play, blanket, hug, smile, story, pretend, teddy bear, lost,found, nap;and Doctor ’s Of ce:Tommy,bike, hurt, mother,hurry,doctor,cry, shot, mother, toy, stop, hold) and two scenarios aboutadult romanticpartners (e.g., Jane and Bob ’s Camping Trip: Jane, Bob, bags, hurry, tent,wind, collapse, upset, camp re, shadow,sounds,hug;andSue ’s Accident:Sue, road, accident, hospital, wait, Mike, tears, doctor, home, dinner, bed,hug). The remaining two story outlines were included as ller stories,

presenting non-attachment relevant scenarios. Word prompts were

27E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 6/11

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 7/11

Data analytic strategy

The AAI data can be presented either categorically or continuously,using the summary scale scores that determine categorical classi ca-tion.Thecoherence ofmindscale isone ofthe primary scalesusedin de-termining autonomous classi cations on the AAI. As such, it was used inall data analyses requiring continuous variables. This approach is inkeeping with a taxometric study of the AAI showing that the variation

underlying autonomous individuals’

representations of their earlyattachment history is more consistent with a dimensional model ratherthan a taxonic model ( Roisman, Fraley, & Belsky, 2007 ).

Partial correlations were used to test associations between differentlevels of attachment representations and maternal sensitivity, whilecontrolling for covariates. The bootstrapping approach, a nonparametricresampling procedure used for estimation and hypothesis testing, wasused to test mediational hypotheses ( Preacher & Hayes, 2008 ). Thisapproach was selected because of its advantages over the causal stepsapproach (e.g., Baron & Kenny, 1986 ); namely, it provides a direct testof the mediation hypothesis, allows for the statistical control of covari-ates, and has increased statistical power over the causal steps approach(Dearing & Hamilton, 2006 ). The bootstrapping method is also pre-ferred over the Sobel test when the sample size is small to moderate,as in the present study. The bootstrapping method draws a large num-ber (e.g., 1000 or more) of arti cial samples from the original datathrough a process of sampling with replacement. The mean direct andindirect (i.e.,mediation)effects andtheir con denceintervals are calcu-lated using estimates from the bootstrap samples (5000 bootsrappingsamples were used). The null hypothesis of no indirect effect is rejectedif thebootstrappedcon denceinterval does notcontain zero ( Dearing &Hamilton, 2006; Preacher & Hayes, 2008 ). Finally, SSP classi cationswere collapsed into a dichotomous variable (i.e., secure/insecure, withdisorganized infantscategorized as insecure regardless of their secondaryclassi cation), and binary logistic regression analyses were conducted toevaluate moderation. Beta weights ( β ), standard errors ( SE ), Wald ’schi-square statistics, and p values are reported.

Results

Results are presented in three sections. Descriptive statistics andpreliminary data analyses examining associationsbetween independent,dependent, and demographicvariables arepresented rst.Next, analysesevaluating the study ’s main objectives are presented, beginning with anexamination of associations between maternal cognitive structures andmaternal sensitivity. These are then followed by analyses examining themediated-moderator model; namely (a) the roleof maternalattachmentrepresentations (i.e., AAI coherence) as a mediator between maternalsecure base scripts and insight about infant behavior, and (b) the role of maternal insightasa moderator of theassociationbetweenmaternalsen-sitivity and secure attachment relationships.

Descriptive statistics and preliminary analyses

The majority of mothers in the present study were classi ed asautonomous (48.6%; n = 34) on the AAI. Of the remaining mothers,20.0% (n = 14) were classi ed as dismissing, 4.3% ( n = 3) as preoccu-pied, and 27.1% as unresolved ( n = 19). The majority of infants wereclassi ed as secure in the SSP (50.0%; n = 33), while 50% were classi-

ed as insecure, speci cally, avoidant ( n = 10, 15.2%), ambivalent(n = 1, 1.5%), or disorganized ( n = 22, 33.3%).

A multivariate analysis of variance revealed no signi cant differencesin household income, maternal age at the commencement of the study,or maternal education level, according to SSPclassi cationsand maternalsensitivity. However, signi cant correlations were observed betweenmothers ’ secure base script score and several demographic variables, in-cluding maternal age ( r = .36, p b .01), education ( r = .25, p b .05),

and household income ( r = .33, p b

.01). Maternal coherence of mind

scores were also signi cantly correlated with maternal age ( r = .30, p b .05) and household income ( r = .26, p b .05). Maternal insight intoinfant behavior scores were signi cantly correlated with education level( r = .29, p b .05) and household income ( r = .29, p b .05). Based onthese ndings, maternal age, education level, and household incomewere used as covariates in relevant data analyses.

Replication analyses: associations between maternal sensitivity and attach-

ment representations, secure base scripts, and maternal insight into infant behavior

Partial correlations between maternal sensitivity and mothers ’ AAIcoherence, secure base scripts, and maternal insight, controlling formaternal age and education level, as well as household income arepresented in Table 1 . As expected, both AAI coherence and secure basescripts were signi cantly associated with maternal sensitivity. Speci -cally, mothers with well-elaborated secure base scripts were more sen-sitive in their interactions with their infants, and mothers who had ahighly coherent conceptualization of their own attachment historywere more sensitive in their interactions with their infants. Contraryto expectations, however, maternal insight into infant behavior wasnot directly associated with maternal sensitivity.

Maternal cognitive structures, maternal sensitivity and the attachment relationship

a. Mediation analyses. As recommended by Hayes (2009) , we onlyreport theresults of theboostrappingapproach. This approach provideda test of the indirect effect of mothers ’ secure base script (predictorvariable) on their insight into infant behavior (outcome variable)through their attachment representations (potential mediator vari-able), while statistically controlling for maternal age, education andhousehold income. Results revealed that there were no partial effectsof the control variables of maternal age ( b = -.04, SE = .03, n.s.), edu-cation ( b = .1160, SE = .08, n.s.), and household income ( b = .09,SE = .06, n.s.), suggesting that these variables did not uniquely con-tribute to the prediction of maternal insight into infant behavior.As previously suggested, one advantage of the bootstrapping ap-

proach is that it provides a formal test of indirect effects. After control-ling for effects of maternal age, education and household income, theindirect effect was signi cant with a point estimate of .08 and 95% BCaCI of .01– 24. 1 Thus, a well-elaborated secure base script leads to in-creased coherence regarding one ’s own attachment history, which inturn, leads to greater insight into the psychological motives underlyingthe infant ’s behavior during parent-child interactions.

b. Moderation analyses. It was also predicted that mothers ’ insightinto their infants ’ behavior would moderate the association betweenmaternal sensitivity and secure attachment relationships. To test thisprediction, hierarchical binary logistic regression analyses were con-ducted, with infant attachment (i.e., secure/non-secure) as the outcomevariable. The covariates maternal education and household incomewere entered as predictor variables in the rst step. The variables

maternal insight and sensitive behavior were centered and entered inthe second step along with the product of these two centered variablesrepresenting the interaction term. As detailed in Table 2 , resultsrevealed that neither maternal education, nor household income con-tributed to the prediction of secure attachment classi cations whenconsidered alone, or when considered with the other predictor vari-ables. There was no main effect of maternal insight into infant behavior,but there was a marginal main effect of maternal sensitivity, suggestingthat in general, mothers who were sensitive in their interactions withtheir infants were more likely to be in secure relationships with their

1 To con rm the direction of our mediational model, we tested the alternative modelthat securebase scriptsmight mediate the associationbetween maternal attachmentrep-resentations and maternal insight into infant behavior. The indirect effect in the alterna-tive model was not signi cant, suggesting that secure base scripts does not mediate the

association between maternal attachment representations and maternal insight.

29E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 8/11

infants. As expected, the interaction between maternal insight into in-fant behavior and maternal sensitivity was signi cant.

To aid in the interpretation of the interaction, the probability of asecure attachment relationship as a function of maternal insight intoinfant behavior and maternal sensitivity is presented in Fig. 2. One stan-dard deviation above and below the mean wasestablished as cut pointsfor maternal sensitivity. Maternal insight was divided into low,medium,and high based on cut points representing the bottom, middle, and topthird of the distribution. An examination of Fig. 2 suggests that contraryto expectations, infants of mothers whodisplayed lowsensitivityduringparent- child interactions bene ted most from high levels of maternalinsight into their infant ’s behavior. In other words, for these mothers,high levels of maternal insight into their infant ’s behavior increasedthe probability of a secure attachment relationship. Medium levels of maternal insight also increased the probability of a secure attachmentrelationship formotherswhodisplayed lowlevels of maternalsensitivity,although to a lesser extent. On the other hand, and again contrary to ex-pectations, mothers who were highly sensitive in their interactions withtheir infants at 10-months were more likely to be in secure attachmentrelationship with their infants, regardless of the quality of their cognitiveinsight into their infant ’s behavior. Rather surprisingly, it appears thathighly sensitive mothers showing low levels of insight into their infants ’

behavior were most likely to be in secure attachment relationships withtheir infants, althoughit is not clear whether thisdifference is statisticallysigni cant.

Discussion

The broad purpose of thepresent study wasto further clarify theroleof maternal cognition in the intergenerational transmission of attach-ment. Results from the present study replicated previous ndingsshowing associations between AAI coherence, secure base scripts, andmaternal sensitivity. In contrast to previous research, no direct associa-tion was found between maternal insight into infant behavior andmaternal sensitivity. As predicted, there was evidence of a mediator-moderator model, such that a) AAI coherence mediated the associationbetween securebase scripts andmaternal insight into infantbehaviour;and b) there was an interaction between maternal insight into infantbehavior and maternal sensitivity to predict secure attachment

relationships. However, the speci c natureof the interaction wassome-what surprising. Each of these ndings will be discussed in turn.

Mothers with a rich and well-elaborated secure base script weresigni cantly more likely to be sensitive in their interactions with theirinfants. Although the association reported here was modest comparedto that found in a previous study( Coppola et al., 2006 ), it provides addi-tional evidence of a direct association between concurrent measures of maternal secure base scripts and maternal sensitivity. Mothers ’ attach-ment representations were also signi cantly associated with maternalsensitivity; such that mothers who were more coherent on the AAIwere signi cantly more likely to be sensitive in their interactions withtheir infants approximately 6 months later. This nding is consistentwith numerous earlier studies showing correlations of a similar magni-tude (see van IJzendoorn, 1995 for review ).

Contrary to expectations, however, the present study did not repli-cate previous research showing a direct association between maternalinsight into infant behavior and maternal sensitivity (e.g., Koren-Karieet al., 2002; Laranjo et al., 2008; Meins et al., 2001 ). There are severalpossible explanations for this nding. First, in the current study, mater-nal insight was assessed at infant age 3 months, which is temporallyearlier than other studies using similar measures (i.e., 12-months;Koren-Karie et al., 2002 ). Second, the current study used an adaptedversion of Oppenheim and Koren-Karie ’s (2002) insightfulness mea-sure. Thus, of primary importance in accounting for these divergent

ndings may be differences in operationalizations of maternal insight,and differences in measures of maternal sensitivity. For instance, al-thoughaccuracyor appropriateness of maternal mental statecommentswas considered in each coding scheme, valence of the comments wasnot a primary feature of the coding scheme used in the current study.This can be contrasted with the coding scheme used by Koren-Karieet al. (2002) , which emphasized acceptance, warmth, and positivity of attributions in their ratings of insightfulness. On the other hand,

Laranjo et al. (2008) , following guidelines outlined by Meins et al.(2001) , used the numberof appropriate comments on the infant ’s men-tal state as a measure of her understanding of her infant ’s mental life.

A third reason for these divergent ndings may be differences inwhen maternal insight and sensitivity were measured. Maternal insightis often conceptualized as a cognitivecomponent of maternal sensitivity(Meins et al., 2001 ). Certainly in Ainsworth et al. (1971) original con-ceptualization of maternal sensitivity these concepts were closelyintertwined. However, more recently it has been argued that “ respon-siveness to the child ’s physical and emotional needs should be clearlydistinguished from mothers ’ capacity or willingness to engage withtheir infants at a mental level ” (Meins et al., 2001 , p. 638, originalemphasis). Based on these arguments and our interest in evaluatingthespeci c contributionsof each concept in theintergenerational trans-

mission of attachment, particular care was made to reduce overlap in

Table 2Binarylogisticregression analyses examiningthe interaction betweenmaternal insight in-to infant behavior and maternal sensitivity in the prediction of secure attachmentrelationships.

Predictors B SE Wald ’s χ 2

Step 1Constant − 2.014 2.32 0.751Maternal education 0.176 0.18 0.967Household income − 0.082 0.13 0.398

Step 2Constant − 2.072 2.64 0.614Maternal education 0.189 0.20 0.906Household income 0.091 0.14 0.418Maternal insight into infant behavior − 0.210 0.32 0.329Maternal sensitivity 0.962 0.51 3.521 +

Maternal insight × sensitivity − 0.999 0.49 4.143*

p b

.05, +

p b

.10.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

lowlow sensitivity high sensitivity

P r o

b a

b i l i t y o

f S e c u r e

A t t a c

h m e n

t

low insight

med insight

high insight

Fig. 2. Probabilityof secure attachmentas a function of maternal insightinto infant behaviorand maternal sensitivity.

Table 1Partial correlations between maternal attachment representations and maternal sensitivity,controlling for maternal age, maternal education, and household income.

2. 3. 4.

1. Secure base scripts .26* .12 .25*2. AAI coherence .30** .22*3. Maternal insight into infant behavior .084. Maternal sensitivity

** p ≤ .01, * p ≤ .05.

30 E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 9/11

their measurement. Speci cally, maternal sensitivity was assessed at adifferent point in time than maternal insight, and our measure of maternal sensitivity was selected because of its primary focus onobservable maternal behaviour ( Pederson et al., 1999 ). Moreover, ma-ternal insight into infant behavior was evaluated using only maternalmental state comments generated while reviewing videotaped interac-tion and not during the interaction itself as was the case in previousstudies. As such, evaluations of maternal insight remained clearly dis-

tinct from evaluations of maternal sensitivity in the present study.Finally, given that previous ndings have shown unique linksbetween maternal insight into infant behavior and secure infant attach-ment ( Koren-Karie et al., 2002 ), as well links between concepts akin tomaternal insightfulness and other features of maternal sensitivity notencapsulated by maternal sensitivity ( Grienenberger, Kelly, & Slade,2005 ), it may well be that associations between maternal insight of in-fant behavior and maternal sensitivity are more complex and indirect(Lundy, 2003 ). Indeed, the present study showed an interactionbetween maternal insight and maternal sensitivity in the prediction of secure infant attachment. These ndings are discussed in more detailbelow.

A mediated-moderator model of transmission

Analyses testing the rst part of the mediated-moderator model of attachment transmission showed that, as predicted, maternal attach-ment representations mediated the association between secure basescripts and maternal insight into infant behavior. In other words, these

ndings suggest that a well elaborated secure base script leads to in-creased coherence regarding one ’s own attachment history, which inturn, leads to greater insight into the psychological motives underlyinginfant behavior during parent-child interactions. However, thismediatedpath accounted foronly 23% of thevariance, suggestingthat other factorsmay also be important in the prediction of maternal insight into infantbehavior. Since conceptualizations of maternal insight, and closely re-lated conceptssuch as re ective functioning (e.g., thecapacity to under-stand behavior in terms of underlying mental states or intentions;Fonagy & Target, 1997 ), are akin to other meta-cognitive capacitiessuch as mind-reading or theory of mind, it seems reasonable to suggestthat maternal insight into infant behavior may also be in uenced bymore general cognitive factors, such as general intelligence, and/ or cog-nitive exibility.Future research in thisarea would bene t frominclusionof these other more general cognitive factors in models of intergenera-tional transmission.

Findings from the present study supported the hypothesis that ma-ternal insight into infant behavior interacts with maternal sensitivityin the prediction of secure infant attachment; however, the speci c na-ture of the interaction was rather unexpected. For instance, contrary toexpectations mothers who were highly sensitive in their interactionswith their infants were more likely to be in secure attachment relation-ships with their infants, regardless of the extent to which they demon-strated insight into their child ’s behavior six months earlier. These

ndings, together with evidence of a direct link from mothers ’ securebase scripts to maternal sensitivity, suggest that for a subset of mothersproceduralknowledge based on their securebase script maybe primaryin the promotion of sensitive behavior, and in turn, secure attachment.For this subset of mothers, the extent to which they show insightabout their infant ’s behavior did not add to the prediction of secure at-tachment. On the other hand, high levels of maternal insight into infantbehavior may procure some resilience on infants of mothers lowin sen-sitivity. That is, high maternal insight increased the probability of beingin secure attachment relationships with their infants for mothers whodisplayed low levels of maternal sensitivity. Similarly, medium levelsof maternal insight also increased the probability of being in secure at-tachment relationships with their infants for mothers who displayedlow levels of maternal sensitivity, although to a lesser extent. These re-

sults are consistent with previous ndi ngs showing that mothers in a

high risk sample who showed high levels of re ective functioningwere considerably more likely to be in secure attachment relationshipswith their infants than mothers of equally high risk who showed lowlevels of re ective functioning ( Fonagy, Steele, Steele, Higgitt, &Target, 1994 ). Nonetheless, given the small sample size of the currentstudy and limited research on maternal insight into infant behaviorand attachment transmission, such ndings must be viewed as prelim-inary and in need of replication.

Although there were several notable strengths to the current study,such as its longitudinal and multi-method design, there are also somelimitations. First, our sample was relatively small. Second, mother-child dyads were predominantly white and middle-class and as result,the generalizability of our ndings beyond this homogenous group islimited. Third, a large proportion of infants in the current communitysample were insecure-disorganized (33%) Speci c hypotheses regardinglinks betweenmaternal cognition and behavior and disorganized attach-ment were not explored in the current study. It will be important forfuture studies to elucidate such linkages as there is currently little theo-retical or empirical research in this regard. Potential questions include:to what extent do multiple levels of maternal cognition and behavior in-

uence the development of attachment disorganization? How do theselevels work in combination with maternal sensitivity to promote orreduce the development of disorganized attachment? One notablestudy by Grienenberger et al. (2005) examined the relation betweenmaternal re ective functioning and disrupted maternal behavior, anestablished precursor of disorganized attachment (e.g., Madigan,Moran, & Pederson, 2006b ). Grienenberger et al. (2005) found thatmothers who displayed higher levels of re ective functioning alsoshowed lower maternal disrupted (and extremely insensitive) behav-ior, as assessed with the AMBIANCE measure ( Lyons-Ruth, Bronfman,& Parsons, 1999 ). Furthermore, the association between maternal re-

ective functioning and the security of the attachment relationshipwas mediated by maternal disrupted behavior. These results are espe-cially noteworthy because extant theoretical models portray disruptedmaternal behavior as a determinant of disorganized attachment rela-tionships rather than secure attachment relationships (e.g. Lyons-Ruth, Yellin, Melnick, & Atwood, 2005 ; see also, Moran, Forbes, Evans,Tarabulsy, & Madigan, 2008 ).

The current line of research suggests several important applicationsforclinicalpractice,both foraugmenting caregiverwell-being,as well asthe caregiver-child relationship. Research by Mikulincer and Shaver(2003, 2010) suggests that adult mental health can be bolstered byexposure to secure-based mental representations (e.g., secure basescripts). Priming studies that activate the perception of attachment

gure availability (i.e., security priming; Mikulincer & Shaver, 2003 )have shown to not only reduce stress, but also improve positive mood,self-concept, compassionate, empathetic behaviour, and mental health(Mikulincer & Shaver, 2010 ). Thus, regardless of an individual ’s secureor insecure attachment status, enhancing the perception of an attach-ment gure ’s availability and responsivity can positively in uence indi-vidual well-being.

In regardsto augmenting components of parenting to improve childoutcomes, a growing body of evidence suggests that relatively brief interventions focused primarily on enhancing maternal sensitivity areeffective at promoting the development of secure infant attachment(e .g., Bakermans-Kranenburg, van IJzendoorn, & Juffer, 2003; Heinickeet al., 1999; Moss et al., 2011 ). Results from the current study suggestthat the dual focus of enhancing maternal sensitivity, as well as recog-nizing and understanding children ’s internal states (e.g., the Circle of Security intervention; Hoffman, Marvin, Cooper, & Powell, 2006 ),could be particularly bene cial for promoting secure parent-childattachment relationships. In the current study, high levels of maternalinsight were not directly related to sensitive behaviour; however,when combined with low levels of sensitivity, it increased the prob-ability of a secure infant attachment. Thus, ideally, attachment-based

interventions should be both experiential and behaviorally based, as

31E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 10/11

7/21/2019 Hawkins 2015 JADP

http://slidepdf.com/reader/full/hawkins-2015-jadp 11/11

child attachment, and behavioral outcomes for maltreated children: A randomizedcontrol trial. Development and Psychopathology , 23, 195– 210.

Oppenheim, D., & Koren-Karie, N. (2002). Mothers' insightfulness regarding theirchildren's internal worlds: The capacity underlying secure child-mother relationships.Infant Mental Health Journal , 23(6), 593 – 605.

Pederson, D., Gleason, K., Moran, G., & Bento, S. (1998). Maternal attachment representa-tions, maternal sensitivity, and the infant-mother attachment relationship.Developmental Psychology , 34, 925– 933.

Pederson, D. R., Moran,G., & Bento,S. (1999). Maternal Behavior Q-sort Manual Version 3.1. Retrieved November 3, 2004, from http://www.ssc.uwo.ca/psychology/faculty/ pedmor/mbqmanual.html

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessingand comparing indirect effects in multiple mediator models. Behavior ResearchMethods , 40(3), 879 – 891.

Raval,V., Goldberg,S., Atkinson,L., Benoit, D.,Myhal, N.,Poulton,L., et al.(2001). Maternalattachment, maternal responsiveness and infant attachment. Infant Behavior & Development , 24, 281– 304.

Roisman, G. I.,Fraley, R. C., & Belsky,J. (2007). A taxometric study of theAdultAttachmentInterview. Developmental Psychology , 43(3), 675 – 686.

Sanders, M. R., & Dadds, M. R. (1992). Children ’s and parents ’ cognitions about family in-teraction: An evaluation of video-mediated recall and thought listing procedures inthe assessment of conduct-disordered children. Journal of Clinical Child Psychology, 21(4), 371 – 379.

Schank, R., & Abelson, R. (1977). Scripts, plans, goals, and understanding. New York, NY:Wiley.

Schechter, D. S., Meyers, M. M., Brunelli, S. A., Coates, S. W., Zeanah, C. H., Davis, M., et al.(2006). Traumatized mothers can change their minds about their toddlers: Under-standing how a novel use of videofeeback supports positive changeof maternal attri-butions. Infant Mental Health Journal , 27 (5), 429 – 447.

Slade, A. (2005). Parental re ective functioning: An introduction. Attachment & HumanDevelopment , 7 (3), 269 – 281.

Slade, A., Grienenberger, J., Bernbach, E., Levy, D., & Locker, A. (2005). Maternal re ectivefunctioning, attachment, and the transmission gap: A preliminary study. Attachment & Human Development , 7 (3), 283 – 298.

Tarabulsy,G., Provost, M.,Bordeleau, S.,Trudel-Fitzgerald, C.,Moran, G.,Pederson, D.,et al.(2009). Validation of a short version of the maternal sensitivity Q-set applied to abrief video record of mother – infant interaction. Infant Behavior & Development , 32,132 – 136.

Tini, M., Corcoran,D., Rodrigues-Doolabh,L., & Waters,E. (2003,March). Maternal attach-ment scriptsand infantsecure base behavior.In H.Waters & E.Waters (Chairs), Script-

like representations of secure base experience: Evidence of cross-age, cross-cultural,and behavioral links. Poster symposium presented at the Biennial Meetings of theSociety for Research in Child Development, Tampa, FL.

Tronick, E. D., Als, H., & Brazelton, T. B. (1977). Mutuality in mother-infant interaction. Journal of Communication , 27 (2), 74 – 79.

vanIJzendoorn, M. H. (1995). Adult attachment representations, parental responsiveness,and infant attachment: A meta-analysis on the predictive validity of the adult attach-ment interview. Psychological Bulletin , 117 , 387– 403.

Vaughn, B., Verissimo, M., Coppola, G., Bost, K., McBride, B., Krzysik, L., et al. (2006a). Ma-ternal attachment script representations: Longitudinal stability and associations withstylistic features of maternal narratives. Attachment & Human Development , 8,

199–

208.Vaughn, B. E., Waters, H. S., Coppola, G., Cassidy, J., Bost, K. K., & Verissimo, M. (2006b).Script-like attachment representations and behavior in families and across cultures:Studies of parental secure base narratives. Attachment & Human Development , 8(3),179 – 184.

Waters, E. (1995). Theattachment Q-set (Version 3). In E. Waters, B. Vaughn, G. Posada, &K. Kondo-Ikemura (Eds.), Caregiving,cultural, and cognitive perspectiveson secure-basebehavior and working models: New growing points of attachment theory and research.Monographs of the Society for Research in Child Development , 60. (pp. 234 – 246) (2 – 3,Serial No. 244).

Waters, T. E., Brockmeyer, S. L., & Crowell, J. A. (2013). AAI coherence predicts caregivingand care seeking behavior: Secure base script knowledge helps explain why. Attachment & Human Development , 15(3), 316 – 331.

Waters, H. S.,& Rodrigues, L. M. (2004). Narrative assessment of adult attachmentrepresen-tations: Thescoring of secure basescript content . Unpublishedmanuscript: Departmentof Psychology, State University of New York at Stony Brook.

Waters, H. S., Rodrigues, L., & Ridgeway, D. (1998). Cognitive underpinnings of narrativeattachment assessment. Journal of Experimental Child Psychology , 71, 211– 234.

Waters, H. S.,& Waters, E. (2006). The attachment working models concept: Among otherthings, we build script-like representations of secure base experiences. Attachment & Human Development , 8(3), 185 – 197.

Wong, M., Bost, K., Shin, N., Verisoomo, M., Maia, J., Monteiro, L., et al. (2011). Preschoolchildren ’s mentalrepresentations of attachment:antecedentsin their secure base be-haviors and maternal attachment scripts. Attachment & Human Development , 13 ,489 – 502.

33E. Hawkins et al. / Journal of Applied Developmental Psychology 39 (2015) 24 – 33