MATERNAL POSTPARTUM DEPRESSIVESYMPTOMS AND 4-MONTH

MOTHER–INFANT INTERACTION

Beatrice Beebe, PhDNew York State Psychiatric Institute

Frank Lachmann, PhDInstitute for the Psychoanalytic Study of

Subjectivity

Joseph Jaffe, MD,Sara Markese, PhD, and

Karen A. Buck, MANew York State Psychiatric Institute

Henian Chen, MD, PhDUniversity of South Florida

Patricia Cohen, PhD, Stanley Feldstein, PhD, andHoward Andrews, PhD

New York State Psychiatric Institute

Associations of 6-week maternal depressive symptoms (CES-D) with 4-monthmother�infant self- and interactive contingency patterns during face-to-faceplay were investigated in a community sample of 132 dyads. Self- and inter-

Beatrice Beebe, PhD, New York State Psychiatric Institute; Frank Lachmann, PhD, Institute for thePsychoanalytic Study of Subjectivity; Joseph Jaffe, MD, Sara Markese, PhD, Karen A. Buck, MA,Patricia Cohen, PhD, Stanley Feldstein, PhD, and Howard Andrews, PhD, New York StatePsychiatric Institute; Henian Chen, MD, PhD, College of Public Health, University of South Florida.

This research was supported by The National Institute of Mental Health (Grant MH56130), ThePsychoanalytic Fund, the Kohler Foundation, the Edward Aldwell Mother-Infant Research Fund,and the Los Angeles Infant Research and Psychoanalysis Fund. We thank our filming/coding team:Caroline Flaster, Donna Demetri-Friedman, Helen Demetriades, Nancy Freeman, Patricia Good-man, Michaela Hager-Budny, Elizabeth Helbraun, Allyson Hentel, Tammy Kaminer, Sandra TriggsKano, Limor Kaufman-Balamuth, Greg Kushnick, Lisa Marquette, Jill Putterman, Jane Roth,Shanee Stepakoff, and Lauren Ellman. We also thank our lab assistants: Claire Jaffe, DaniellaPolyak, Julia Reuben, Priscilla Caldwell, Jessica Lateck, Carol Scheik-Gamble, Danny Sims, JakeFreeman, Brianna Hailey, Alla Chavarga, Nidhi Parashar, Josianne Moise, Sarah Temech, HopeIgleheart, Yana Kuchirko, Elizabeth White, Greer Raggio, Kate Lieberman, Alina Pavlakos,Adrianne Lange, Kari Gray, Jennifer Lyne, Annee Ackerman, Fernanda Lucchese, Sarah Miller,Max Malitzky, Sam Marcus, Michael Klein, Kara Levin, Matthew Kirkpatrick, Lauren Cooper,Helen Weng, Iskra Smiljanic, and Christy Meyer. Finally, we thank our students and colleagues whocontributed: Sara Hahn-Burke, Nancy Freeman, Michael Ritter, Glenn Bromley, Robert Gallaghan,Naomi Cohen, Paulette Landesman, Tina Lupi, Jillian Miller, Alan Phelan, Danielle Phelan,Katherine Weinberg, Edward Tronick, Doris Silverman, Anni Bergman, Lin Reicher, GeorgeDowning, Estelle Shane, and the Monday afternoon seminar, who consulted on our findings.

Correspondence concerning this article should be addressed to Beatrice Beebe, PhD, New YorkState Psychiatric Institute #108, 1051 Riverside Drive, New York, NY 10032. E-mail:beebebe@pi.cpmc.columbia.edu

Psychoanalytic Psychology © 2012 American Psychological Association2012, Vol. 29, No. 4, 383–407 0736-9735/12/$12.00 DOI: 10.1037/a0029387

383

active contingency were defined as predictability within (self-contingency) andbetween (interactive contingency) two partners’ behavioral steams over time.Infant and mother attention (gaze), affect (facial, vocal), spatial orientation, andtouch behaviors were coded second-by-second from split-screen videotape, anda multimodal measure of facial�visual “engagement” was constructed. Withhigher depressive symptoms, the self-contingency of both partners was loweredin most modalities; there was a lowered ability to anticipate one’s own nextmove, which is metaphorically a “destabilization.” With higher depressivesymptoms, interactive contingency values were both heightened (in some mo-dalities) and lowered (in others), varying by partner, and consistent with anoptimum midrange model. Thus, interactive contingency patterns associatedwith depressive symptoms manifested as both heightened and dampened coor-dination with the partner. With higher depressive symptoms, interactive con-tingency showed the following patterns: (a) mother and infant reciprocal ori-entational sensitivity; (b) mother and infant reciprocal intermodaldiscordance—a lowered gaze coordination but heightened affective coordina-tion; and (c) “infant approach, mother withdraw”—as infants heightened, butmothers lowered, touch coordination with the partner’s touch. The analysis ofseparate modalities revealed striking, complex intermodal discordances whichwere forms of intrapersonal and dyadic conflict, with relevance for therapeuticintervention.

Keywords: mother–infant interaction, maternal depression, microanalysis

Our purpose is to bring to the attention of a clinical audience aspects of a larger empiricalstudy on maternal depression and mother�infant communication. Associations betweenmaternal depressive symptoms reported at 6 weeks postpartum and mother�infant face-to-face play at 4 months were examined in a community sample collected by Beebe et al.(2008). Maternal depression is of particular interest to clinicians because it is predictiveof childhood social and cognitive difficulties. Detailed second-by-second microanalysis ofvideotaped interactions identified a new range of subtle intrapersonal and interpersonaldifficulties in mother�infant communication that are predicted by maternal postpartumdepressive symptoms. This work enriches the psychoanalytic understanding of the con-sequences of maternal depression for mother�infant communication disturbances.

The study we describe examined a community sample, rather than a psychopathologysample, so that the findings have more general relevance.1 In the general population, theincidence of maternal postpartum depression is 10–20%, depending on the study (Camp-bell & Cohn, 1991; Horwitz, Briggs-Gowan, Storfer-Isser, & Carter, 2007; Dietz et al.,2007). Thus, in considering the relevance for adult treatment, the average patient (not theextremely disturbed patient) has at least a 10% to 20% chance of having had a mother withpostpartum depressive symptoms. A better understanding of the kinds of early commu-nication disturbances that are likely in this group can inform the adult clinician.

1 This paper is based on a prior publication: Beebe, B., Jaffe, J., Buck, K., Chen, H., Cohen,P., Feldstein, S., Andrews, H. (2008). Six-week postpartum maternal depressive symptoms and4-month mother�infant self- and interactive contingency. Infant Mental Health Journal, 29,442–471.

384 BEEBE ET AL.

There is a large body of psychoanalytic, psychiatric, and psychological literature,some of which will be reviewed in this paper, on maternal depression and its consequencesfor the developing infant. Infants of depressed mothers are at risk for social, emotional,and cognitive difficulty, whether depression is defined by clinician-based diagnosis orself-report (Gitlin & Pasnau, 1989; Murray & Cooper, 1997). Infants of depressed mothersshow less positive affect, less optimal social and object engagement, and greater likeli-hood of insecure attachment (see Campbell & Cohn, 1991; Field, 1995; Lyons-Ruth,Repacholi, McCleod, & Silva, 1991; Tronick, 1989; Zlochower & Cohn, 1996). Com-pared to controls, depressed mothers and their infants spend more time in negative statesand match negative states more than positive ones (Cohn, Campbell, Matias, & Hopkins,1990; Field, Healy, Goldstein, & Guthertz, 1990). Infants of depressed mothers haveself-regulatory disturbances, such as perinatal complications, suggesting difficulty frombirth; by 6 months, they have elevated heart rates and cortisol levels (Field, 1995; Murray& Cooper, 1997).

Many of the above studies examined maternal depression and mother�infant inter-active regulation using microanalysis of videotape, which was our approach, as well.However, these studies are inconsistent in their findings, perhaps because studies differwith respect to statistical approach, infant age, severity of maternal depression, andsocioeconomic status (SES) of the participants. Cohn and Tronick’s (1989) small, high-risk, depressed sample showed lowered interactive regulation in depressed dyads at 6–7months, whereas Cohn et al.’s (1990) middle-class depressed sample showed no differ-ences in interactive regulation at 2 months. Field, Healy, and LeBlanc’s (1989) low SESstudy showed lower interactive regulation in depressed dyads at 3 months, but Field etal.’s (1990) low SES study showed no difference in interactive regulation at 3 months.

In contrast to the emphasis on high-risk and low SES samples in the literature, weexamined a low-risk mother�infant community sample in our study. A unique aspect ofour approach was an examination of multiple communication modalities during face-to-face play, separately capturing attention, affect, orientation, and touch. We focused on anexamination of both self- and interactive regulation. Self- and interactive regulation arecuriously separate in the research literature. Interactive regulation is the main theme andpredicts social and cognitive outcomes (Bakeman, Adamson, Brown, & Eldridge, 1989;Cohn, Campbell, Matias, & Hopkins, 1990; Cohn & Tronick, 1988; Jaffe, Beebe,Feldstein, Crown, & Jasnow, 2001; Lewis & Feiring, 1989; Leyendecker, Lamb, Fracasso,Scholmerich, & Larson, 1997; Malatesta, Culver, Rich, & Shepard, 1989; Martin, 1981;Tronick, 1989). However, it is essential to integrate both self- and interactive forms ofregulation. Each person must monitor the partner and regulate his or her own inner state.Either or both forms of regulation may be misregulated in each partner (Gianino &Tronick, 1988; Beebe et al., 2007; Beebe et al., 2010). Rather than locating the source ofdifficulty in one or the other partner by examining, for example, infant self-regulatorydifficulties or maternal insensitivity, we pry apart the relative contributions of self- andinteractive regulation of both partners. If the relative salience of self- versus interactivedifficulties varies with different clinical pictures, such as depression, therapeutic inter-vention for mother�infant dyads can be more finely focused.

Self-regulation is important, but we do not know exactly how it is accomplished (Fox,1994; Thompson, 1994). There are many definitions of self-regulation, such as activationor dampening of degree of arousal and capacity to down-regulate negative affect (Kopp,1989; Stifter & Spinrad, 2002; Thompson, 1994). Another definition is based on theexamination of how infants cope under stress, such as in the still-face experiment

385MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

(Weinberg, Tronick, Cohn, & Olson, 1999). However, any behavioral pattern can beviewed as simultaneously self-regulatory and communicative (Overton, 1998).

Our definition of regulation is predictability of behavior over time, or contingency,(Gottman, 1981; Tronick, 1989). Infants detect contingencies from birth, and by 4 months,are adept at discriminating degrees thereof (Bigelow, 1998; Watson, 1985). Interactiveregulation is defined as the predictability of each partner’s behavior from that of the otherover time (lagged cross-correlation). We adopt the term interactive contingency. Self-regulation is defined narrowly as predictability of a person’s behavior over time (auto-correlation). Because we study the dyadic face-to-face exchange, our definition of self-regulation is limited to the individual’s behavior in the presence of a particular partner. Weadopt the term self-contingency to distinguish our meaning of self-regulation from othermeanings. Temperament approaches to self-regulation are outside our scope (Fox, 1994;Kopp, 1989).

The Optimum Midrange Model of Interactive Regulation

Competing hypotheses in the literature suggest that (a) high interpersonal contingenciesare optimal for communication (Chapple, 1970) or that (b) high contingencies indexcommunicative stress (Gottman, 1981). However, using vocal rhythm interactive contin-gency in 4-month face-to-face interactions to predict 12-month infant attachment, Jaffe etal. (2001) and Beebe et al. (2000) found that midrange degrees of contingency wereoptimal for security and that higher and lower degrees of contingency predicted insecurity.Similarly, midrange (vs. higher or lower) degrees of 4-month vocal rhythm contingencyin this data set were found to predict secure (vs. insecure) 4-year attachment representa-tions (Markese, Beebe, Jaffe, & Feldstein, 2008). In relation to insecure attachment, weinterpret (a) high interactive contingency as “vigilance,” a dyadic effort to create moremoment-to-moment predictability in the interaction, and (b) low interactive contingencyas inhibition or withdrawal.

Other research converges on an optimum midrange model of mother�infant interac-tive contingency. In predicting 1-year attachment from 3- and 9-month maternal measures,Lewis and Feiring (1989) found that 3-month secure infants showed midrange sociabilityand object play and that mothers of secure infants showed midrange responsiveness.Measuring maternal (but not infant) facial contingency, Malatesta et al. (1989) found thathigh maternal contingency predicted avoidant attachment, whereas midrange maternalcontingency predicted secure attachment. Sander (1995) reported that relatively more“loosely coupled” (midrange) mother�infant dyads were more resilient, whereas in moretightly coupled dyads, any disruption might shatter the system. Roe, Roe, Drivas, andBronstein (1990) found that only moderately talkative mothers both initiated vocalizationsand allowed their infants to do the same; these infants vocalized preferentially to theirmothers (vs. strangers), and at 3 and 5 years exhibited higher cognitive scores. Across 4,8, and 12 months, Leyendecker et al. (1997) found that secure dyads showed midrangedegrees of contingency, whereas insecure dyads received higher or lower scores. Beebe etal. (2010), analyzing attention, affect, spatial orientation, and touch, found that insecuritywas predicted by 4-month mother and infant self- and interactive contingencies that wereheightened in some modalities and lowered in others. Hane, Feldstein, and Dernetz (2003)found midrange vocal coordination was related to higher maternal sensitivity at 4 months.Beebe et al. (2007) found that 6-week maternal self-criticism was associated at 4 months

386 BEEBE ET AL.

with heightened maternal interactive contingencies in some modalities and lowered inothers.

Although the above studies address degrees of interactive contingency, few have usedthe auto-correlation measure of self-contingency as a variable in its own right. (Self-contingency indexes self-predictability, or the stability of one’s own rhythms of behav-ing.) Instead, auto-correlation has been statistically controlled and removed (see Gottman,1981). In research on adult conversation, Warner (1992) was unusual in integrating“within-actor” and “between-partner” contingencies. Beebe et al. (2007, 2008) found thatinfant and mother self-contingency were lowered in contexts of maternal distress.

Approach of the Study

In the present study we investigated whether specific modalities of the face-to-faceexchange (attention, affect, orientation and touch) may differ in patterns of self- andinteractive contingency, thus defining different aspects of disturbance associated withmaternal depression. Consistent with the literature, we also investigated associations ofmaternal depression with qualitative features of behavior, such as incidence of maternalintrusive touch. We constructed a multimodal measure of facial�visual “engagement,”(Beebe & Gerstman, 1980) as well, which is consistent with the literature (see “monadicphases” of Tronick, Als, & Brazelton, 1980).

Our central approach was the separate analysis of modalities (see Weinberg &Tronick, 1994) through ordinalized behavioral scales of mother and infant interac-tions: (a) attention— gaze at or away from the partner’s face, (b) affect—facial andvocal, (c) touch, and (d) spatial orientation. Such specificity might yield moreinformation and be more clinically useful. It is also easier to observe specificmodalities than a constructed variable.

Mother and infant gaze and facial affect are central modalities of face-to-faceexchange (Stern, 1971; Tronick, 1989). Infant vocal affect, particularly vocal distress,is an important form of infant communication (Mumme, Fernald, & Herrera, 1996).Infant touch is important in infant self-regulation and self-soothing (Tronick, 1989).Maternal touch is a central but less studied modality and was examined in relation toinfant measures of vocal affect, touch, and engagement (see Malphurs, Raag, Field,Pickens, & Pelaez-Nogueras, 1996; Stepakoff, Beebe, & Jaffe, 2000). We exploredmaternal spatial orientation (sitting upright, leaning forward, or looming in close tothe infant’s face) in relation to infant head orientation, from en face to arch (Beebe &Stern, 1977; Demetriades, 2003; Kushnick, 2002; Weinberg & Tronick, 1998). When-ever possible, we attempted to pair the same modality for mother and infant: gaze,facial affect, engagement, touch, and orientation. However, we also examined infantvocal affect in relation to mother facial affect as a second way of exploring theinfant’s emotional response to the mother’s face. Following Tronick’s (1989) hypoth-esis that touch has a self-comforting function and investigating whether maternaldepression affects the likelihood that infant touch modulates infant vocal distress, weexplored one pattern within the infant: infant touch and infant vocal affect.

This approach allowed us to understand better what the infant’s experience might be.It allowed us to “unpack” interactions which might be coded as interactive errors (Tronick,1989) or communication errors (Lyons-Ruth, Bronfman, & Parsons, 1999). It identifiedintermodal discrepancies, which may be intrapersonal or dyadic. For example, we wereable to identify difficulties in the regulation of attention, separate from the regulation of

387MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

affect. Interactive contingency of gaze examines the degree to which each partner followsthe direction of the other’s visual attention, on and off the partner’s face. Interactivecontingency of affect was examined as two patterns of affective “mirroring,” the degreeto which partners share direction of affective change: (a) facial affect mirroring and (b)cross-modal maternal facial affect�infant vocal affect mirroring. The coordination ofinfant touch and maternal touch examines the degree to which mothers touch moreaffectionately as infants touch more (and vice versa) and whether the relative affectionatequality of maternal touch affects infant likelihood of touch (and vice versa). In thiscontext, we can examine whether maternal touch may be a compensatory modality whenattention or affect were problematic. Maternal contingent spatial coordination with infanthead orientation measures mutual approach and withdrawal patterns (mother’s likelihoodof leaning forward or looming in as the infant reorients toward enface, and reciprocally,mother’s likelihood of sitting back in an upright posture as the infant orients away). Infanthead orientation coordination with maternal spatial orientation measures the infant’slikelihood of being involved in a maternal approach�infant withdrawal pattern; in otherwords, it measures the infant’s likelihood of arching away as mother looms in (in Beebeand Stern’s [1977] “chase and dodge” pattern) and reciprocally, the infant’s likelihood ofreorienting as mother sits back.

Hypotheses

Hypothesis 1: Building on Jaffe et al. (2001), where both heightened and loweredinteractive contingencies were associated with insecure attachment and midrangevalues were associated with secure attachment, we hypothesized that maternal depres-sion would bias the mother�infant communication system toward both heightenedself- and interactive contingency in some modalities, and lowered in others. We alsoexplored whether maternal depression is associated with differences in qualitativefeatures of mother and infant behavior, such as intrusive maternal touch.

Hypothesis 2: Because low scores of maternal depression may be ambiguous (Shedler,Mayman, & Manis, 1993), we used nonlinear analyses (using depression as a quadraticmeasure) to explore whether contingency patterns where depression scores movetoward the low pole may look similar to contingency patterns where depression scoresmove toward the high pole.

Hypothesis 1 proposes an “optimum midrange” model of contingency: Both low andhigh contingencies, in different communication modalities, may be associated with higherdepression. This approach is a “conceptual” midrange model of contingency. It does notuse nonlinear models. Instead, if the depression subgroup shows contingencies that areboth higher than controls in some modalities and lower than controls in other modalities,then the control group is in this sense operating more in the midrange and less at theextremes.

Low scores of maternal depression may be ambiguous (Shedler, Mayman, & Manis,1993). Hypothesis 2 proposes that mothers scoring at the low and high poles of depressionmay have similar altered contingencies when compared to mothers scoring midrange indepression. This approach constitutes another type of optimum midrange model ofdepression, using a statistically nonlinear approach to depression.

388 BEEBE ET AL.

Method

Participants

Recruitment

Within 24 hours of delivering a healthy, full-term, singleton infant without major com-plications, 152 primiparous mothers were recruited from Columbia University MedicalCenter for a study of infant social development (1992 – 1998).2 Subjects were 18 years orolder, married (or living with partner), and had a home telephone. At 6 weeks, all 152mothers were telephoned and given the Center for Epidemiological Studies DepressionScale (CES-D). At 4 months, 132 (of the 152) mothers and infants were videotaped in thelab. No differences were found in ethnicity, education, or infant gender between the 132participants and the 152 recruited.

Demographic Description

Mothers were 53.0% White, 28.0% Hispanic, 17.5% Black, and 1.5% Asian; they werewell-educated (3.8% without High School diploma, 8.3% without college, 28.8% somecollege, 59.1% college degree or more); their mean age was 29, SD � 6.5, range �18–45. Of 132 infants, 58 were female.

Procedure

Scheduling of the lab visit took into account infants’ eating and sleeping patterns.Mothers were seated opposite the infant, who sat in an infant seat on a table. Motherswere instructed to play with their infants as they would at home, but without the useof toys, for approximately 10 minutes. Two video cameras generated a split-screenview of the interaction. Following the filming, mothers again filled out the CES-Dscale in the lab.

Measurement of Maternal Depressive Symptoms

The CES-D (Radloff, 1977) measures self-reported current nonspecific distress,“depressive symptoms,” not clinically diagnosed depression (Campbell & Cohn,1991). A score of 16� at 6 weeks defined a hi-CES-D group (n � 34, 25.8% of thesample). Controls (n � 98) had CES-D values of 0 – 15.3 We used 6-week (vs.4-month) CES-D4 scores to see whether depressive symptoms at this early stageaffected the dyad 2-1/2 months later; 6-week and 4-month CES-D scores werecorrelated, r � .47, p � .01.

2 This data set is entirely different from that reported in Jaffe et al. (2001).3 Preliminary analyses explored CES-D at 6 weeks and at 4 months as continuous and

categorical (16�) variables. We chose 6-week CES-D to see if maternal depressive symptoms at thisearly stage might affect the dyad almost 2 months later. Six-week symptoms (16�) were higher thanthose at 4 months, and preliminary analyses showed 6-week CES-D to be more discriminating inmother and infant facial affect and gaze regression models. We chose categorical analyses because6-week CES-D (16�) is a widely-accepted clinical cut-off. Furthermore, in preliminary gaze andfacial affect regression models, categorical analyses were significant whereas continuous ones werenot.

4 Missing values for 6-week CES-D (n � 32) were estimated from 8-week CES-D values (n �27) available on a small subset assessed at 8 rather than 6 weeks, or from 4-month CES-D values(n � 5) corrected by the 6-week group mean, using regression analyses to impute missing scores.

389MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Behavioral Coding

We examined separate communication modalities of attention, affect, spatial orien-tation, and touch, as well as a composite measure of facial�visual engagement. Thefirst 2-1/2 uninterrupted continuous play minutes of videotaped mother�infant inter-action were coded on a 1-s time base by coders blind to CES-D status, using Tronickand Weinberg’s (1990) timing rules. Behavioral codes were used to create ordinalizedscales for data analysis (required by time-series techniques). Definitions of behavioralscales follow. Categories were lumped (indicated by /) when necessary to avoid smallfrequencies. Gaze: on�off partner’s face; for infant, gaze-off was further divided intogaze at object versus off subject. The following scales are ordinalized from high tolow. Mother facial affect: mock surprise, smile 3, smile 2, smile 1, “oh” face, positiveattention, neutral, “woe” face, negative face (frown, grimace, compressed lips). Infantfacial affect: high positive, low positive, interest/ neutral, mild negative (frown,grimace), negative (precry, cry-face). Infant vocal affect: positive/ neutral, none,fuss/whimper, angry protest/cry. Mother spatial orientation: sitting upright, leaningforward, looming in. Infant head orientation: en face, en face � head down, 30 – 60avert, 30 – 60 avert � head down, 60 –90 avert, arch. A dyadic code, “chase anddodge,” was defined as a minimum 2-s sequence in which infant averts head 30degrees or more from vis-a-vis and mother moves head or body in the direction of theinfant’s movement (Kushnick, 2002). Infant touch: none, 1, or 2 � of the followingbehaviors within one second: touch/ suck own skin, touch mother, touch object(Hentel, Beebe, & Jaffe, 2000). Mother touch (examples in parentheses): affectionate(stroke, kiss), static (hold, provide finger for infant to hold), playful (tap, tickle), none,caregive, jiggle/ bounce, infant-directed oral touch, object-mediated, centripetal(touch body center: face, body, head), rough (scratch, push, pinch), high intensity/intrusive. Codes “affectionate” through “jiggle/bounce” occur in locations of hands,arms, feet, legs; thereafter, touch may occur in locations of infant face, head, orcentral body (Stepakoff et al., 2000). Multimodal facial�visual engagement scaleswere also constructed. Infant engagement was ordinalized from a high of “highpositive engagement” to a low of “cry.” Mother engagement was ordinalized from“mock surprise” to “neutral/negative off” (Beebe & Gerstman, 1980).

Reliability estimates of ordinalized scales were assessed in 30 randomly selecteddyads (in 3 waves to prevent coder “drift”). Mean kappas per variable follow: forinfants, gaze � .80, facial affect � .78, vocal affect � .89, touch � .75, headorientation � .71; for mothers, gaze � .83, facial affect � .68, touch � .90, spatialorientation � .89; dyadic maternal “chase” and infant “dodge” � .89.

Using ordinalized scales, eight mother�infant “modality pairings” and a ninthintrapersonal infant pairing were generated for data analysis: (a) infant gaze�mothergaze, (b) infant facial affect�mother facial affect, (c) infant vocal affect�motherfacial affect, (d) infant engagement�mother engagement, (e) infant engagement-�mother touch, (f) infant vocal affect�mother touch, (g) infant touch�mother touch,(h) infant head orientation�mother spatial orientation, and (i) infant touch�infantvocal affect.

Self- and Interactive Contingency

Each person’s behavior is affected both by one’s own prior behavior (self-contingency)and by that of the partner (interactive contingency; Thomas & Malone, 1979; Thomas &

390 BEEBE ET AL.

Martin, 1976). Time-series methods are designed to partition these two sources ofvariance. Because any behavior pattern in a face-to-face encounter may participatesimultaneously in self- and interactive contingency functions, every behavior must beassessed for both functions. Self- and interactive contingency are defined, using lagcorrelations, as predictability within (autocorrelation) and between (lagged cross-correlation) two partners’ behavioral streams over time. Multilevel time-series modelswere employed.5 In these models, the prior 3 seconds were used as a weighted lag topredict the current second, in an iterative process.6 Thus, self-contingency assesses thedegree to which the prior 3 seconds predict the current second, within the individual’s ownbehavioral stream. Interactive contingency assesses the degree to which the prior 3seconds of the partner’s behavior predicts the current second of the individual’s behavioralstream. For further details, see Chen and Cohen (2006), Beebe et al. (2008).

Linear equations analyzed effects of depressive symptoms categorically (presence/absence of depression), and results were interpreted as characterizing the hi-CES-D (16�)group versus controls. Nonlinear equations analyzed effects of depression as a continuousvariable, and results were interpreted as characterizing the high and low poles of CES-Dscores, compared to dyads where mothers scored midrange in CES-D. The 6-week CES-Ddistribution was skewed, skewness � 1.20, SE .21, mean � 12.34, SD � 8.41, me-dian � 10.46, mode � 9.00. Representation of subjects at the high and low ends wasadequate for our conclusions, but there was more discrimination at the high end. Eightsubjects endorsed 0, 1, or 2 CES-D items. Significant standardized estimates (�) of thelinear and nonlinear effects of 6-week CES-D on 4-month self- and interactive contin-gency for the 9 modality pairings are presented in Tables 1 and 2, respectively.

5 Multilevel models are designed to address patterns over time (here, the course of behaviorsecond-by-second). They have more power than traditional time-series techniques, take into accounterror structures, and estimate individual effects with empirical Bayesian (maximum likelihood)techniques (rather than Ordinary Least Squares), which take into account prior distributions (Singer,1998). Because the prior probability of error is greatest for the extreme parameters, this methodtends to pull in such extremes. Advantages of this approach include the following: (a) multipletime-series (in our case, self- and interactive contingency) can be modeled simultaneously, (b) anaverage effect of key parameters (e.g. infant behavior contingent on mother behavior) is estimatedfor the group and allows the investigator to ask how that group mean changes in the context of otherfactors such as maternal depression, (c) control variables (such as demographics) and their condi-tional effects can be included as necessary, (d) nonlinear relations can be examined in the sameanalyses, and (e) more appropriate statistical model assumptions are made (see Beebe et al., 2007,2008; Chen and Cohen, 2006).

6 Preliminary analyses estimated the number of seconds over which lagged effects weresignificant. For each dependent variable, measures of prior self or partner behavior, “laggedvariables,” were computed as a weighted average of the recent prior seconds. Typically, the prior 3seconds sufficed to account for lagged effects on subsequent behavior. Across modality pairings,Mother was significant at 2-3 lags (seconds); longer lags were not significant. Significant Infant lagsvaried: for self-contingency, 2 lags (touch), 3 (face, gaze), and 4 (vocal affect); for interactivecontingency, 3 (M face3 I vocal affect), 5 (M facial affect3 I facial affect), 6 (M gaze3 I gaze).Although some modality pairings showed lags longer than 3 seconds, the amount of varianceaccounted for was very small. No more than 3 lags were used in any weighted mean lag in order tomaintain consistent sample size.

391MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Table 1Maternal 6-Week Depressive Symptoms (CES-D 16�) and 4-Month Self- andInteractive Contingency: Linear Effects

INFANT MOTHERβ SE β β SE β

Pattern (1) Infant Gaze Mother GazeCESD .398* .165 CESD .154 .129

I→I/CESD -.227* .109 M→M/CESD .293 .197M→I/CESD -.165* .083 I→M/CESD -.363* .152

Pattern (2) Infant Facial Affect Mother Facial AffectCESD .143 .401 CESD .248 .353

I→I/CESD .006 .015 M→M/CESD -.047** .016

M→I/CESD .017 .019 I→M/CESD .039** .015

Pattern (3) Infant Vocal Affect Mother Facial Affect

CESD -.009 .063 CESD .339 .298

I→I/CESD .012 .015 M→M/CESD -.032† .018

M→I/CESD .0001 .001 I→M/CESD .581* .285

Pattern (4) Infant Engagement Mother Engagement

CESD .218 .170 CESD .105 .077

I→I/CESD -.033* .014 M→M/CESD -.028 .020

M→I/CESD -.017 .036 I→M/CESD .005 .009

Pattern (5) Infant Engagement Mother Touch

CESD .236 .166 CESD -.065 .074

I→I/CESD -.032* .014 M→M/CESD -.064*** .012

M→I/CESD -.038† .023 I→M/CESD -.008 .008

Pattern (6) Infant Vocal Affect Mother Touch

CESD .001 .017 CESD -.061 .078

I→I/CESD -.021 .016 M→M/CESD -.055*** .012

M→I/CESD -.003 .003 I→M/CESD -.090 .058

392 BEEBE ET AL.

Results and Discussion

We studied maternal 6-week self-reported depressive symptoms (CES-D) rather thanclinically diagnosed depression. Thus, we used the term depression to refer toself-reported depression. The symptoms reported indicated current but nonspecificdistress, such as difficulty sleeping or eating, crying spells, fearfulness, feeling lonely,and feeling like a failure. In a large, diverse community sample, we hoped for a moregeneralizable sample than those of many studies of maternal depression which focuson a specific clinical or demographic subgroup. However, the sample was highlyeducated and with a high incidence of depressive symptoms (25%) for a communitysample (Campbell & Cohn, 1991). Higher depressive scores were associated withlower education levels.

Table 1 (continued)

Pattern (7) Infant Touch Mother Touch

CESD .006 .012 CESD -.075 .079

I→I/CESD -.013 .011 M→M/CESD -.071*** .012

M→I/CESD .006* .003 I→M/CESD -.054 .062

Pattern (8) Infant Head Orientation Mother Spatial Orientation

CESD .096 .051 CESD -.014 .029

I→I/CESD -.046** .015 M→M/CESD -.059*** .013

M→I/CESD -.006 .033 I→MCESD -.018** .006

INFANT INFANT

Pattern (9) Infant Vocal Affect Infant Touch

CESD .004 .018 CESD .005 .013

I→I/CESD -.003 .016 I→I/CESD -.007 .011

I→I/CESD -.021 .014 I→I/CESD -.008 .014

Note. Estimated fixed effects of maternal depression (CES-D 16�) and interaction with M 3 M, I 3 M (orI 3 I, M 3 I) based on the “basic models,” controlling for demographic variables and infant gender.CES-D � Center for Epidemiological Studies Depression Scale.CESD � main effect of CESD on the mean of the video code.I 3 I/CESD � effect of CESD on infant self-contingency.M 3 I/CESD � effect of CESD on infant interactive contingency.M 3 I � Mother behavior (lagged) predicts Infant behavior (current second).I 3 M � Infant behavior (lagged) predicts Mother behavior (current second).All parameter entries are maximum likelihood estimates fitted using SAS PROC MIXED.Maternal depression at 6 weeks.CESD � 16 � coded 1; CESD � 16 coded 0.Negative signs indicate lower mean values of variable in question (main effects) or lower estimates of self- andinteractive contingency.To illustrate the picture of higher depression in pattern (1), infant gaze�mother gaze, infants of hi-CES-D (16�)mothers showed lowered gaze self-contingency, compared to infants of control mothers (CES-D � 15) (Table 1,I 3 I/CESD � � �.227, p � .05).† p � .10. � p � .05. �� p � .01. ��� p � .001.

393MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Table 2Maternal 6-Week Depressive Symptoms (CES-D) and 4-Month Self- and InteractiveContingency: Nonlinear Effects

INFANT MOTHERΒ SE β β SE β

Pattern (1) Infant Gaze Mother GazeCESD² .008 .054 CESD² -.057 .043

I→I 3.570*** .062 M→M 2.725*** .097I→I/CESD² .022 .039 M→M/CESD² -.169** .058

M→I .494*** .140 I→M .568*** .084M→I/CESD² .104 .090 I→M/CESD² .023 .051

Pattern (2) Infant Facial Affect Mother Facial AffectCESD² -.060 .131 CESD² .038 .112

I→I .651*** .010 M→M .562*** .009

I→I/CESD² .003 .005 M→M/CESD² -.008 .005

M→I .046*** .010 I→M .116*** .009

M→I/CESD² .003 .006 I→M/CESD² .017** .005

Pattern (3) Infant Vocal Affect Mother Facial Affect

CESD² -.053** .020 CESD² .029 .093

I→I .662*** .010 M→M .632*** .008

I→I/CESD² -.017*** .005 M→M/CESD² -.005 .005

M→I .001* .001 I→M 1.213*** .160

M→I/CESD² .001** .0003 I→M/CESD² .139 .097

Pattern (4) Infant Engagement Mother Engagement

CESD² .069 .054 CESD² -.023 .024

I→I .688*** .008 M→M .484*** .012

I→I/CESD² -.009* .005 M→M/CESD² -.014* .006

M→I .041* .020 I→M .056*** .006

M→I/CESD² .024* .012 I→M/CESD² .007* .003

394 BEEBE ET AL.

Table 2 (continued)

Pattern (5) Infant Engagement Mother Touch

CESD² .079 .053 CESD² -.018 .024

I→I .698*** .008 M→M .738*** .007

I→I/CESD² -.006 .005 M→M/CESD² .001 .004

M→I .009 .013 I→M .006 .005

M→I/CESD² .004 .007 I→M/CESD² -.002 .003

Pattern (6) Infant Vocal Affect Mother Touch

CESD² .009 .005 CESD² -.022 .025

I→I .699*** .013 M→M .738*** .007

I→I/CESD² -.013** .005 M→M/CESD² -.00064 .004

M→I .001 .002 I→M .082** .032

M→I/CESD² .001 .001 I→M/CESD² -.012 .019

Pattern (7) Infant Touch Mother Touch

CESD² .004 .004 CESD² -.017 .027

I→I .707*** .006 M→M .730*** .007

I→I/CESD² .006 .004 M→M/CESD² .002 .004

M→I .001 .002 I→M .194*** .035

M→I/CESD² -.002* .001 I→M/CESD² -.114*** .022

Pattern (8) Infant Head Mother Spatial

CESD² .037 .047 CESD² .011 .301

I→I .642*** <.001 M→M .703*** <.001

I→I/CESD² -.001 .864 M→M/CESD² .022*** <.001

M→I -.007 .723 I→M -.005 .144

M→I/CESD² .014 .237 I→M/CESD² -.006 .018

M→I/CESD² .013 .010 I→M/CESD² -.004 .008

(table continues)

395MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Contingency was measured as predictability over time. Infants perceive temporalsequences and degrees of contingencies, and quickly come to expect when events willoccur (DeCasper & Carstens, 1981; Haith, Hazan, & Goodman, 1988; Watson, 1985).Degree of self-contingency generates expectancies of predictability or stability of one’sown behavior. Degree of interactive contingency generates expectancies of how predict-ably the partner changes in relation to one’s own changes, and vice versa, metaphoricallyorganizing expectancies of “how I affect you,” and “how you affect me.”

In previous work, we interpreted high coordination as an effort to create morepredictability in the dyad under conditions of novelty, challenge, or threat; we interpretedlow coordination as withdrawal or inhibition (Jaffe et al., 2001). Heightened and loweredinteractive contingency were translated into the metaphors of “vigilant” and “withdrawn(inhibited),” respectively. Vigilance for the social signals of the other is an importantaspect of social intelligence likely to be an evolutionary advantage in conditions ofuncertainty or threat, and it is likely to be accompanied by emotional activation (Ohman,2002). A partner’s lowered interactive coordination is interpreted as a “withdrawal” or“distancing” from the individual, compromising the individual’s interactive agency. Theindividual is thereby less able to anticipate the consequences of his own actions for thepartner. Lowered self-contingency is translated into the metaphor of “self-destabilization,”a lowered ability to anticipate one’s own next move.

Table 2 (continued)

INFANT INFANT

Pattern (9) Infant Vocal Affect Infant Touch

CESD² .012 .005 CESD² .003 .004

IVoc→IVoc .694*** .012 ITch→ITch .774*** .006IVoc→IVoc/CESD² -.016** .005 ITch→ITch/CESD² .005 .004

ITch→IVoc -.004 .012 IVoc→ITch .025*** .008Itch→IVoc/CESD² .001* .005 IVoc→ITch/CESD² .001 .005

Note. Standardized estimated fixed nonlinear (NL) effects (�) of maternal depression and interaction with M3M, I 3 M (or I 3 I, M 3 I) based on the “basic models,” controlling for demographic variables and infantgender.SE � � Standard Error of the Beta. CES-D � Center for Epidemiological Studies Depression Scale.CESD � maternal depression at 6 weeks (the scale was centered by its mean).I 3 M � Infant behavior (lagged) predicts Mother behavior (current second).M 3 I � Mother behavior (lagged) predicts Infant behavior (current second).CESD2 � NL main effect of CESD on video codes.I 3 I/CESD2 � NL effect of CESD on infant self-contingency.M 3 I/CESD2 � NL effect of CESD on infant interactive contingency.All parameter entries are maximum likelihood estimates fitted using SAS GLIMMIX Macro. Bothdepression as a linear effect, and its linear conditional effect on self- and interactive contingency, wereincluded in all NL models but are not presented here. However we do include estimates of self-andinteractive contingency to facilitate interpretation of NL conditional effects of CESD2 on these estimates.Since only nonlinear results are of interest in this table, significant nonlinear results are bolded. Negativesigns indicate lower mean values of the variable in question, or lower estimates of self- and interactivecontingency. To illustrate the picture of higher depression in pattern (1), infant gaze� mother gaze, withhigher CES-D, maternal gaze self-contingency was lowered (Table 2, M3M/CESD2 � � �.169, p �.01),compared to mothers scoring midrange.� p � .05. �� p � .01. ��� p � .001.

396 BEEBE ET AL.

Contingency processes are essential to social communication. Infants are highlysensitive to the ways in which their behaviors are contingently responded to (Hains &Muir, 1996; Haith et al., 1988; Murray & Trevarthen, 1985; Tarabulsy et al., 1996;Tamis-LeMonda, Bornstein, & Baumwell, 2001). The prediction of events and thecreation of expectancies about their time course is a foundation of the infant’s commu-nicative capacity, facilitating information processing, memory, and the procedural repre-sentation of interpersonal events (Fagen, Morrongiello, Rovee-Collier, & Gekoski, 1984;Feldman, 2007; Hay, 1997; Jaffe et al., 2001; Lewis & Goldberg, 1969; Murray & Cooper,1997; Tronick, 1989). Although rarely studied together, both self- and interactive con-tingency processes generate expectancies essential to communication. Behavioral rhythmsand contingencies, both within and between individuals, provide the ongoing temporalinformation necessary to predict and coordinate with one’s partner so that each cananticipate how the self and the other will proceed (Feldman, 2006; Tronick, 1989; Warner,1992). Thus, the findings presented below of 4-month mother and infant self- andinteractive contingency patterns altered by postpartum maternal depression define oneprocess by which maternal depressive symptoms can be transmitted to the infant and alterthe trajectory of development. In the communication patterns documented below, infantexpectancies of self- and interactive contingency were different in infants whose mothersreported postpartum depressive symptoms, compared to infants whose mothers did not.

The Optimum Midrange Model of Self- and Interactive Regulation

We hypothesized that, across the system of both partners and all communication modal-ities, 6-week maternal depression would bias the 4-month communication system towardboth heightened contingency values in some modalities, and lowered contingency valuesin others. Findings for interactive contingency were consistent with our hypothesis, as wedetail below. Thus, interactive contingency can be “excessive” (vigilant) or “insufficient”(withdrawn), consistent with Jaffe et al. (2001), who documented that excessive andinsufficient degrees of vocally rhythmic interactive contingency predicted insecure attach-ment.

However, self-contingency did not fit our hypothesis. With maternal depression,self-contingency was lowered for mother and infant. This is an important finding,consistent with Beebe et al. (2007), who documented similar lowered self-contingencieswith maternal self-criticism. Lower self-contingency makes it harder for each individualto anticipate his or her own, as well as the partner’s, ongoing behavioral stream. Withlower self-contingency, the cues by which one knows oneself are less predictable, and oneknows less what to expect of oneself, yielding a decreased sense of coherence over time(personal communication, D. Silverman, November 18, 2004).

Maternal 6-Week Depressive Symptoms and 4-Month Interactive Contingency

We present here the results of analyses in which mothers who endorsed depressivesymptoms are compared to those who did not, presented in Tables 1 and 2, andsummarized in Figure 1. Negative signs indicate lowered contingency estimates withhigher CES-D. Nonlinear findings are interpreted here only for the high end of increasingCES-D scores (see below for the low end).

In Figure 1, arrows which curve from infant to mother represent mother interactivecontingency (vice versa for infant); arrows which curve back into one partner’sbehavior represent self-contingency. The notation I3 M for interactive contingencyindicates that infant behavior in the prior few seconds predicted maternal behavior in

397MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Figure 1. Maternal 6-week CES-D and Altered Self- and Interactive Contingency. Arrows whichcurve from infant to mother represent mother interactive contingency (vice-versa for infant);arrows which curve back into one partner’s behavior represent self-contingency. The notation I3M for interactive contingency indicates that lagged infant behavior in the prior few secondspredicted maternal behavior in the current moment Mother coordinates with infant; M 3 I �lagged mother behavior predicts infant behavior: Infant coordinates with mother. Summary of alllinear (L) (Table 1) and nonlinear (NL) (Table 2) findings. NL findings for high end of depressiononly. L � NL indicates effects that are both linear and nonlinear. Gz � Gaze; Face � FacialAffect; Vc A � Vocal Affect; Eng � Engagement; Tch � Touch; I Head � Infant HeadOrientation; M Sptl � Mother Spatial Orientation. A broken arrow means that if mother isdepressed (CES-D 16�) vs. non-depressed, contingency is lower. An unbroken arrow means thatif mother is depressed (CES-D 16�) vs. non-depressed, contingency is higher. No arrow meansthere is no significant conditional effect of maternal depression on contingency. Numbers (1)through (9) indicate the 9 modality pairings examined, grouped by domains: attention (pattern 1),affect (2, 3, 4), mother touch (5, 6, 7), spatial orientation (8), and infant intrapersonal (9).a self-contingency; b interactive contingency; c Across the group, there was a significant negativeassociation of lagged I Head 3 M Spatial. That is, as infants moved from the head orientationposition of en face (highest code) toward 90° aversion or arch (lowest code), mothers moved fromloom (lowest code) to forward or upright (highest code). Reciprocally, as infants moved from 90°aversion or arch toward en face, mothers moved from upright toward forward or loom. Thedepressed (vs. non-depressed) group showed a significant increase in this negative association,and thus, an even stronger negative correlation. This association is bolded in the figure to showthat the association is strengthened in the depressed group; d As noted in the results, there was asignificant nonlinear, as well as linear, effect of CES-D on mother spatial self-contingency. Alinear decrease in self-contingency was modified by a nonlinear increase. With increasingdepression, mothers tended to have lower contingency, the linear effect. However, as depressionincreased, the linear decline flattened out, and there was no further decline. Thus, the primaryeffect is the linear decrease in self-contingency.

398 BEEBE ET AL.

the current moment: Mother coordinates with infant (vice versa for M3 I). Brokenarrows represent findings where depressive symptoms generated lowered contingencyvalues; unbroken arrows represent heightened contingency values (compared tocontrols for linear equations and compared to dyads where mothers scored midrangefor nonlinear equations). Absence of arrows represents no effects of CES-D. Bracketsin the margins of Figure 1 demarcate five domains: attention (pairing 1), affect(pairing 2, 3, 4), mother touch in relation to infant behavior (5, 6, 7), orientation (8),and infant intrapersonal vocal affect and touch (9).

Gaze

Depressed (vs. control) mothers and their infants both lowered their coordination ofpatterns of looking at and away from the partner’s face, a form of mutual interactivewithdrawal. However, each partner looked at the other’s face for a greater percentage ofthe time, compared to control dyads, as if vigilantly watchful. Moreover, both depressedmothers and their infants lowered their gaze self-contingency. Thus, both mothers andinfants were less predictable to themselves and to the partner. Looking more, butcoordinating less, is interpreted as a form of “noninteractive looking.” Attention to thepartner’s direction of gaze at and away from one’s own face is a critical foundation of theface-to-face encounter (Brazelton, Koslowski, & Main, 1974; Tronick, 1989) and iscompromised in depressed mothers and their infants.

Increased looking at the partner is an unusual finding, generally inconsistent with thedepression literature which finds less attending behavior (Cohn et al., 1990), particularlyin infants of low SES samples (see Field, 1995), who are likely to be different from ourwell-educated community sample. More looking on the part of both partners may indicatewariness, and longer infant looking may indicate less competent visual processing(Colombo, Richman, Shaddy, Greenhoot, & Maikranz, 2001). However, the finding oflowered interactive gaze contingencies is consistent with the literature.

Facial Affect

Despite lowered gaze coordination, depressed (vs. control) mothers heightened theirfacial coordination with infant facial and vocal affect, tending to brighten and soberas infants did. Thus, compared to controls, depressed mothers were “overly thrilled”when their infants were positive, and “overly disappointed” when their infants soberedor became negative. Moreover, depressed mothers showed less “woe-face,” an ex-pression of facial empathy for infant distress. While depressed mothers heightenedtheir facial interactive coordination, they lowered their facial self-contingency, aninteractive versus self-contingency imbalance. Thus, they were overly facially coor-dinated with infants while sacrificing their own facial stability—metaphorically“destabilized”—with lowered ability to anticipate their own next move. Infants ofdepressed mothers heightened their vocal affect coordination with maternal facialaffect. This was a significant but subtle effect.

Facial�Visual Engagement

Depressed mothers and their infants reciprocally heightened their facial�visual engage-ment coordination, a mutual vigilance in which both partners were overly reactive to theother. Mothers might have been overly thrilled when infants were positive and looking atthem, and overly disappointed when infants were negative and looking away. Moreover,

399MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

both partners were overly coordinated with the other in engagement while sacrificingself-stability.

Depressed mothers and their infants, thus, both showed a peculiar intermodal discor-dance in interactive contingency of attention versus affect. They lowered their contingentcoordination of gaze patterns at or away from the partner’s face, but they heightened theiremotional contingent coordination of facial/vocal/engagement shifts. Whereas attentionand emotion usually operate together as a “package” (Beebe & Gerstman, 1980), this splitattention�emotion coordination is a contradictory communication pattern, likely confus-ing to both partners.

Touch

Frequency of infant touch (none, one, or 2� touches of self, partner, or object) wasanalyzed in relation to maternal touch, from affectionate to intrusive. Infants of depressedmothers showed heightened contingent touch coordination with mother touch, comparedto controls. These infants were more likely to touch as maternal touch became moreaffectionate and less likely to touch as maternal touch became less affectionate. Thisfinding is striking on two counts. First, control infants did not coordinate their touchpatterns with mother touch so that control infants were free to use touch independent ofmother’s touch pattern. Second, depressed mothers had more intrusive touch, so that theirinfants were highly coordinated with a more intrusive form of maternal touch. Theseinfants were less likely to touch as maternal touch became more intrusive. More intrusivetouch in depressed mothers is consistent with other studies (see Field, 1995; Hart, Jones,Field, & Lundy, 1999; Malphurs et al., 1996). In depressed dyads in our study, infants’high coordination with maternal intrusive touch might have been an effort to manage it.But these infants were not free to use touch as needed, particularly when mothers wereintrusive. These infants were likely to experience a lessened efficacy in ability to regulatetheir own state through touch. For control infants, touch was uncoupled from maternaltouch, and it was more at infant disposal as needed. Strikingly, for their part, depressedmothers decreased their touch coordination with infant touch so that their infants werelikely to experience a decreased agency in their ability to affect maternal touch patterns.Thus, maternal touch was more intrusive and less sensitively coordinated with infant touchpatterns. The combination of heightened infant touch coordination, and lowered maternaltouch coordination, is an infant approach�mother withdrawal pattern, a form of dyadicconflict in the realm of touch.

Spatial Orientation

Depressed mothers were more spatially coordinated with infant head orientation thancontrol mothers. As infants oriented away, from en face toward arching away, depressedmothers were more likely to move from looming in toward sitting upright; reciprocally,as infants moved from arching away toward en face, depressed mothers were more likelyto move from sitting upright to leaning forward or looming in. Thus, depressed motherswere more sensitive to infant head orientation. They were more likely to draw back asinfants moved away, and they were more likely to move forward as infants reoriented.This maternal spatial orientation vigilance is consistent with the depressed mother’sfacial vigilance. Both findings indicate a greater reactivity to the infant’s level ofcontact. However, it is striking that depressed mothers coordinated more with infanthead orientation, but less with infant direction of visual regard, another peculiardiscordance.

400 BEEBE ET AL.

For their part, infants of depressed (vs. control) mothers were more likely to remainen face, whereas control infants were more free to venture away from en face. Thus, adyadic orientational sensitivity emerged in depressed pairs. To summarize, both depressedmothers and their infants showed an orientational�attentional discordance. They bothshowed orientational vigilance and more looking at the partner’s face but a loweredlikelihood of visual coordination with the partner’s pattern of looking and looking away.

Infant Intrapersonal Vocal Affect–Touch

Infants of both depressed and control mothers showed a positive correlation betweeninfant vocal affect and infant touch. When infants touched more in the previous fewseconds, infant vocal affect was more positive in the subsequent second (and viceversa). Thus, more infant touch functioned as a coping mechanism associated with lessvocal distress. This association was heightened in infants of depressed mothers. Thisfinding suggests a compensatory infant attempt to manage vocal distress throughtouch. However, at inevitable moments when infants touched less, depressed infantshad a greater likelihood of vocal distress. This was a significant, but subtle effect.

Summarizing the picture of depressed mothers, mothers (a) dampened (withdrew)their gaze coordination with infant gaze, which disturbed their monitoring of infantavailability for visual engagement; (b) dampened their touch coordination with infanttouch in the context of intrusive maternal touch; (c) showed less facial empathy through“woe face”; (d) spent more time looking at the infant’s face; and (e) heightened theircoordination with infant facial and vocal affect, engagement, and head orientation. Thesefindings together suggest that these depressed mothers were more self-preoccupied. It isas if they were overusing the infant’s affective state to regulate their own, perhaps usingthe infant’s face to see if they were loved. They seemed overly thrilled when infants werepositive and overly disappointed when infants became sober or negative. Mothers may becharacterized as “emotionally hovering” and spatially overreactive, but at the same timethey were relatively oblivious to infant visual availability and touch. The self-contingencyof the mothers’ own behavioral rhythms was destabilized in gaze, facial affect, facial�vi-sual engagement, and touch, likely yielding an experience of lowered self-coherencewithin the face-to-face interaction.

Summarizing the picture of infants of depressed mothers, infants showed the complexcombination of (a) orientational and attentional (more time looking) vigilance, (b) inter-personal attentional avoidance (lowered coordination of looking and looking away), (c)heightened affective coordination with maternal affect, and (d) heightened touch coordi-nation with maternal intrusive touch. The greater amount of time spent looking at motherand of enface orientation (but with lowered interpersonal coordination of looking patterns)suggests infant wariness. But infants simultaneously heightened their emotional and touchcoordination, showing an interpersonal vigilance. These infant patterns document strikingintermodal discordances within the infant. The self-contingency of the infants’ ownbehavioral rhythms was destabilized in looking patterns, vocal affect, and facial�visualengagement, likely yielding an experience of lowered self-coherence within the face-to-face interaction.

Summarizing the picture of the dyad of depressed mothers and their infants, there isa striking parallel in their patterns. Both depressed mothers and their infants showed thepeculiar noninterpersonal gaze pattern of more time spent looking but lowered contingentcoordination of looking patterns. Both partners also showed vigilant coordination inemotion. Thus, both partners showed attention�emotion discordance in lowered attention

401MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

coordination but heightened emotion coordination. Both partners showed heightenedsensitivity in orientation. Only touch showed a dyadic conflict, in which mothers withdrewtheir touch coordination, while infants heightened their coordination of touch withintrusive maternal touch. Thus, overall, infants made adaptations similar to those of theirmothers.

Evaluating High and Low Poles of Maternal Self-Reported DepressiveSymptoms

Because self-report scales are vulnerable to denial (Shedler, Mayman, & Manis, 1993), wehypothesized that maternal report of few or no depressive symptoms may be associatedwith communication difficulties similar to those associated with maternal report of manysymptoms. Using nonlinear analyses to interpret findings at the low end of the CES-Dscale, half the findings (15 out of 29) were nonlinear; they are presented in Table 2 andsummarized in Figure 1. In these significant nonlinear findings, altered self- and interac-tive contingency patterns were similar (in direction, but not necessarily absolute amount)at both high and low poles of depressive symptoms, compared to mothers with a midrangedegree of depressive symptoms.

Although we are intrigued by these nonlinear findings, which indicate difficulty at thevery low pole of symptoms and possibly measure maternal “denial of distress,” we remaincautious. Some mothers reporting very few or no symptoms may indeed be less vulner-able, whereas others may be using denial. This is a subject of controversy. For example,Tronick, Beeghly, Weinberg, and Olson (1997) argued that low scores on self-reportscales indicate postpartum “exuberance,” whereas Pickens and Field (1993) showed thatinfant facial expressions of low-scoring mothers were more negative. This debate deservesfurther consideration.

Limitations

Whereas most of the literature examines concurrent maternal distress and mother�infantcommunication patterns, we examined maternal 6-week depressive symptoms in relationto 4-month communication patterns. The effects of 6-week maternal depressive symptomsmay differ from those of 4-month symptoms. Nevertheless, it is important that maternaldepressive symptoms at this early stage are associated with interaction disturbances 2-1/2months later.

Significance for Mental Health and Intervention

An ongoing program translating research findings into brief videotape-assisted clinicalinterventions with mother-infant treatment pairs, and with mothers widowed on 9–11 andtheir young children, has been reported (Beebe, 2003, 2005; Cohen & Beebe, 2002;Beebe, Cohen, & Markese, 2011). The current findings can improve the specificity ofclinical interventions in similar community samples. Mothers with depressive symptomscan be helped to pay more attention to the infant’s necessary cycling of looking andlooking away. Perhaps mothers do not track the infant’s “look�look away” patternbecause the infant’s look away is interpreted as a rejection. These mothers can besensitized to the importance of infant touch patterns as a source of infant self-comfort.More affectionate maternal patterns of touch may restore the infant’s touch as his own,uncoupled from the mother’s touch pattern. Mothers can be helped to lessen theirvigilance of the infant’s head orientation toward and away, and to the infant’s emotionalups and downs, particularly if it seems that the mother is desperate for the infant’s love,

402 BEEBE ET AL.

overly thrilled as the infant becomes more engaged and positive, and overly disappointedas the infant becomes visually disengaged and negative. Mothers can be helped to “hold”the infant with more midrange rather than vigilant facial coordination, through bothinterpretive as well as educational strategies (see Beebe, 2003, 2005; Cohen & Beebe,2002). The findings of lowered self-contingency are also clinically useful. Mothers can behelped to appreciate the importance of their own—and their infant’s—stability andpredictability, which can be facilitated through educational, interpretive, and imagingstrategies.

Applicability of these findings to the treatment of adults is more speculative. However,the findings may be useful to the clinician when the adult patient comes with a history ofa mother with postpartum depression. Is it possible that the clinician might see residualaspects of these nonverbal patterns in the patient, although presumably reorganized bylater development. For example, does the patient show discordance with an emotionalvigilance but a lowered gaze coordination? Might the patient show the peculiar nonin-terpersonal form of looking, that is, looking more, but coordinating looking patterns less,with the therapist? Might the patient expect some form of intrusion, analogous to themother’s less affectionate, more intrusive touch? Might the patient’s forms of touchself-regulation be too “hooked” to aspects of the therapist’s own behaviors, and thus, lessavailable to the patient as he or she might need such self-regulatory touch? Thesespeculations await further exploration.

Conclusion

Four-month mother�infant face-to-face communication was substantially altered in dyadswhere mothers reported depressive symptoms at 6 weeks. Patterns of mother and infantself- and interactive contingency were equally informative and were largely cocreated byboth partners. Interactive contingency findings were consistent with our hypothesis thatdepression biases the system toward the poles of contingency—toward both highercontingency values in some modalities and lower contingency values in others. Thus,interactive contingency can be excessive (vigilant), as well as insufficient (withdrawn), aswe previously showed (Jaffe et al., 2001). This finding is consistent with our optimummidrange of interactive regulation. However, self-contingency was consistently loweredin depressed dyads, a robust finding not consistent with our hypothesis. The analysis ofseparate modalities revealed striking, complex, intermodal discordances, which wereforms of intrapersonal and dyadic conflict.

References

Bakeman, R., Adamson, L., Brown, J., & Eldridge, M. (1989). Can early interaction predict? Howand how much? In Bornstein, M. & Krasnegor, N. (Eds.), Stability and continuity in mentaldevelopment (pp. 235–248). Hillsdale, NJ: Lawrence Erlbaum.

Beebe, B. (2003). Brief mother�infant treatment using psychoanalytically informed video micro-analysis. Infant Mental Health Journal, 24, 24–52. doi:10.1002/imhj.10042

Beebe, B. (2005). Mother�infant research informs mother�infant treatment. The PsychoanalyticStudy of the Child, 60, 7–46.

Beebe, B., Cohen, P., & Markese, S. (2011). Mothers, infants and young children of September 11,2001: A primary prevention project. Journal of Infant, Child, and Adolescent Psychotherapy, 10,145–387.

Beebe, B., & Gerstman, L. J. (1980). The “packaging” of maternal stimulation in relation to infantfacial�visual engagement: A case study at four months. Merrill-Palmer Quarterly, 26, 321–339.

403MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Beebe, B., Jaffe, J., Buck, K., Chen, H., Cohen, P., Blatt, . . . Andrews, H. (2007). Six-weekpostpartum maternal self-criticism and dependency and 4-month mother�infant self- and inter-active contingencies. Developmental Psychology, 43, 1360 –1376. doi:10.1037/0012-1649.43.6.1360

Beebe, B., Jaffe, J., Buck, K., Chen, H., Cohen, P., Feldstein, S., & Andrews, H. (2008). Six-weekpostpartum maternal depressive symptoms and 4-month mother�infant self- and interactivecontingency. Infant Mental Health Journal, 29, 442–471. doi:10.1002/imhj.20191

Beebe, B., Jaffe, J., Lachmann, F., Feldstein, S., Crown, C., & Jasnow, J. (2000). Systems modelsin development and psychoanalysis: The case of vocal rhythm coordination and attachment.Infant Mental Health Journal, 21, 99–122. doi:10.1002/(SICI)1097-0355(200001/04)21:1/2�99::AID-IMHJ11�3.0.CO;2-#

Beebe, B., Jaffe, J., Markese, S., Buck, K., Chen, H., Cohen, P., . . . Feldstein, S. (2010). The originsof 12-month attachment: A microanalysis of 4-month mother�infant interaction. Attachment &Human Development, 12, 3–141. doi:10.1080/14616730903338985

Beebe, B., & Stern, D. (1977). Engagement�disengagement and early object experiences. In N.Freedman & S. Grand (Eds.), Communicative structures and psychic structures (pp. 35–55).New York, NY: Plenum Press.

Bigelow, A. E. (1998). Infants’ sensitivity to familiar imperfect contingencies in social interaction.Infant Behavior and Development, 21, 149–162. doi:10.1016/S0163-6383(98)90060-1

Brazelton, T. B., Koslowski, B., & Main, M. (1974). The origins of reciprocity. In M. Lewis & L.Rosenblum (Eds.), The effects of the infant on its caregiver (pp. 137–154). New York, NY:Wiley-Interscience.

Campbell, S. B., & Cohn, J. F. (1991). Prevalence and correlates of postpartum depression infirst-time mothers. Journal of Abnormal Psychology, 100, 594 –599. doi:10.1037/0021-843X.100.4.594

Chapple, E. (1970). Culture and biological man. New York, NY: Holt, Rhinehart, & Winston.Chen, H., & Cohen, P. (2006). Using individual growth models to analyze the change in quality of

life from adolescence to adulthood. Health and Quality of Life Outcomes, 4.Cohen, P. M., & Beebe, B. (2002). Video feedback with a depressed mother and her infant: A

collaborative individual psychoanalytic and mother–infant treatment. Journal of Infant, Child,and Adolescent Psychotherapy, 2, 1–55. doi:10.1080/15289168.2002.10486404

Cohn, J. F., Campbell, S. B., Matias, R., & Hopkins, J. (1990). Face-to-face interactions ofpostpartum depressed & nondepressed mother–infant pairs at 2 months. Developmental Psy-chology, 26, 15–23. doi:10.1037/0012-1649.26.1.15

Cohn, J. F., & Tronick, E. Z. (1988). Mother–infant face-to-face interaction: Influence is bidirec-tional and unrelated to periodic cycles in either partner’s behavior. Developmental Psychol-ogy, 24, 386–392. doi:10.1037/0012-1649.24.3.386

Cohn, J. F., & Tronick, E. Z. (1989). Specificity of infants’ response to mothers’ affective behavior.Journal of the American Academy of Child & Adolescent Psychiatry, 28, 242–248. doi:10.1097/00004583-198903000-00016

Colombo, J., Richman, W. A., Shaddy, D. J., Greenhoot, A. F., & Maikranz, J. M. (2001). Heartrate-defined phases of attention, look duration, and infant performances in the paired-comparisonparadigm. Child Development, 72, 1605–1616. doi:10.1111/1467-8624.00368

DeCasper, A. J., & Carstens, A. A. (1981). Contingencies of stimulation: Effects on learning andemotion in neonates. Infant Behavior and Development, 4, 19–35.

Demetriades, H. (2003). Maternal anxiety and maternal spatial proximity in 4-month mother�infantface-to-face interaction (Unpublished doctoral dissertation). Long Island University, Brookville, NY.

Dietz, P. M., Williams, S. B., Callaghan, W. M., Bachman, D. J., Whitlock, E. P., & Hornbrook,M. C. (2007). Clinically identified maternal depression before, during and after pregnanciesending in live births. The American Journal of Psychiatry, 164, 1515–1520. doi:10.1176/appi.ajp.2007.06111893

Fagen, J. W., Morrongiello, B. A., Rovee-Collier, C., & Gekoski, M. J. (1984). Expectancies and

404 BEEBE ET AL.

memory retrieval in three-month-old infants. Child Development, 55, 936–943. doi:10.2307/1130145

Feldman, R. (2006). From biological rhythms to social rhythms: Physiological precursors ofmother–infant synchrony. Developmental Psychology, 42, 175–188. doi:10.1037/0012-1649.42.1.175

Feldman, R. (2007). Parent�infant synchrony and the construction of shared timing: Physiologicalprecursors, developmental outcomes, and risk conditions. Journal of Child Psychology andPsychiatry, 48, 329–354. doi:10.1111/j.1469-7610.2006.01701.x

Field, T. (1995). Infants of depressed mothers. Infant Behavior and Development, 18, 1–13.doi:10.1016/0163-6383(95)90003-9

Field, T., Healy, B. T., Goldstein, S., & Guthertz, M. (1990). Behavior�state matching & synchronyin mother�infant interactions of nondepressed vs. depressed dyads. Developmental Psychol-ogy, 26, 7–14. doi:10.1037/0012-1649.26.1.7

Field, T., Healy, B. T., & LeBlanc, W. G. (1989). Sharing and synchrony of behavioral states andheart rate in nondepressed vs. depressed mother�infant interaction. Infant Behavior and Devel-opment, 12, 357–376.

Fox, N. (1994). The development of emotion regulation. Monographs of the Society for Researchin Child Development, 59(2–3, Serial No. 240).

Gianino, A., & Tronick, E. (1988). The mutual regulation model: The infant’s self and interactiveregulation, coping and defense. In T. Field, P. McCabe, & N. Schneiderman (Eds.), Stress andcoping (pp. 47–68). Hillsdale, NJ: Erlbaum.

Gitlin, M. J., & Pasnau, R. O. (1989). Psychiatric syndromes linked to reproductive function inwomen: A review of current knowledge. The American Journal of Psychiatry, 146, 1413–1422.

Gottman, J. (1981). Time-series analysis. Cambridge, United Kingdom: Cambridge UniversityPress.

Hains, S. M. J., & Muir, D. W. (1996). Effects of stimulus contingency in infant–adult interactions.Infant Behavior and Development, 19, 49–61. doi:10.1016/S0163-6383(96)90043-0

Haith, M. M., Hazan, C., & Goodman, G. S. (1988). Expectation and anticipation of dynamic visualevents by 3.5 month old babies. Child Development, 59, 467–479. doi:10.2307/1130325

Hane, A. A., Feldstein, S., & Dernetz, V. H. (2003). The relation between coordinated interpersonaltiming and maternal sensitivity in four-month-old infants. Journal of Psycholinguistic Research,32, 525–539. doi:10.1023/A:1025494200272

Hart, S., Jones, N. A., Field, T., & Lundy, B. (1999). One-year-old infants of intrusive andwithdrawn depressed mothers. Child Psychiatry and Human Development, 30, 111–120. doi:10.1023/A:1021902418770

Hay, D. (1997). Postpartum depression and cognitive development. In L. Murray & P. Cooper(Eds.), Postpartum depression and child development (pp. 85–120). New York, NY: GuilfordPress.

Hentel, A., Beebe, B., & Jaffe, J. (2000, July). Maternal depression at 6 weeks is associated withinfant self-comfort at 4 months. Paper presented at the International Conference on InfantStudies, Brighton, England.

Horwitz, S. M., Briggs-Gowan, M. J., Storfer-Isser, A., & Carter, A. S. (2007). Prevalence,correlates, and persistence of maternal depression. Journal of Women’s Health, 16, 678–691.doi:10.1089/jwh.2006.0185

Jaffe, J., Beebe, B., Feldstein, S., Crown, C., & Jasnow, M. (2001). Rhythms of dialogue in infancy.Monographs of the Society for Research in Child Development, 66(2, Serial No. 264), 1–132.

Kopp, C. B. (1989). Regulation of distress and negative emotions: A developmental view. Devel-opmental Psychology, 25, 343–354. doi:10.1037/0012-1649.25.3.343

Kushnick, G. (2002). Maternal spatial intrusion patterns in mother�infant face-to-face play:Maternal dependency, depression, and mother�infant chase and dodge (Unpublished doctoraldissertation). Long Island University, Brookville, NY.

Lewis, M., & Feiring, C. (1989). Infant, mother, and mother�infant interaction behavior andsubsequent attachment. Child Development, 60, 831–837. doi:10.2307/1131024

405MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY

Lewis, M., & Goldberg, S. (1969). Perceptual�cognitive development in infancy: A generalizedexpectancy model as a function of the mother–infant interaction. Merrill-Palmer Quarterly, 15,81–100.

Leyendecker, B., Lamb, M. E., Fracasso, M. P., Scholmerich, A., & Larson, D. (1997). Playfulinteraction and the antecedents of attachment: A longitudinal study of Central American andEuro-American mothers and infants. Merrill-Palmer Quarterly, 43, 24–47.

Lyons-Ruth, K., Bronfman, E., & Parsons, E. (1999). Maternal disrupted affective communication,maternal frightened or frightening behavior, and disorganized infant attachment strategies.Monographs of the Society for Research in Child Development, 64(3, Serial No. 258), p.67.

Lyons-Ruth, K., Repacholi, B., McCleod, S., & Silva, E. (1991). Disorganized attachment behaviorin infancy: Short-term stability and maternal and infant correlates and risk-related subtypes.Development and Psychopathology, 3, 377–396. doi:10.1017/S0954579400007586

Malatesta, C., Culver, C., Rich, J., & Shepard, B. (1989). The development of emotion expressionduring the first two years of life. Monographs of the Society for Research Child Development,54(1–2, Serial No. 219).

Malphurs, J. E., Raag, T., Field, T., Pickens, J., & Pelaez-Nogueras, M. (1996). Touch by intrusiveand withdrawn mothers with depressive symptoms. Early Development and Parenting, 5,111–115.

Markese, S., Beebe, B., Jaffe, J., & Feldstein, S. (2008, March). Rhythms of dialogue in infancy andattachment narratives in childhood. Paper presented at the XVI Biennial International Confer-ence on Infant Studies, Vancouver, Canada.

Martin, J. (1981). A longitudinal study of the consequences of early mother�infant interactions: Amicroanalytic approach. Monograph Series of the Society for Research in Child Development,46(3, Serial No. 190), 1–52.

Mumme, D. L., Fernald, A., & Herrera, C. (1996). Infants’ responses to facial and vocal emotionalsignals in a social referencing paradigm. Child Development, 67, 3219–3237. doi:10.2307/1131775

Murray, L., & Cooper, P. J. (1997). The role of infant and maternal factors in postpartum depression,mother�infant interactions, and infant outcome. In L. Murray & P. J. Cooper (Eds.), Postpartumdepression and child development (pp. 111–135). New York, NY: Guilford Press.

Murray, L., & Trevarthen, C. (1985). Emotional regulation of interactions between two-month-oldsand their mothers. In T. Field & N. Fox (Eds.), Social perception in infants (pp. 177–197).Norwood, NJ: Ablex.

Ohman, A. (2002). Automaticity and the amygdala: Nonconscious responses to emotional faces.Current Directions in Psychological Science, 11, 62–66.

Overton, W. (1998). Developmental psychology: Philosophy, concepts and methodology. In R. M.Lerner (Ed.), Theoretical models of human development: Vol. 1. Handbook of Child Psychology(5th ed., pp. 107–188). New York, NY: Wiley.

Pickens, J., & T. Field. (1993). Facial expressivity in infants of “depressed” mothers. DevelopmentalPsychology, 29, 986–988. doi:10.1037/0012-1649.29.6.986

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in thegeneral population. Applied Psychological Measurement, 1, 385– 401. doi:10.1177/014662167700100306

Roe, K. V., Roe, A., Drivas, A., & Bronstein, R. (1990). A curvilinear relationship between maternalvocal stimulation and 3 month olds’ cognitive processing. Infant Mental Health Journal, 11,175–189. doi:10.1002/1097-0355(199022)11:2�175::AID-IMHJ2280110208�3.0.CO;2-2

Sander, L. (April, 1995). Thinking about developmental process: Wholeness, specificity and orga-nization of conscious experiencing. Address to the annual meeting of American PsychologicalAssociation, Division 39, Santa Monica, CA.

Sander, L. L. (April, 1995). Thinking about developmental process: Wholeness, specificity, and theorganization of conscious experiencing. Address to the annual meeting of the Division ofPsychoanalysis, American Psychological Association, Santa Monica, CA.

406 BEEBE ET AL.

Shedler, J., Mayman, M., & Manis, M. (1993). The illusion of mental health. American Psychol-ogist, 48, 1117–1131. doi:10.1037/0003-066X.48.11.1117

Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, andindividual growth models. Journal of Educational and Behavioral Statistics, 24, 323–355.

Stepakoff, S., Beebe, B., & Jaffe, J. (2000, July). Mother�infant tactile communication at fourmonths: Infant gender, maternal ethnicity, and maternal depression. Paper presented at theInternational Conference of Infant Studies, Brighton, England.

Stern, D. N. (1971). A microanalysis of the mother�infant interaction: Behavior regulating socialcontact between a mother and her 3 1/2-month-old twins. Journal of the American Academy ofChild Psychiatry, 10, 501–517. doi:10.1016/S0002-7138(09)61752-0

Stifter, C. A., & Spinrad, T. L. (2002). The effect of excessive crying on the development of emotionregulation. Infancy, 3, 133–152. doi:10.1207/S15327078IN0302_2

Tamis-LeMonda, C. S., Bornstein, M. H., & Baumwell, L. (2001). Maternal responsiveness andchildren’s achievement of language milestones. Child Development, 72, 748–767. doi:10.1111/1467-8624.00313

Tarabulsy, G. M., Tessier, R., & Kappas, A. (1996). Contingency detection and the contingentorganization of behavior interactions: Implications for socioemotional development in infancy.Psychological Bulletin, 120, 25–41. doi:10.1037/0033-2909.120.1.25

Thomas, E. A., & Malone, T. W. (1979). On the dynamics of two-person interactions. PsychologicalReview, 86, 331–360. doi:10.1037/0033-295X.86.4.331

Thomas, E. A., & Martin, J. A. (1976). Analyses of parent�infant interaction. PsychologicalReview, 83, 141–156. doi:10.1037/0033-295X.83.2.141

Thompson, R. (1994). Emotion regulation: A theme in search of definition. In N. Fox (Ed.), TheDevelopment of Emotion Regulation. Monograph Series of the Society for Research in ChildDevelopment, 59(2–3, Serial No. 240), 25–52.

Tronick, E. Z. (1989). Emotions and emotional communication in infants. American Psycholo-gist, 44, 112–119. doi:10.1037/0003-066X.44.2.112

Tronick, E. Z., Als, H., & Brazelton, T. B. (1980). Monadic phases: A structural descriptive analysisof infant�mother face to face interaction. Merrill-Palmer Quarterly, 26, 3–24.

Tronick, E. Z., Beeghly, M., Weinberg, M. K., & Olson, K. L. (1997). Postpartum exuberance: Notall women in a highly positive emotional state in the postpartum period are denying depressionand distress. Infant Mental Health Journal, 18, 406 – 423. doi:10.1002/(SICI)1097-0355(199724)18:4�406::AID-IMHJ7�3.0.CO;2-I

Tronick, E. Z., & Weinberg, M. K. (1990). The infant regulatory scoring system (Unpublishedmanuscript). Children’s Hospital, Harvard Medical School, Boston, MA.

Warner, R. M. (1992). Sequential analysis of social interaction: Assessing internal versus socialdeterminants of behavior. Journal of Personality and Social Psychology, 63, 51–60. doi:10.1037/0022-3514.63.1.51

Watson, J. (1985). Contingency perception in early social development. In T. Field & N. Fox (Eds.),Social perception in infants (pp. 157–176). Norwood, NJ: Ablex.

Weinberg, M. K., & Tronick, E. Z. (1994). Beyond the face: An empirical study of infant affectiveconfigurations of facial, vocal, gestural, and regulatory behaviors. Child Development, 65,1503–1515. doi:10.2307/1131514

Weinberg, M. K., & Tronick, E. Z. (1998). The impact of maternal psychiatric illness on infantdevelopment. The Journal of Clinical Psychology, 59, 53–61.

Weinberg, M. K., Tronick, E. Z., Cohn, J. F., & Olson, K. L. (1999). Gender differences inemotional expressivity and self-regulation during early infancy. Developmental Psychology, 35,175–188. doi:10.1037/0012-1649.35.1.175

Zlochower, A. J., & Cohn, J. F. (1996). Vocal timing in face-to-face interaction of clinicallydepressed and nondepressed mothers and their 4-month old infants. Infant Behavior andDevelopment, 19, 371–374. doi:10.1016/S0163-6383(96)90035-1

407MATERNAL DEPRESSION AND MOTHER�INFANT CONTINGENCY