The Validity of Prenatal Assessments of Mothers’ Emotional, Cognitive, and Physiological Reactions to Infant Cry

INTRODUCTION

A number of investigators have demonstrated that women’s responses to cry stimuli during the prenatal period predict subsequent parenting outcomes (e.g., Ablow et al., 2013; Hechler et al.2019; Leerkes, 2010) and child outcomes (Leerkes et al., 2011; Ostlund et al., 2019). The focus on cry stimuli is based on evidence that infant cry cues are highly salient and reliably evoke broad individual differences in adult affect, cognition, physiology, neural responding, and behavior (e.g., Barr et al., 2014; Dudek et al., 2016; Groh & Roisman, 2009). Moreover, how adults respond to infant distress cues is a significant predictor of infants’ subsequent attachment security, emotion regulation, and behavioral and social competence (Leerkes et al., 2009; McElwain & Booth-La Force, 2006). Thus, the use of prenatal cry paradigms is particularly appealing based on the notion that identifying women at risk for compromised parenting during the prenatal period could allow for screening and preventative efforts prior to infants’ birth and/or intervention efforts in the early postpartum that could result in enhanced parenting and infant outcomes (Hechler et al., 2019; Rutherford et al., 2018). An apparent justification for this approach is the assumption that how women respond to cry stimuli prenatally is stable or reflective of their likely postnatal responses to their own infant (Pedersen et al., 1996); however, this proposition has rarely been tested directly (Hechler et al., 2019). Thus, the goals of the present study are to examine the extent to which first-time pregnant women’s emotional, cognitive, and physiological responses assessed in response to videos of crying infants (1) are stable when re-assessed in response to the same videos at 6 months postpartum after parenting experience has accrued; (2) converge (i.e., correlate) with parallel emotional, cognitive, and physiological responses assessed in response to their own infants during distress-eliciting caregiving tasks and a video-recall interview from those tasks (e.g., convergence between prenatal empathy in response to standard clips and postnatal empathy in response to own child); and (3) predict maternal sensitivity across the infants first 2 years of life.

Prior Research Using Prenatal Cry Paradigms

The existing literature assessing adults’ response to infant cry stimuli is vast. Given the purpose of this paper, we focus our review on studies of parents that included prenatal assessments of responses to cry stimuli in relation to postpartum outcomes. We identified 11 prior publications predicting parent/parenting outcomes (Ablow et al., 2013; Bjertrup et al., 2021; Bos et al., 2018; Dudek & Haley, 2020; Hechler et al., 2019; Leerkes, 2010; Leerkes et al., 2015; Pearson et al., 2011; Pedersen et al., 1996; Rutherford et al., 2018; Verhage et al., 2015) and two predicting infant outcomes (Leerkes et al., 2011; Ostlund et al., 2019) from parental responses to prenatal cry stimuli. Prenatal infant cry stimuli in these publications included audio-recordings of cry sounds (Bjertrup et al., 2021; Pedersen et al.,1996; Verhage et al., 2015), photographs/still images of infant facial expressions (Bjertrup et al., 2021; Dudek & Haley, 2020; Pearson et al., 2011; Rutherford et al., 2018), video recordings that include sound and visual images (Ablow et al., 2013; Leerkes, 2010; Leerkes et al., 2015; Ostlund et al., 2019), and infant dolls (called simulators) that are programmed to cry and allow for direct interaction (Bos et al., 2018; Hechler et al., 2019). The latter two reports, both from the same study, are the only ones that include both mothers and fathers in their sample; the others focus exclusively on mothers. Four of the identified studies focused exclusively on first-time mothers (Ablow et al., 2013; Leerkes, 2010; Pedersen et al., 1996; Verhage et al., 2015). Results of each of the studies cited above suggest that responses to prenatal cry stimuli demonstrate predictive validity to important postnatal outcomes as elaborated below.

Generally, women who report more negative cognitive (e.g., higher perceived intensity, negative attributions) and emotional reactions (e.g., higher aversion, irritation) to prenatal cry stimuli view their own infants more negatively (Pedersen et al., 1996), are more susceptible to the negative effects of infant difficult temperament on postnatal parenting efficacy (Verhage et al., 2015), and are more likely to experience postpartum depression (Bjertrup et al., 2021). Furthermore, women who report negative reactions to crying prenatally are subsequently observed to be less sensitive in response to their own infants (Leerkes, 2010; Leerkes et al., 2015) and have infants who are at greater risk of forming insecure attachments (Leerkes et al., 2011). In contrast, pregnant women who report more empathy via questionnaires with follow-up interview probing are better skilled at recognizing infant distress states, make more appropriate attributions about the causes of infant crying, are more behaviorally responsive to prenatal cry stimuli, and are subsequently observed to be more sensitive in response to their own infants (Hechler et al., 2019; Leerkes, 2010; Leerkes et al., 2015). Likewise, expectant mothers who demonstrated better physiological regulation, via greater RSA withdrawal from baseline to video cry stimuli, were subsequently rated more sensitive when interacting with their own infants (Ablow et al., 2013) and had newborns who were more regulated as demonstrated via low infant arousal during a standard neonatal assessment (Ostlund et al., 2019). Furthermore, expectant mothers’ better regulated arousal via the combination of high skin conductance arousal paired with high RSA withdrawal in response to video cry stimuli predicted subsequent higher maternal sensitivity toward their infants during distress eliciting tasks (Leerkes et al., 2015).

Finally, greater attention to infant distress versus non-distress in the prenatal period predicts subsequent adaptive parenting outcomes. For instance, longer reaction time on an attention task when exposed to distressed infant faces versus non-distressed infant faces behind the target images (suggesting biased attention to distressed versus non-distressed infants) during pregnancy was associated with higher reported bonding to one’s infant 3 to 6 months post birth (Pearson et al., 2011). Moreover, expectant mothers who demonstrated heightened cortical responses (i.e., greater right hemisphere N170 event related potential (ERP) amplitude reflecting greater processing/attention) in reaction to infant distress faces relative to non-distress faces during a Go/No-Go task were observed to be more sensitive when interacting with their own infants (Dudek & Haley, 2020). In contrast, attention to infant neutral faces, but not to distress faces, assessed via the late positive potential ERP, predicted compromised postnatal reflective functioning about infant states (Rutherford et al., 2018). That prenatal responses to cry stimuli predicted subsequent parenting and child outcomes in 12 out of the 13 identified studies illustrates the value of such an approach in developmental science, but important methodological questions remain.

Methodological Considerations

To our knowledge, only one prior study has explicitly tested the notion that prenatal emotional, cognitive, and physiological responses to cry stimuli predict how parents respond to similar stimuli from their own infants (Leerkes, 2010) and one has demonstrated that prenatal behavioral responses to a doll simulator predict subsequent parenting quality (Hechler et al., 2019). However, none examined the stability of responses to the same stimuli from the prenatal to postnatal period. Given that relatively stable adult characteristics like personality (Leerkes et al., 2015; Zeifman, 2003) and prior experiences like the parenting one received in childhood and adult attachment representations (Ablow et al., 2013; Groh & Roisman, 2009; Leerkes & Siepak, 2006) predict emotional, cognitive, physiological, and behavioral responses to cry stimuli, it is reasonable to expect some degree of stability in these responses over time and convergence with responses to their own infants’ cries. However, in a prior study using similar methodology, Leerkes (2010) reported small-to-moderate associations between parallel emotional and cognitive responses to infant distress assessed prenatally in response to standard cry clips and postnatally in response to mothers’ own infants (rs ranged from .10 to .38). Leerkes (2010) postulated that experiential and methodological factors may contribute to change over time or a lack of convergence. First, mothers’ emotional and cognitive responses to crying may change across the postpartum period as mothers gain experience caregiving, and if this is the case prenatal measures may have lower predictive validity to subsequent outcomes. Second, it is quite difficult to create own infant cry stimuli for all participants in a study that have the desired features that match standard cry stimuli. In particular, not all infants become distressed, nor intensely distressed for sustained periods during laboratory visits. These differences in the amount and intensity of infant distress across own infant and standard stimuli may also contribute to limited convergence of measures.

In an effort to address these methodological issues, we (1) examined rank order stability in how mothers respond to standard cry stimuli from the prenatal period until their own infants are 6 months old and (2) considered mothers’ prior experience with infants and observed infant distress at 6 months as moderators of the convergence between women’s prenatal responses to standard cry stimuli and postnatal responses to their own infant in an effort to inform future research. We reasoned that associations between parallel measures (e.g., prenatal and postnatal efficacy) would be higher among women who entered parenting with more prior experience with infants because their reactions to crying may be less likely to shift post-birth compared to women who had little prior experience with infants. Likewise, we predicted that associations between parallel measures would be higher among women whose infants displayed higher distress at 6 months because the postnatal stimuli are more similar to the standard cry stimuli, which were selected because of clear and persistent infant distress, than is the case for women whose own infants displayed lower distress at 6 months.

In addition, the prior study by Leerkes (2010) is the only one to directly compare the relative predictive validity of responses to prenatal cry stimuli and postnatal cry stimuli in relation to subsequent parenting; demonstrating they were comparable. However, this work was conducted in a relatively small and non-diverse sample; thus, replication is warranted. Such work is important, in part, given claims that the value of assessing prenatal responses to cry stimuli is in its potential to identify parents who would benefit from early, including prenatal, intervention to enhance parenting outcomes.

The Present Study

We draw on data from a diverse sample of 259 primiparous mothers followed prospectively from the third trimester until their children were 2 years old. The primary goals of the larger study were to test a comprehensive model of the origins of maternal sensitivity to infant distress (Leerkes et al., 2015). Focal maternal responses to infant distress that were informed by the social information processing perspective (Lemerise & Arsenio, 2000) and Dix’s (1991) affective organization of parenting model included accurate distress detection, empathy, negative emotions, attributions, and efficacy. In addition, two physiological reactions, changes in skin conductance level (SCL) and respiratory sinus arrythmia (RSA) when exposed to distress videos relative to a resting baseline, were selected because they capture immediate physiological responses to a stressor believed to reflect affective processes and should not be affected by social desirability. Increases in SCL, when presented with a challenge, indicate activation of the sympathetic nervous system and are believed to reflect emotional arousal (Stern et al., 2001); decreases in RSA, when presented with a challenge, are a parasympathetic nervous system response believed to reflect self-regulation (Porges, 2007). In our comprehensive model, we asserted that high SCL arousal accompanied by poor RSA regulation would predict a pattern of self-focused as opposed to infant-focused social cognition (i.e., low distress detection and empathy; high negative and minimizing attributions) which would in turn predict less sensitive responding to infants during distressing tasks.

During the prenatal period, women were exposed to videos of distressed infants to whom they had no relationship, and these responses were assessed. When infants were 6 months, all participating mothers were observed interacting with their own infants during distress-eliciting tasks. RSA and SCL were assessed during caregiving; the other focal measures were assessed via a video-recall interview after the observation. Then, a subset of the sample was re-exposed to the standard cry videos used in the prenatal assessment to assess rank order stability of reactions to distress across the transition to parenthood. Maternal sensitivity was observed during distress eliciting tasks when infants were 6 months, 1 year, and 2 years old. Thus, as illustrated in Figure 1, this data set presents unique opportunities to directly examine the stability (goal 1), convergent validity (goal 2), and relative predictive validity of assessing mothers’ reactions to prenatal distress stimuli (goal 3).

Figure 1.

Overview of study goals and methods.

In relation to goal 1, we hypothesized moderate-to-high rank order stability in responses to the same stimuli over time given the absence of methodological differences. In terms of goal 2, we anticipated modest-to-moderate associations based on prior research (Leerkes, 2010). We also tested moderators that may alter convergence as noted above. We predicted that associations between parallel measures would be higher among women who entered parenting with more prior caregiving experience and had infants who displayed higher distress at 6 months. Finally, we examined the predictive validity of the prenatal cry stimuli in relation to maternal sensitivity and compared the magnitude of these associations with the associations between parallel measures assessed in response to mothers’ own children and their subsequent parenting (goal 3). Based on prior research (Leerkes, 2010), we anticipated relatively comparable predictive validity to subsequent parenting from both prenatal standard video method and the 6-month observation video recall with own infant method. We hypothesized that higher accurate distress detection, empathy, situational/emotional attributions, efficacy, SCL arousal, and RSA withdrawal would be associated with greater maternal sensitivity, whereas greater negative emotional reactions, minimizing attributions, and negative attributions would be negatively associated with maternal sensitivity.

METHOD

Participants

Participants were 259 primiparous mothers (128 European American, 123 African American, 8 multiracial) and their infants from the southeastern United States. Mothers ranged in age from 18 to 44 years (M = 25.1, SD = 5.41). Twenty-seven percent had a high school diploma or less, 27% had attended but not completed college, and 46% had a 4- year college degree. The majority (57%) of mothers were married or living with their child’s father, 24% were in a relationship but not living with their child’s father, and 19% were single. Annual family income ranged from less than $2,000 to over $100,000; median income was $35,000. Although all mothers were primiparous, they varied in the extent to which they reported prior experience caring for infants: 22% reported no or little experience, 37% reported some, and 41% reported a lot of prior experience caring for infants. All participating infants were full term and healthy; 51% were female. Of the initial 259 participants (a sample size designed to adequately power testing the conceptual model reported in Leerkes et al., 2015), 211 mother-infant dyads participated in the 6-month observation, 207 at 1 year, and 199 at 2 years. There were 232 mothers with one or more observation of maternal sensitivity available. There were no differences on ethnicity, maternal age, income-to-needs, marital status, or experience with infants between mothers who completed the 6-month observation and interview and those that did not or mothers with at least one observational measure of sensitivity versus none.

Procedures

Expectant mothers were recruited at childbirth classes offered in the local hospital and public health department, breastfeeding classes offered through the Special Supplemental Nutrition Program for Women, Infants and Children, obstetric practices, and word of mouth. Upon enrollment in the study, women were mailed consent forms and a packet of questionnaires including measures of demographics. Women visited our laboratory for an interview 6 to 8 weeks prior to their due date. At this visit, mothers’ electrodermal activity and heart rate were recorded during a 2-min resting baseline while they viewed video clips of crying infants and as they were interviewed about their emotional and cognitive responses to the videos. Mothers and infants visited our laboratory for a videotaped observation of mother-infant interaction when infants were about 6 months (M = 6.39 months, SD = .75), 1 year (M = 13.90 months, SD = 1.05), and 2 years old (M = 27.32 months, SD = 2.52). At each visit, mothers and infants engaged in a 7-min free play, followed by 2 to 3 tasks designed to elicit infant distress. Immediately after the 6-month observation, mothers participated in an audiotaped video-recall interview in which they viewed the videotapes of each distress task and answered a series of questions to assess their emotional and cognitive responses during the interactions. Then, a subset of 103 mothers was re-administered the prenatal interview. Mothers’ electrodermal activity and heart rate were recorded at 6 months during a 2-min resting baseline while interacting with their infants and during the video-based interviews. Mothers received cash compensation ranging from $50 to $100 and small gifts at the end of each laboratory visit. All procedures were approved by the Institutional Review Board at UNC Greensboro (Protocol # 09-0035).

Prenatal Emotion Interview.

Electrodes were placed on mothers’ right collarbone and under each ribcage to record their heart rate, and two velcro strips were placed on the middle segments of two adjacent fingers of mothers’ non-dominant hand to record skin conductance level. These were connected to the Biolog 3991x (UFI, Morro Bay, CA) which stored physiological data for subsequent download to a computer using their General Purpose Physiological software v1.1. Expectant mothers engaged in a 2-min baseline procedure during which they were asked to be calm for 2 min while the interviewer left the room. Next, mothers viewed four 1-min video-clips of infants. The order of presentation was fixed given prior evidence suggesting no order differences when a similar method was used (Leerkes et al., 2004). Each mother saw two clips of infants exhibiting anger (assessed during an arm restraint procedure, one from the study reported in Leerkes & Crockenberg, 2003, and the other from the study reported in Swingler et al., 2014) followed by two clips of infants exhibiting fear (during a novel toy - bumble ball - approach, both from the study reported in Leerkes & Crockenberg, 2003), the first of each was European American and the second was African American. All infants were full-term at birth, typically developing, 5 or 6 months old when the videos were made, 2 were female and 2 were male, and all were dressed in gender neutral clothing. The infants were seated upright, and their face and torso were visible in all clips; in addition, the experimenter’s or mother’s hands were visible in the anger clips, and the bumble ball was visible in the fear clips. The average fundamental frequency was 800.00 Hz (range = 650.17–950.33) for the anger clips and 691.67 Hz (range = 450.33–1,000.76) for the fear clips. The average sound pressure level was 78.70 dB adjusted (dBA) (range = 65.00–89.9) for the anger clips and 71.85 dBA (range = 65.00–85.00) for the fear clips. As elaborated below in the measures section, after viewing each clip, mothers were interviewed and completed rating forms.

6-month Observation and Interview(s).

Mothers and infants engaged in a 7-min free play interaction, then infants were secured in an infant seat and mothers and infants participated in 3 distress-eliciting tasks. The first distress task was a 4-min arm restraint procedure designed to elicit infant frustration. The experimenter knelt in front of the infant seat and gently held the infant’s forearms immobile while keeping her head down and not interacting with the infant. The second distress task was a novel toy approach designed to elicit infant fear. The infant was tucked into a table with a barrier that prevented the toy from touching the infant. A remote control-operated dump truck with flashing lights, motion, and sound and an action figure seated on top approached the infant three times. Then, the truck’s horn, ignition, and a voice sounded, and music played while the truck vibrated and its lights flashed. The silent and still truck remained within the infant’s reach for 1 min. The entire task lasted 4 min. During the first min of both tasks, the mother was instructed to remain uninvolved unless she wanted to end the activity. Then, the experimenter signaled to the mother that she could interact with her infant as she pleased. The fear and anger tasks were inspired by tasks described by Goldsmith and Rothbart (1996) but were designed to be continuous rather than brief repeated trials. The final distress eliciting task was the Still Face procedure (Tronick et al., 1978). Mothers’ seats were moved across from their infant. Mothers were instructed to play with their infant as they normally would for 2 min, then to look at their infant with a still face for 2 min, and finally to play with their infant as they normally would for 2 min.

Immediately after the 6-month observation, the mother and experimenter moved to an adjacent room for the audiotaped video-recall interview while another research assistant cared for the infant. After mothers viewed the videotapes of each task (arm restraint, novel toy, and still face re-engagement) in their entirety, the experimenter asked the mother the same series of questions and asked her to complete the same series of questionnaires to assess emotional and cognitive responses during the distress-eliciting tasks. Infant affect in these videos is described below. Then, the first 103 mothers enrolled in the study who maintained participation at 6 months were re-administered the prenatal emotion interview using the 4 standard cry video clips as described above. (The decision not to re-administer to all participants was a function of a budget cut from the funder which necessitated scaling back on incentives and staff.)

1-year Observation.

Mothers and infants participated in a 7-min free-play interaction followed by 2 distress-eliciting tasks. The first task was a 4-min attractive toy in a jar procedure designed to elicit infant frustration. The researcher offered the infant an interactive toy phone. Once the infant was interested in the phone, the researcher placed it in a clear plastic jar and closed the lid so the infant could see but not touch the toy. The researcher prompted the infant to open the jar. After 4 min, the researcher opened the jar and allowed the infant to play with the phone. Next, during the novel character approach designed to elicit fear, the researcher left the room and a research assistant dressed in a green monster costume entered the room and engaged in a series of approaches toward and attempts to interact with the infant for 4 min. During the first min of both tasks, the mother was instructed to remain uninvolved unless she wanted to end the activity. Then, the experimenter signaled the mother that she could interact as she pleased.

2-year Observation.

Mothers and infants participated in a 7-min free-play interaction with age appropriate toys followed by 3 distress-eliciting tasks. The first task was a toy clean up, designed to elicit frustration. After the free-play period, the experimenter brought in 2 large storage containers and instructed mothers to get their child to clean up all of the toys in any way they wanted, but they had to involve their child. The task ended when 5 min was over or when all the toys were in the containers (M = 4.30 min, SD = 1.03 min). The second task was the attractive toy in a locked box, designed to elicit frustration. Children selected one of two attractive toys, after being allowed to play with it for a moment, the experimenter locked it in a clear container and gave the child a set of keys with the instruction that the child could play with the toy when the child opened the box. The correct key was not on the key ring. For 4 min, the experimenter prompted the child to use the keys to open the box. The third task was the spider approach, designed to elicit fear. The experimenter left the room and placed a stuffed spider attached to a remote-control car immediately inside of the door to the observation room. For 20 s, the spider remained still near the door. Then, the spider repeatedly approached to within 2 feet of the child, retreated from the child and paused until 3.5 min had passed. During the last 30 s, the experimenter returned to the room and asked the child to touch the motionless spider 3 times in a neutral voice. During the first min of the latter two tasks, the mother was instructed to remain uninvolved unless she wanted to end the activity. Then, the experimenter signaled the mother that she could interact as desired for the remaining 3 min.

Measures

A brief summary of the procedures and measures for the interviews using the standard cry clips and the video recall interview with own infant is presented in Table 1.

Table 1.

Summary of key interview procedures/measures.

	Prenatal and 6 Month Standard Clips	6 Month Own Infant
Procedure	Watch each 1-min clip and answer questions after each	Watch video of each interactive task in entirety and answer questions after each
Accurate Distress Detection	Average of the following after being standardized: # of clips mother minimized (rated as not distressed) (reversed) # clips in which mother identified correct dominant emotion (fear or anger) Weighted dominant emotion error score (0 = no error, 1 = other negative emotion, 2 = non-emotion term, 3 = positive emotion term) summed across clips (reversed)	Average of the following: #seconds infant was distressed according to trained raters and mothers rated infant as being neutral and or positive the entire time, averaged across tasks (reversed) #of seconds infants was distressed according to trained raters and mother did not list a negative emotion term to describe infant, averaged across tasks (reversed)
Empathy	Mean rating on 3 emotions (empathy, sympathy, sad), if coded as infant-oriented, averaged across 4 clips	Same as prenatal, but averaged across 3 tasks
Mother-Oriented Emotions	Mean rating on 9 emotions (angry, irritated, frustrated, annoyed, disgusted, anxious, concerned, nervous, worried) if coded as mother-oriented, averaged across 4 clips	Same as prenatal, but averaged across 3 tasks
Situational Emotional Attributions	Mean rating on 4 items across, averaged across 4 clips: upset by the situation, no one was helping baby, trying to show he/she needs help; had no way to feel better	Same as prenatal, but averaged across 3 tasks
Negative Attributions	Mean rating on 7 items across, averaged across 4 clips: spoiled, difficult temperament, trying to make mother’s life difficult, unreasonable, crying on purpose, selfish, just wanted attention	Same as prenatal, but averaged across 3 tasks
Minimizing Attributions	Mean rating on 5 items across, averaged across 4 clips: having a bad day, in a bad mood, tired, hungry, not feeling well	Same as prenatal, but averaged across 3 tasks
Efficacy	Mean rating on 4 items, averaged across 4 clips: know what baby wants or needs, understand what baby is feeling, know how to handle the situation, handle the situation effectively	Same as prenatal, but averaged across 3 tasks
SCL and RSA (relative to resting baseline)	Recorded while watching each clip, averaged across clips	Recorded while interacting with infant during distressing tasks prior to interview, averaged across tasks

Physiological Arousal.

SCL was continuously recorded in microsiemens on the Biolog at a sampling rate of 100Hz, and average SCL during each video was calculated. Difference scores were calculated for each of the four video-clips for the prenatal interview and its repeat at 6 months and for the 3 interactive tasks at 6 months by subtracting the baseline SCL score from the clip/task scores. Positive scores indicate an increase in arousal from baseline to clip or from baseline to distress-eliciting task. Change scores for each clip or for each task were averaged to yield a single measure of physiological arousal; αs = .93, .96, and .92, respectively, for the prenatal interview, 6-month observation, and 6-month interview. Data were missing due to equipment malfunctions for 3 mothers prenatally and 2 at 6 months.

Physiological Regulation.

Mothers’ electrocardiogram was recorded at a sampling rate of 1 kHz. A data file containing the interbeat intervals (IBI), or the time between R-waves, was transferred to a computer for artifact editing (resulting from movement) and analyzed using the CardioEdit v 1.5. software (Brain Body Center, University of Illinois at Chicago). A staff member was trained to reliably edit artifacts using the Inter-Beat-Interval Training Program v. 3 provided by the Brain Body Center. Estimates of RSA were calculated using Porges’s (1985) method. Heart period (HP) was derived from the IBI data, then an algorithm was applied to the sequential HP data. A band-pass filter extracted the variance of HP within the frequency band of spontaneous respiration (.12–.40 Hz) in adults. RSA, in ms², was calculated for every 15-s epoch during baseline and during each of the video clips and was then averaged across epochs within a task of interest. Tasks in which more than 10% artifact editing was needed were treated as missing data and later imputed based on maternal ethnicity and RSA from the other clips/tasks; this occurred for 3 mothers parentally and 10 at 6 months. Vagal withdrawal scores were calculated for each clip/observational task by subtracting the average RSA from the average RSA during baseline. Change scores for each clip/observational task were averaged to yield a single score; αs = .87, .81, and .89, respectively for the prenatal interview, 6-month observation, and 6-month interview. Higher scores indicate greater vagal withdrawal and better physiological regulation. RSA data were completely missing for 2 mothers prenatally and 3 mothers at 6 months due to equipment malfunctions or poor electrode placement.

Emotional Reactions to Distress.

After watching each of the four 1-min standard video clips or 3 distress eliciting tasks with own infant, mothers rated how strongly they felt 17 emotions (e.g., sad, irritated, concerned) on a 4-point scale ranging from 1 (not at all) to 4 (very strongly) using a paper questionnaire. Then, mothers were asked to describe verbally why they felt each emotion. Each explanation was coded as infant-oriented (i.e., concerns about the infant’s welfare, a desire to help the infant, sympathy or empathy for the infant, or feeling pleasure or pride in the infant’s behavior; e.g., I was sad for the baby) or mother-oriented (i.e., self-focused concerns, negative reactions about the infant, or responses that are of interest or importance to the mother but not the infant; e.g., I was sad for myself that I had to watch that) based on procedures developed by Dix et al., (2004). Reliability for infant versus mother-orientation on each distinct emotion was calculated based on 40 double-coded interview transcripts for the prenatal and 6-month interviews and 20 interviews for the repeat of the prenatal interview conducted on the subset of mothers at 6 months. Kappas ranged from .65 to 1.0 (M kappas = .85 prenatally, .94 for 6-month own infants, and .97 for 6-month standard clips). Two emotional reaction scores were calculated for each clip/task by averaging mothers’ intensity ratings for appropriate emotions: empathy (i.e., infant-oriented empathy, sympathy, sadness) and mother-oriented negative emotions (i.e., angry, irritated, frustrated, annoyed, disgusted, anxious, concerned, nervous, worried). Then, parallel scores were averaged across the 4 clips or 3 tasks to yield composite scores of empathy (αs = .87 for prenatal standard clips, .74 for 6 month own infants, and .86 for 6 month standard clips ) and mother-oriented negative emotions (αs = .80 for prenatal standard clips, .70 for 6 month own infants, and .74 for 6 month standard clips).

Accurate Distress Detection.

Accurate distress detection was assessed using procedures developed by Leerkes (2010). After viewing each standard video clip, both prenatally and at 6 months, mothers rated infant distress on a scale ranging from 1 (high positive) to 7 (high negative). Mothers were coded as minimizing or underrating distress if they gave a global rating of 1, 2, 3, or 4, indicating they did not recognize that the infant was distressed. The number of clips they minimized was summed for a possible range from 0 to 4 (M = 0.50, SD = 0.77, and M = 0.31, SD = 0.67, for prenatal and 6 months, respectively). For each clip, mothers were asked to identify the infant’s dominant emotion from a list of 20 emotion terms (e.g., interested, happy, sad, angry). Angry, frustrated, irritated, annoyed, or disgusted were coded as accurate for the anger clips. Afraid, scared, nervous, anxious, wary, or worried were coded as accurate for the fear clips. The number of times mothers accurately identified the dominant negative emotion was summed across clips for a possible range from 0 to 4 (M = 2.59, SD = .99, and M = 2.93, SD = .90, for prenatal and 6 months, respectively). When mothers named an incorrect dominant emotion, the nature of their error was coded into one of three weighted categories: another negative (e.g., indicated angry for fear clip; coded 1), non-emotion (e.g., tired, hungry, or bored; coded 2), or positive emotion/state (e.g., happy; coded 3). A total weighted error score was computed by summing across the 4 clips resulting in a possible range from 0 to 12 (M = 2.51, SD = 1.61, and M = 2.04, SD = 1.34, for prenatal and 6 months, respectively). These variables were standardized and averaged (reverse scoring minimization and dominant emotion ratings) to create an accurate distress detection composite for the prenatal and 6-month interviews using the standard cry clips.

A different approach was used to score distress detection for the video-recall interview about own infant because there was insufficient time in the moment to screen or edit the videos to focus only on distress segments. Mothers were asked to indicate how frequently their infants were distressed during each interactive task on a 7-point scale from 0 (never) to 6 (the whole time) and to indicate all emotions the infant displayed during each task using a list of 20 emotion terms (e.g., happy, sad, angry). To score distress detection, mothers’ responses about their infants’ state were compared to ratings made by reliably trained infant affect coders (described below). If an infant was distressed according to our raters (i.e., a score of 5, 6, or 7), and the mother rated the infant as never distressed (i.e., under-rating) or failed to indicate the infant felt specific negative emotions like sadness, fear, anger (i.e., under-identification), the number of s the infants was rated as distressed by us was recorded to reflect the magnitude of her detection error. That is, not noting an infant was distressed if the infant cried for 30 s is a bigger error than not noting the infant only cried for 5 s. Mothers who did not make these errors were scored as 0. These scores were calculated for each caregiving task and then summed across tasks. The two types of detection errors correlated, r (206) = .20, p < .01, and were averaged. This score was multiplied by −1 so high scores reflect more accurate distress detection. In contrast to the prenatal measure, there was no measure related to correctly noting the dominant negative emotion or the type of dominant emotion errors made because not all infants displayed negative emotions, and we could not code specific emotions for infants in the sample because doing so requires moment-to-moment facial affect coding.

Causal Attributions about Crying.

After watching each standard video clip/task, mothers rated the extent to which they agreed with 18 statements about why the infant from each clip or their own infant was crying on a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree), developed by Leerkes and Siepak (2006). Three attribution scores were created for each clip/task: emotion minimizing attributions includes five items (i.e., having a bad day, in a bad mood, tired, hungry, not feeling well); negative/internal attributions includes seven items (i.e., spoiled, difficult temperament, trying to make mother’s life difficult, unreasonable, crying on purpose, selfish, just wanted attention); and situational/emotional attributions includes four items (i.e., upset by the situation, no one was helping the baby, trying to show he/she needs help, had no way to feel better). Scores were averaged across the 4 clips or 3 tasks. Alphas for all attributions scales and interviews were acceptable, ranging from .83 to .95.

Efficacy.

After each video clip, mothers were asked to rate how good they thought they would be at the following 4 items developed by Leerkes et al., (2004): knowing what baby wants or needs, understanding what baby is feeling, knowing how to handle the situation, and handling the situation effectively. For the standard clips they were asked to imagine how good they would feel if their own infant were feeling like the infant in the video; for the videos of their own infant they were asked to rate how good they felt during the interactive task. The response scale ranged from 1 (not good at all) to 4 (very good). Scores were averaged across the 4 standard clips or 3 distress-eliciting tasks; αs = 93, .87, and .95 for the prenatal interview, 6-month interview about own infant, and 6-month interview about the standard clips, respectively.

Infant Affect.

Infant affect was continuously rated/coded from digital media files of the 6-month distress-eliciting tasks using INTERACT 9 (Mangold, Arnstorf, Germany). The average rating of infant affect during the arm restraint task, novelty task, and still face re-engagement episode was used as an index of infant affect which was rated on a 7-point scale based on facial expressions, vocalizations, and body movements ranging from 1 (high positive affect) (e.g., intense smile, laughing or squealing) to 7 (high negative affect) (e.g., screams, wails, sobs intensely). This system was adapted from Braungart-Rieker and Stifter (1996). Coders were trained to 80% agreement with the first author based on videos deliberately selected to illustrate the full range of infant affect prior to coding independently. Then, inter-rater reliability was established between coders based on 34 randomly selected double-coded tapes; weighted kappa = .76. Only 8 (4%) infants did not become distressed, and the average duration of distress across the tasks was 2 min (range = 0 to 7.75 min) demonstrating the effectiveness of the tasks at eliciting distress from mothers’ own infants at 6 months but also the variability in infant distress. However, in contrast to the standard prenatal video stimuli, very few infants (17%) displayed high distress (coded as 7) at some point and most (87%) infants displayed positive affect at some point during the distress-eliciting tasks. This illustrates the importance of considering infant affect as a moderator of convergence.

Maternal Sensitivity.

Maternal sensitivity during each interactive task at each time point was rated using Ainsworth’s 9-point sensitivity scale in which coders consider a mother’s awareness of and apparent interpretation of infant cues and the appropriateness and promptness with which she responds to those cues (Ainsworth et al., 1978). To illustrate, a rating of 1 (highly insensitive) is characterized by rare, slow, inappropriate, fragmented and incomplete responses to infant signals suggesting the mother’s behavior is driven by her own needs, whereas a rating of 9 (highly sensitive) is characterized by consistently appropriate and temporally contingent responses to infant cues that appear to reflect a mothers awareness, accurate interpretation, and prioritization of infant cues. At each time point, 15 to 20% of randomly selected videos were double coded to assess inter-rater reliability via intraclass correlation coefficients (ICC), after coders underwent training as described above. Mean ICC across all waves and tasks was .88. A single measure of maternal sensitivity during distress tasks over the first 2 years was created by averaging the scores from each task and time point (α = .91).

Analytic Plan

Descriptive statistics, one set of correlations, and multiple regressions were calculated in SPSS version 26 (IBM, Armonk, NY). The remaining correlations were calculated in Mplus (version 8; Muthén & Muthén, 1998-2017). Prior to hypothesis testing, the demographics and prenatal responses of the subsample of mothers who were re-administered the interview at 6 months using standard cry clips (n = 103) were compared to those who were not (n = 108) using t-tests for continuous variables and chi squared for ethnicity. Then, descriptive statistics were calculated for all primary variables.

To address goal 1, stability in maternal responses to the standard videoclips from the prenatal period to 6 months postpartum were examined via zero-order correlations among the subsample of 103 women who provided this data at both time points in SPSS. Correlations were calculated again in Mplus using full information maximum likelihood (FIML) so that any available data was used to estimate the correlations for the 211 participants who participated at 6 months. This approach was used to compare the results with and without FIML and to facilitate a subsequent comparison with convergent validity correlations for the 211 mothers with relevant data at 6 months.

To address goal 2, the extent to which pregnant women’s responses to the standard cry videos converged with parallel responses about their own infant at 6 months was examined via zero-order correlations. Then, for each measure, the magnitude of convergent validity was compared to the magnitude of rank order stability using Fisher’s r-to-z transformation for non-independent samples. Next, hierarchical multiple regression was used to examine the extent to which convergence between parallel measures varied as a function of prior experience with infants and observed infant distress at 6 months. Nine regressions were conducted; one predicting each of the 9 responses to own infant at 6 months. Each regression included the parallel response from the prenatal period (for example prenatal accurate distress detection in response to the standard clips predicting accurate distress detection in response to own infant at 6 months), the two moderators of interest (prior experience, observed infant affect), and then their interaction terms with the prenatal interview response. Variables were centered and interaction terms were calculated as the product of the variables of interest. Given the number of interactions tested, the Benjamini-Hochberg procedure (McDonald, 2009) was used to adjust the p-values within the set of 9 interactions involving prior experience with infants and within the 9 interactions involving observed infant distress. Significant interactions were probed by calculating the simple slopes at + and − 1 SD of the M.

To address goal 3, the predictive validity of responses to both the prenatal standard cry videos and own infant video-recall and maternal sensitivity during distress-eliciting tasks was examined via zero-order correlations. Given the number of correlations tested, the Benjamini-Hochberg correction was again applied to the set of 9 correlations from responses to prenatal cry clips to sensitivity and to the set of 9 correlations from own infant video recall at 6 months. Then, the magnitude of the association between each parallel measure assessed prenatally using the standard cry stimuli and at 6 months using video recall with own infant stimuli in relation to maternal sensitivity was compared using Fisher’s r-to-z transformation for non-independent samples.

RESULTS

Preliminary Analyses

Of the 13 analyses conducted to compare the subset of women who were re-administered the interview with standard cry clips at 6 months postpartum to those who were not, only 1 statistically significant difference emerged. Mothers who were re-administered the interview at 6 months had a higher income to needs ratio (M = 3.31, SD = 2.20) than mothers who were not re-administered the survey (M = 2.69, SD = 2.06), t (101) = 2.03, p < .05, d = .29. Thus, the subgroup was generally representative of the larger sample. Descriptive statistics on primary variables are presented in Table 2.

Table 2.

Descriptive statistics.

	Standard Cry Clips						Own Infant
	Prenatal N = 259			6 Months n = 103			6 Months n = 211
	M	SD	Possible Range	M	SD	Possible Range	M	SD	Possible Range
Accurate distress detection1	0.05	.81	−2.66 – 1.03	0.07	.76	−2.78 – .82	−7.83	13.97	−90.42 – 0
Empathy	2.16	.60	1 – 4	1.78	.56	1 – 4	1.52	.51	1 – 4
Negative Emotions	1.11	.15	1 – 4	1.03	.08	1 – 4	1.11	.13	1 – 4
Minimizing Attributions	1.58	.53	1 – 4	1.31	.42	1 – 4	1.42	.43	1 – 4
Negative Attributions	1.41	.37	1 – 4	1.30	.35	1 – 4	1.30	.29	1 – 4
Situational Attributions	3.21	.43	1 – 4	3.03	.51	1 – 4	2.04	.60	1 – 4
Efficacy	3.44	.39	1 – 4	3.57	.44	1 – 4	3.46	.40	1 – 4
Skin Conductance1	2.48	2.11	−5.22 – 9.13	1.25	1.85	−2.26 – 7.37	3.23	2.26	−1.85 – 2.37
RSA Withdrawal1	.73	.91	−3.15 – 4.41	−.26	.88	−2.31 – 2.82	.38	.73	−4.30 – 10.49
Experience with infants	2.15	.87	0 – 3
Observed infant distress							4.25	.41	1 – 7
							6M to 2 Year n = 232
Maternal Sensitivity							5.31	1.39	1 – 9

¹Actual range presented because values are based on a composite of z scores or reflect change scores.

^*p < .05.

^**p < .01.

Goal 1: Rank Order Stability in Response to Standard Clips

As demonstrated in the first and second columns of Table 3, the values and statistical significance of stability correlations were comparable across analytic approaches. Furthermore, with the exception of RSA withdrawal, all responses to the standard infant cry video clips demonstrated moderate and statistically significant stability from the prenatal period to 6 months. The mean correlation was .34, 95% CI [.30, .38], for those with complete data (n = 103) and .35, 95% CI [.32, .37], when FIML was used (n = 211).

Table 3.

Zero-order correlations.

	Rank Order Stability		Convergent Validity	Predictive Validity to Observed Maternal Sensitivity 6M to 2Y
Measure	Standard Cry Clips Prenatal & 6 Months n = 1031	Standard Cry Clips Prenatal & 6 Months n = 2112	Prenatal Standard Cry Clips & Own Infant 6 Months n = 211	From Prenatal Standard Cry Clips n = 232	From Own Infant Video Recall at 6 Months n = 232
Accurate distress detection	.34**	.37**	.22**	.19**	.20**
Empathy	.41**	.41**	.27**	.19**	.20**
Negative Emotions	.54**	.51**	.34**	−.08	−.03
Minimizing Attributions	.47**	.47**	.38**	−.19**	−.17**
Negative Attributions	.42**	.43**	.39**	−.22**	−.29**
Situational Attributions	.32**	.32**	.26**	.09	.02
Efficacy	.36**	.37**	.38**	−.04	.002
Skin Conductance	.30**	.33**	.37**	.12	.25**
RSA Withdrawal	−.10	−.09	−.09	.07	.01

¹Correlation from SPSS on subsample only;

²same correlations from MPLUS using FIML.

^*p < .05.

^**p < .01.

Goal 2: Convergence Between Prenatal Responses to Standard Videos and Own Infant Clips

The correlations for convergent validity are reported in the third column of Table 3 for the full sample. With the exception of RSA withdrawal, all parallel responses demonstrated statistically significant convergence that was modest to moderate in magnitude. The mean correlation was .28, 95% CI [.26, .30]. Notably, the magnitude of these associations was comparable to the rank order stability correlations based on Fisher’s r-to-z for non-independent samples comparing correlation coefficients in columns 2 and 3, with two exceptions. Rank order stability using the standard clips at both times was higher than was convergent validity from the prenatal standard cry clips to video recall with own infant for both empathy, r(207) = .41, p < .01 versus r(207) = .27, p < .01, z = 2.14, p < .05, and negative emotions, r(207) = .51, p < .01 versus r(207) =.34, p < .01 , z = 2.84, p < .01.

With respect to moderators of convergence, of the 18 interactions tested across the 9 models, only 4 were statistically significant, two each for prior experience with infants (in relation to minimizing and negative attributions) and for observed infant distress (in relation to negative emotions and minimizing attributions). The statistically significant interactions are summarized below, and the complete set of regression results is available in a supplemental table.

First, prior experience with infants moderated the associations between both minimizing attributions, β = .19, p = .002, and negative attributions, β = .14, p = .03, assessed prenatally in response to the standard clips and the parallel measure assessed at 6 months in response to own infant. Consistent with prediction, convergence was stronger among mothers with greater prior experience with infants than among mothers with less prior experience, β = .55, p < .001 versus β = .22, p = .01, for minimizing attributions and β = .49, p < .001, versus β = .18, p = .07, for negative attributions. Second, observed infant affect moderated the associations between both negative emotions, β = .20, p = .001, and minimizing attributions, β = .14, p = .02, assessed prenatally in response to the standard clips and the parallel measure assessed at 6 months in response to own infant. Both were consistent with prediction, such that convergence was stronger among mothers whose infants displayed greater distress than among mothers whose infants displayed less distress, β = .56, p < .001, versus β = .11, p = .21, for negative emotions and β = .48, p < .001, versus β = .23, p = .01, for minimizing attributions.

Goal 3: Predictive Validity to Observed Sensitivity

The predictive validity correlation for responses to the prenatal standard cry videos and own infant video-recall in relation to maternal sensitivity during distress-eliciting tasks are presented in columns 4 and 5 of Table 3, respectively. Overall, 4 out of 9 associations between prenatal responses to the standard videos and subsequent maternal sensitivity were statistically significant. Given that the valences of most correlations were consistent with predictions (e.g., empathy correlated positively whereas negative attributions correlated negatively with sensitivity), the absolute values were used to calculate the mean correlation with one exception. That is, the negative valence was maintained for the unexpected association between prenatal efficacy and sensitivity. The mean correlation between prenatal measures and sensitivity was .12, 95% CI [.11, .13]. Likewise, 5 out of 9 associations between video recall responses about one’s own infant assessed at 6 months and maternal sensitivity were statistically significant and the valences of all associations were consistent with expectation. The mean absolute value of these associations was .13, 95% CI [.12, .14].

Generally, there was high consistency in the pattern of associations for specific responses to crying regardless of assessment method and timing. That is, the following responses to standard videos prenatally and video recall of own infant at 6 months correlated with sensitivity significantly and at a comparable magnitude (prenatal standard cry videos/6-month own infant): accuracy, r(230) = .19, p < .01/ r(230) = .20, p < .01; empathy, r(230) = .19, p < .01/ r(230) = .20, p < .01; minimizing attributions, r(230) = −.19, p < .01/ r(230) = −.17, p = .01; and negative attributions, r(230) = −.22, p < .01/ r(230) = −.29, p < .01. Likewise, the following responses were not significantly related with sensitivity regardless of method/timing: negative emotions, r(230) = −.08, p = .24/ r(230) = −.03 p =.70; situational attributions, r(230) =.09, p = .17/ r(230) =.02, p = .79; efficacy, r(230) = −.04, p = .50/ r(230) =.002, p = .98; and RSA withdrawal, r(230) =.07, p = .28/ r(230) =.01, p =.84. The one exception to this overall pattern of similarity was SCL. Maternal SCL during the 6-month caregiving tasks was a significant predictor of maternal sensitivity, r = .25, p <. 01, whereas prenatal SCL was not, r = .12, p = .08, and this was the only pair of correlations that differed, albeit marginally, z = 1.70, p = .09.

DISCUSSION

The goals of the current study were to examine (1) the rank order stability of first-time pregnant women’s prenatal responses to infant cry videos when reassessed at 6 months postpartum, (2) convergent validity with emotional, cognitive, and physiological responses towards their own infants during distress-eliciting tasks and video-recall interview, and (3) predictive validity to maternal sensitivity when infants were 6 months to 2 years old. Women’s responses to cry cues during the prenatal period predict later parenting outcomes, but the stability of these responses has rarely been tested (Hechler et al., 2019). The present study utilized mother-report of cognitive and emotional responses to crying cues, observations of mothers and infants during distress-eliciting tasks, and physiological responses to assess stability and validity of prenatal responses to standard videos and own infant clips. Overall, results support our hypotheses, such that there was generally moderate rank order stability between mothers’ prenatal and 6 month responses to the standard cry stimuli, evidence of convergence between mothers’ prenatal cry stimuli responses and 6-month own infant stimuli for most measures, and comparable predictive validity from both methods to maternal sensitivity for all but one measure.

Goal 1: Rank Order Stability in Responses to Standard Stimuli

Primarily consistent with the first hypothesis, there was moderate stability in mothers’ responses to the standard cry stimuli except for RSA, which had a near-zero association. Thus, mothers’ rank level of emotional and cognitive responses to infant distress as well as their physiological arousal, as indexed by SCL, remained relatively stable from the third trimester to 6 months postpartum. To our knowledge this is the first study to assess and demonstrate rank order stability in cognitive, emotional, and physiological reactions to infant cry stimuli from the prenatal to postnatal period. These results suggest that, although mothers may become more skilled in responding to infant distress with experience, their rank order performance remains fairly comparable over time (Bornstein et al., 2017). The lack of stability for RSA is somewhat puzzling but could be due to the order of the of the 6-month assessments which could have led to contrast effects. That is, given the standard cry clips were presented after mothers interacted with their own infants in distressing situations, the standard clips may not have been as moving or viewed as a problem the mother needed to address. In contrast, when these first-time pregnant women were presented with the standard cry clips in the prenatal period they were likely imagining how they would cope with such situations with their own infant.

Goal 2:Convergence Between Prenatal Standard Responses and Postnatal Own Infant Responses

There was generally moderate convergence between prenatal responses to the videos of stranger infants and mothers’ responses on the video recall interview based on videos of their own infants. The correlations in this study were generally stronger than those reported by Leerkes (2010) (Mrs of .30 versus .19 for the 4 constructs that overlapped across the two studies), although only significantly stronger for mothers’ negative emotional reactions to crying (rs = .35 in the current study versus .10 in the prior, z = 2.16, p = .015). Two methodological enhancements may have led to this slight improvement. First, pregnant women responded to four rather than two standard cry videos in the current study which may have led to more reliable measures of women’s responses to crying. Second, three distress eliciting tasks as opposed to two were used in the current study which resulted in a longer average duration of distress than in the earlier study (a difference of approximately 1 min). In other words, the match between the prenatal standard stimuli and own infant stimuli were likely enhanced in the current study, relative to the prior study, which may have led to the slightly increased convergence. Notably, the magnitude of convergent validity correlations was comparable to the stability correlations, with two exceptions. The convergence from the prenatal standard cry videos to video-recall with own infants was lower than stability in responses to standard cry videos for both empathy and negative emotions. As noted below, variability in observed infant distress may account for this difference for negative emotions. Reasons for the difference in empathy are unclear. Nevertheless, of key interest considering the current study goals is the consistent evidence that mothers’ prenatal responses to standard cry stimuli demonstrate significant convergent validity with how mothers respond to their own infants for 8 out of the 9 measures considered. This convergence supports the validity of using mothers’ prenatal reactions to cry stimuli as an indicator of their likely responses to their own infants’ distress.

Furthermore, there was relatively little evidence that maternal or methodological factors moderated the magnitude of convergence. That is, only 2 of 9 tested interactions involving prior experience with infants were significant (i.e., in relation to minimizing and negative attributions). The nature of these interactions was consistent with the view that there would be more convergence over time/across methods among mothers with more experience caring for infants. That is, mothers’ reactions to crying may change more over the transition to parenthood among those who had little prior experience with infants. Although convergence was lower among mothers with low prior experience, as predicted, the associations reflecting convergence were still significant (for minimizing attributions) or marginally significant (for negative attributions), suggesting this maternal characteristic plays a relatively modest role in altering the validity of the approach.

There was similar modest support for the view that the comparability of the prenatal stranger videos and own infant videos in regard to infant distress, a methodological factor, mattered given two of nine interactions were statistically significant. For both negative emotional reactions and minimizing attributions, convergence was higher among mothers whose infants displayed more distress at 6 months. In this case, convergence was not significant for negative emotions among mothers whose infants displayed lower distress. Given the prenatal videos displayed infants who were highly distressed for a considerable period, we believe greater convergence among mothers whose infants were highly distressed at 6 months reflects greater comparability between the stimuli to which mothers responded if their own infants were higher on observed distress. This difference illustrates perhaps the greatest advantage of the prenatal standard video method; that is, it is easier to create standard stimuli in which infants display high and persistent levels of negative affect than it is to create comparable stimuli for mothers of each unique infant. Doing so would likely require much longer observations of infant behavior and time to identify the ideal segments of video footage to use.

Notably, convergence was not apparent for RSA withdrawal as a main effect or when moderators were considered. The lack of convergence for RSA may in fact reflect methodological differences in that the prenatal measure was taken when mothers were passively sitting and watching a video, whereas the 6-month measure was taken when mothers were mobile and interacting with their infant. Comparability in task features when measuring RSA is advised (Burt & Obradović, 2013). The same is true for the conditions under which SCL was assessed, yet convergence is apparent in that case.

Goal 3: Predictive Validity of Both Approaches in Relation to Sensitivity

The prenatal measures assessed in response to stranger infants tended to predict sensitivity comparably to parallel measures assessed at 6 months postpartum during the video recall of own infant. This was demonstrated by the non-significant Fisher’s r-to-z tests. One exception is that SCL assessed while caring for one’s own infant was a statistically significant and somewhat more robust predictor of sensitivity than was SCL assessed while passively watching the standard prenatal videos. This difference in magnitude may reflect the fact that 6-month SCL was assessed while caregiving; it would be useful to assess physiological arousal and regulation prenatally using the infant doll simulator which affords the opportunity to interact with the doll. Baby simulators have successfully elicited cortisol changes in prior research (Bos et al., 2018), and thus may be viable for eliciting changes in SCL too. Notably, SCL arousal was associated with higher rather than lower sensitivity consistent with our view that some amount of arousal is adaptive in that it suggests the parent is aware of/affected by infant crying, and hence motivated to act (Mills-Koonce et al., 2009). However, consistent with the Yerkes-Dodson law as explained by Bornstein and Manian (2013), extremely high or poorly regulated arousal in response to crying predicts less sensitivity/more harshness (Joosen et al., 2013; Leerkes et al., 2016), law which posits.

Higher accurate distress detection and empathy and lower minimizing and negative attributions were significantly associated with sensitivity during the distress tasks, although the correlations tended to be small in magnitude. These associations are consistent with expectation and prior research on the predictors of parenting (Dix et al., 2004; Leerkes, 2010; Lorber & O’Leary, 2005; Rodriguez et al., 2016). In contrast, negative emotions, situational attributions, efficacy, and RSA withdrawal were not significantly associated with sensitivity. This is not to say these factors are not important. Prior evidence from this and other samples demonstrates that these constructs either operate as valuable indicators of broader latent constructs that do predict maternal sensitivity in the case of situational emotional attributions or predict maternal sensitivity via interactions with other constructs as is the case for negative emotions, efficacy, and RSA withdrawal (Leerkes et al., 2016; Leerkes, 2010; Leerkes et al., 2004).

Limitations and Directions for Future Research

Limitations of this study include that the interview with standard cry stimuli was only re-administered to a subset of mothers at 6 months, and these mothers were somewhat more advantaged economically. This may limit generalizability somewhat, as does the fact that our sample only includes first-time expectant mothers. However, this sample is more economically and ethnically diverse than most studies of this type, and correlations from the subsample were consistent even when FIML was applied, reducing concerns that stability could be biased based on sample characteristics. Another limitation is that standard clip interview at 6 months was administered after the video recall interview which may have altered responses due to contrast effects. Despite these limitations, this work is the first to examine stability of non-behavioral responses to cry stimuli across the transition to parenthood, and only the third to examine convergence with parallel measures in response to own infant (Hechler et al., 2019; Leerkes, 2010). Additional research of this type is warranted, particularly with the doll simulator which is a method gaining popularity (Rutherford, 2019) and is particularly appealing in the ability to make the conditions under which physiological measures are assessed more comparable to actively parenting one’s own infant. Finally, additional work of this type is needed with fathers given only one study exists to date using cry stimuli with expectant fathers and mothers, and some differential prediction to parenting was apparent (Bos et al., 2018; Hechler et al., 2019). In future work, it is important that investigators fully describe the nature of the employed cry stimuli following suggestions made by Gabrieli et al. (2019) both to promote the ability to replicate methods and to provide critical details which may affect the interpretation of results and explain discrepant results across studies.

SYNOPSIS.

Objective.

The current study examines the stability, convergent validity, and predictive validity of assessing pregnant women’s emotional, cognitive, and physiological responses to infant cry.

Design.

In an ethnically diverse sample of 259 first-time mothers and their infants, during the prenatal period physiological arousal (skin conductance) and regulation (respiratory sinus arrythmia) were recorded as mothers were exposed to four 1-min videos of crying infants. After each clip, mothers completed questionnaires and were interviewed about their cognitive (attributions, ability to detect distress, efficacy) and emotional responses (empathy, negative emotions). When infants were 6 months old, mothers’ physiological arousal and regulation were assessed while interacting with their own infants during distress-eliciting tasks, then mothers were interviewed about their emotional and cognitive responses using a video-recall method. A subset of mothers (n = 103) was re-administered the prenatal interview using the standard cry videos. Maternal sensitivity was observed during distress-eliciting tasks when infants were 6 months, 1 year, and 2 years old.

Results.

Mothers’ prenatal responses to cry videos were moderately stable until 6 months postpartum, converged with postnatal measures from own infant stimuli, and illustrated modest predictive validity to maternal sensitivity during the first 2 years that was comparable to predictive validity from mothers’ postpartum responses to their own infants.

Conclusions.

How mothers respond to cry stimuli during the prenatal period is reflective of later responses toward their own infants, and as such prenatal cry stimuli are a useful tool for parenting researchers.

IMPLICATIONS FOR PRACTICE AND APPLICATION.

This research demonstrates that first-time expectant mothers’ responses to cry video stimuli were moderately stable until 6 months postpartum, converged with comparable measures derived in response to their own infants, and demonstrated modest predictive validity to maternal sensitivity during the infants first 2 years of life. Although predictive validity was modest, it was comparable to postnatal measures derived from own infant stimuli. These findings support the view that mothers’ responses to cry stimuli during the prenatal period are reflective of their subsequent responses to their own infants (Pedersen et al., 1996), and may be a valuable method to identify mothers at risk for compromised parenting who would benefit from early parenting intervention/education (Hechler et al., 2019). Furthermore, given many such parenting interventions target parental perspective taking and emotion regulation (Bick & Dozier, 2013; Cassidy et al., 2010; Juffer et al., 2008), administering this method pre and post intervention could be a useful way to assess mechanisms of change and pinpoint areas in which interventions could be further enhanced.