Mothers’ Neural and Behavioral Responses to Their Infants’ Distress Cues: The Role of Secure Base Script Knowledge

Abstract

This research examined mothers’ secure base script knowledge—reflected in the ability to generate narratives in which attachment-relevant problems are recognized, competent help is offered, and problems are resolved—and its significance for early-stage processing of infants’ distress cues, using event-related potentials in an emotion oddball task. Mothers with lower secure base script knowledge exhibited (a) a heightened P3b response—reflective of greater allocation of cognitive resources—to their infants’ distressed (but not happy) target facial expressions; (b) a larger P3b response to their infants’ distressed (compared with happy) target facial expressions, which is indicative of allocating disproportional attentional resources to processing their infants’ distress; and (c) poorer accuracy in identifying their infants’ distressed target facial expressions. Findings suggest that mothers’ attachment-relevant biases in processing their infants’ emotion cues are especially tied to infant distress and shed light on underlying mechanisms linking mothers’ attachment representations with sensitive responding to infant distress.

Keywords

adult attachment infant distress event-related potentials information processing attachment script assessment

Infants’ distress signals evolved to promote survival by eliciting caregiving responses (Bowlby, 1969/1982). How caregivers respond to infant distress has predictive significance for children’s socioemotional development (Leerkes, 2011; Leerkes, Blankson, & O’Brien, 2009; McElwain & Booth-LaForce, 2006), and representations of early attachment-relevant experiences predict mothers’ sensitive caregiving (Verhage et al., 2016). Attachment representations are thought to contribute to responding within interpersonal relationships by shaping biases in processing social cues (Bowlby, 1973; Dykas & Cassidy, 2011). Indeed, mothers’ attachment representations are associated with interpretations of and attributions about infant distress (Deoliveira, Moran, & Pederson, 2005; Leerkes et al., 2015). However, research has primarily examined downstream cognitive processes that heighten or minimize early automatic responding to infant distress. In contrast, little is known about the role of mothers’ attachment representations in shaping neural responding implicated in early-stage automatic processing of infant distress, including allocation of attentional resources, an important component of sensitive caregiving (Ainsworth, Blehar, Waters, & Wall, 1978). This study addressed this gap by using event-related potentials (ERPs) to index mothers’ attention to their infants’ distress cues (i.e., facial expressions) as a function of attachment.

Attachment and Social Information Processing

Adults internalize early attachment-relevant experiences in the form of cognitive-affective representations (Main, Kaplan, & Cassidy, 1985), assessed in developmental psychology using narrative-based measures, including the Attachment Script Assessment (ASA; Waters & Rodrigues-Doolabh, 2004) and Adult Attachment Interview (AAI; George, Kaplan, & Main, 1984). Attachment representations are thought to influence how adults obtain, organize, and process social information. Whereas experiences of secure base support from attachment figures promote flexible processing of positive and negative attachment-relevant information in secure individuals, inconsistent or harsh responding to attachment needs contributes to biased processing of attachment-relevant information in insecure individuals (Bowlby, 1980). Indeed, research on attachment-relevant attentional biases indicates that insecure individuals—who dismiss the importance of or become emotionally overwhelmed when discussing attachment-relevant experiences in the AAI—and individuals with lower secure base script knowledge (SBSK)—who produce attachment-relevant narratives lacking a secure base structure in the ASA—exhibit heightened attention to emotional words and images (e.g., social interactions; Maier et al., 2005; Warren et al., 2010; cf. Zeijlmans van Emmichoven, van IJzendoorn, de Ruiter, & Brosschot, 2003). Such evidence arguably reflects a heightened vigilance for emotional cues in insecure individuals, perhaps resulting from a history of negative experiences with attachment figures that contributes to monitoring of the environment for attachment-relevant threats (Maier et al., 2005). The current study extends prior work focused on general emotional cues to examine attachment-relevant, relationship-specific attentional biases in mothers’ processing of infants’ distress cues.

ERPs

ERPs are electrical potentials emanating from the brain and noninvasively recorded on the scalp. The direction, onset, and duration of deflections in the ERP waveform meaningfully reflect specific stages of cognition, making ERPs an ideal tool for elucidating attachment-relevant cognitive biases. The P3b is a well-studied ERP component that occurs approximately 300 ms following a significant stimulus and is prominent over parietal brain regions. The P3b is often examined in oddball tasks, in which participants respond to infrequent target stimuli that are randomly presented in a background of frequently occurring standard and infrequently occurring deviant stimuli. Target stimuli elicit a P3b response, the amplitude of which is proportional to the amount of attentional resources engaged in a task (Polich, 2007).

Although secure mothers exhibit a heightened P3b amplitude to target images of infant faces (vs. nonsocial stimuli), differences in P3b amplitude to target infant emotional expressions according to attachment have not been detected (Fraedrich, Lakatos, & Spangler, 2010; Leyh, Heinisch, Behringer, Reiner, & Spangler, 2016). Importantly, prior research examined mothers’ neural responding to unfamiliar infants. Several studies have demonstrated that mothers exhibit a larger P3b amplitude in response to their own infants than to unfamiliar infants (Grasso, Moser, Dozier, & Simons, 2009; Weisman, Feldman, & Goldstein, 2012). Because mothers allocate fewer attentional resources to processing images of unfamiliar infants, the use of unfamiliar infants in prior research might explain the failure to detect attachment-relevant differences in mothers’ neural processing of infants’ emotion cues. The current study addressed this limitation by targeting images of mothers’ own infants.

Current Study

We examined the significance of mothers’ attachment representations for attention allocation to their infants’ distress cues. Mothers completed two versions of an oddball task comprising images of their infants’ facial expressions (distressed, happy, neutral) in which either distressed or happy expressions served as targets that mothers responded to. Following evidence that adult attachment is more strongly associated with responding to infant distress (vs. nondistress) vocalizations (Groh & Roisman, 2009), we expected that lower SBSK would be associated with (a) heightened processing of infant cues when attention was directed toward infant distress and (b) less flexible processing of infant cues, as reflected in disproportional allocation of cognitive resources, when attention was directed toward infant distress compared with happiness. Specifically, in the distressed-target condition, mothers with lower SBSK were expected to exhibit a heightened P3b amplitude to their infants’ distressed expressions, compared with mothers who had higher SBSK. Moreover, we expected that P3b amplitude would be especially heightened in response to infants’ distressed compared with happy target facial expressions for mothers with lower (compared with higher) SBSK. Mothers with lower SBSK were also expected to exhibit behavioral responses indicative of heightened attention to distress, including greater accuracy and shorter reaction times in detecting distress targets. In light of claims that temperament influences attachment (Kagan, 1982), negative emotionality in infants might impact mothers’ attachment-relevant processing of their infants’ distress. To test this, we examined whether infant temperament influenced associations between mothers’ attachment and attention to infant distress.

Method

Participants

The sample for this study comprised 70 right-handed mothers (M age = 30 years, SD = 4.36) of 6-month-old infants. All infants were full term, of normal birth weight, and normally developing. Mothers were recruited via announcements at hospital birth education and lactation classes, community events for families, advertisements at child-care centers, and advertisements in university newsletters. Seventy percent of mothers were White/Caucasian, 13% were Asian, 10% were Hispanic, 3% were Black/African American, and 4% were multiracial. Mothers’ education ranged from 1 (high school degree) to 5 (advanced degree), with a mean of 4.19 (SD = 0.98). Median reported family income was $51,000 to $60,000 per year and ranged from less than $10,000 to more than $100,000 per year.

Procedure

Mothers participated in two laboratory visits that were approximately 1 week apart. Images of mothers’ infants displaying distressed, happy, and neutral facial expressions were acquired during the first laboratory visit. Between the first and second laboratory visits, mothers completed the Infant Behavior Questionnaire-Revised (Gartstein & Rothbart, 2003), the measure of temperament used in this study. During the second laboratory visit, mothers completed (a) the ASA (Waters & Rodrigues-Doolabh, 2004) while being digitally recorded and (b) an infant emotion oddball task while being physiologically monitored. Seventy-one mothers participated in both laboratory visits, but because 1 mother’s infant became fussy and the mother was unable to complete the second visit, the mother was excluded from the current sample. Our predetermined target sample size of 70 was based on prior research on electrophysiological correlates of adult attachment (Fraedrich et al., 2010; Groh et al., 2015; Leyh et al., 2016) and available funding. Data collection stopped when we reached the enrollment goal.

Measures

ASA

At the start of the ASA, mothers were given six cards, each with the title of a story and a list of 12 words (organized in three columns). They were asked to tell all six stories (order counterbalanced), three of which concerned children’s relationships (“Baby’s Morning,” “Doctor’s Office,” and “Trip to Park”) and three of which concerned adults’ relationships (“Jane and Bob’s Camping Trip,” “The Accident,” and “An Afternoon Shopping”). The two adult stories (“Jane and Bob’s Camping Trip” and “The Accident”) and two child stories (“Baby’s Morning” and “Doctor’s Office”) designed to tap SBSK were coded. Stories were rated for SBSK using the 7-point scale designed by Waters and Rodrigues-Doolabh (2004). A secure base script is one in which a problem occurs and there is a bid for help, help is offered and useful in overcoming the situation, and the situation returns to normal. Narratives that receive the highest score (7) clearly show this structure, whereas narratives that receive the lowest score (1) lack a clear secure base structure. Importantly, SBSK in the ASA and coherence in the AAI are moderately correlated and share common origins in the caregiving environment (Schoenmaker et al., 2015; Steele et al., 2014; Waters & Waters, 2006). A composite score reflecting SBSK (M = 3.92, SD = 0.89) was created by averaging scores across narratives, and this composite score was used in all analyses (α = .70). The authors are trained coders of the ASA; the first author coded all stories, and the second author coded 20% of the stories for reliability purposes. Interrater reliability was high (intraclass correlation coefficient = .96). SBSK was standardized for all analyses.

Infant emotion oddball task

Images of mothers’ infants displaying distressed, happy, and neutral facial expressions were extracted from video recordings of mother-infant interactions during a laboratory visit that occurred approximately 1 week before mothers completed the oddball task. A distressed facial expression was defined as the infant crying and/or having a contorted mouth and lowered or tightened brows. A happy facial expression was defined as smiling or laughing. A neutral facial expression was characterized by relaxed facial features and the absence of distressed and happy features (see Fig. 1). Twenty research assistants rated the valence of the distressed, happy, and neutral images for all 70 infants on a 5-point Likert-type scale ranging from 1 (very happy) to 3 (neutral) to 5 (very distressed). Mean ratings for the happy, neutral, and distressed images were 1.66 (SD = 0.34), 3.05 (SD = 0.11), and 4.63 (SD = 0.30), respectively. Importantly, mothers’ SBSK was not significantly associated with emotional valence ratings for any of the facial expressions (happy: r = −.09, p = .462; neutral: r = .09, p = .478; distressed: r = .04, p = .736).

Fig. 1.

Example photos of infant expressions used in the emotion oddball task: happy, neutral, and distressed (from left to right).

Each mother viewed images of her own infant displaying distressed, happy, and neutral facial expressions in two blocks of 160 trials each. Images appeared one at a time in random order, with the distressed and happy expressions each appearing 25% of the time and the neutral expression appearing the other 50%. The distressed facial expression served as the target stimulus in one block (distressed condition) and the happy facial expression served as the target stimulus in the other block (happy condition; block order was counterbalanced). For each condition, mothers were instructed to indicate targets as quickly and as accurately as possible by pressing a button with their dominant hand. At the beginning of each block, mothers completed a practice round of 10 trials. Images were presented for 600 ms each, separated by an interstimulus interval of 1,500 ms with a random jitter of 200 ms. Each image was positioned in the center of a screen 60 cm in front of the mother at a visual angle of 9° × 10°.

Electroencephalogram recording

Continuous electroencephalographic (EEG) activity was recorded using an ActiveTwo head cap and the ActiveTwo BioSemi system (BioSemi, Amsterdam, The Netherlands). Recordings were taken from 32 scalp electrodes arranged according to the 10-20 system and 2 electrodes placed on the left and right mastoids. Electrodes were placed below the right and left eyes and near the outer canthus of each eye for recording vertical and horizontal electrooculograms. As per BioSemi’s design, the ground electrode during acquisition was formed by the Common Mode Sense active electrode and the Driven Right Leg passive electrode. All bioelectric signals were digitized on a laboratory computer using ActiView software (BioSemi). Sampling was at 512 Hz.

Off-line analysis was performed using ElectroMagnetic Source Estimation Suite Data Editor software (Version 5.5; Source Signal Imaging, San Diego, CA). EEG data were re-referenced to the numeric mean of the mastoids and band-pass filtered, with cutoffs of 0.1 and 30.0 Hz. The EEG was segmented for each trial, beginning 100 ms before and ending 1,300 ms after stimulus onset. An algorithm was implemented using ElectroMagnetic Source Estimation software to remove blinks from the EEG. Trials contaminated by muscle artifacts or blinks occurring within 200 ms after stimulus presentation were excluded (M = 5% for each trial type in each condition). Average ERP waveforms, locked to image onset, were computed for each trial type within each condition at each electrode. The P3b component was defined as the average activity in a 300- to 500-ms window following stimulus presentation. Consistent with evidence that the P3b component associated with allocation of attentional resources is maximal over the centroparietal scalp region (Polich, 2007), P3b amplitude was most prominent over the Pz scalp site (see the Supplemental Material available online), and thus, the Pz electrode was examined in analyses.

Performance data

Two measures of performance were computed for each condition type (distressed target, happy target): (a) average reaction time for correct-response target trials and (b) total number of misses (failures to make a response to a target). These measures of performance were computed to examine the relation between mothers’ SBSK and their speed and accuracy in identifying their infants’ emotional expressions.

Infant temperament

Mothers completed the Infant Behavior Questionnaire-Revised (Gartstein & Rothbart, 2003), a 191-item measure designed to assess infant temperament along 14 dimensions (activity level, distress to limitations, approach, fear, duration of orienting, smiling and laughter, vocal reactivity, sadness, perceptual sensitivity, high-intensity pleasure, low-intensity pleasure, cuddliness, soothability, and falling reactivity). Because of the specific focus of this study, we combined the dimensions that prior factor-analytic research indicates load onto a broadband temperamental dimension reflecting negative emotionality (α = .68, consisting of sadness, distress to limitations, fear, and falling reactivity [reverse-scored]; Gartstein & Rothbart, 2003). Infants’ negative temperamental emotionality was standardized for all analyses.

Statistical analyses

We examined the influence of the following potential covariates: mother ethnicity, family income, and mother education. We also examined the influence of performance data (e.g., number of missed responses) in ERP analyses. Because the direction and significance of effects did not change when we controlled for these variables, they were not included in the analyses presented below. ASA data were missing for 1 participant, and performance data were missing for 1 participant because of technical problems. Missing data were imputed using a single-imputation equation-modeling algorithm. Three mothers’ P3b amplitudes and 1 mother’s total number of missed targets were 3 standard deviations above the mean. To reduce undue influence of these scores while retaining all participant data, we Winsorized outliers, preserving the original order (Tabachnik & Fidell, 2001).

To account for the nested structure of the data, we used hierarchical linear modeling (Version 7.0; Bryk & Raudenbush, 1992). An omnibus model was tested to examine SBSK and negative temperamental emotionality in relation to P3b amplitude and whether associations depended on target condition and infant emotional expression. Specifically, a two-level model with P3b amplitude as the dependent variable was used, in which emotion (distressed, happy, neutral) and condition (distressed target, happy target) were entered as within-subjects variables at Level 1, SBSK and temperament were entered as between-subjects variables at Level 2, and all possible interactions were entered. A significant interaction between SBSK, condition, and emotion was probed in two ways. To test the first hypothesis that the association between mothers’ SBSK and P3b amplitude would be specific to when mothers’ attention was directed toward their infants’ distressed (but not happy) facial expressions, we ran separate models for each condition (distressed target, happy target). To test the second hypothesis that mothers’ SBSK would be associated with differential P3b amplitudes when attention was directed toward their infants’ distressed compared with happy facial expressions, we ran a model comparing mothers’ responding as a function of target type (distressed, happy). Infant temperament was included in these models to further examine potential child-driven effects. To examine SBSK and negative temperamental emotionality in relation to performance, we ran separate models for each performance indicator (reaction time, total missed targets) with condition (distressed target, happy target) as a within-subjects variable, SBSK and temperament as between-subjects variables, and their interactions entered as a cross-level interaction. To further evaluate predictions regarding the specificity of associations to infant distress cues, we used Bayes factors (BFs). This approach provides a principled measure of the relative evidence from data for various models, including those that embed the null hypothesis. Thus, BFs may be used to state positive evidence for the lack of an effect, which is not possible in conventional significance testing used in the hierarchical-linear-modeling analyses described above (Rouder & Morey, 2012; Rouder, Morey, & Wagenmakers, 2016). BFs were calculated using an online calculator developed by Rouder and Morey (2012) on the basis of linear models from Lang, Paulo, Molina, Clyde, and Berger (2008). BFs indicate the extent to which one model is favored over another model; a BF of 1.8 indicates that a model is favored over another nested model at a ratio of 1.8:1 (Rouder et al., 2016).

Results

Mothers’ attachment and ERP data

In the omnibus model predicting mothers’ P3b amplitude, there were significant main effects for condition and emotion (see Table 1), and adding condition, emotion, and their interaction to the model decreased Level 1 residual variance by 42%. P3b amplitude was larger in the distressed-target condition (M = 11.40 μV, SD = 6.69) than in the happy-target condition (M = 11.38 μV, SD = 5.54). In addition, images of infants displaying distressed facial expressions elicited the largest P3b amplitude (M = 14.09 μV, SD = 6.61), followed by images of happy (M = 12.08 μV, SD = 5.55) and neutral (M = 8.00 μV, SD = 4.42) facial expressions. These main effects were qualified by a significant two-way interaction between condition and emotion that was further qualified by a significant three-way interaction between SBSK, condition, and emotion (see Table 1). Adding SBSK to the model decreased Level 2 variance by 1%. To probe this significant three-way interaction, first, we examined whether the association between mothers’ SBSK and P3b amplitude to their infants’ facial expressions was specific to the distressed-target condition; second, we examined whether SBSK was associated with differential P3b amplitudes to distressed versus happy target facial expressions. The main and interactive effects of infants’ negative temperamental emotionality were not significant, and adding temperament to the model did not decrease Level 2 variance, suggesting that the interaction between SBSK, condition, and emotion in relation to P3b amplitude was independent of infant temperament. To further rule out the role of temperament in influencing associations between SBSK and P3b amplitude, we retained it in follow-up analyses.

Table 1.

Results of Hierarchical Linear Models Predicting Mothers’ P3b Amplitude to Infants’ Emotional Facial Expressions

Model and predictor	b	SE	t ratio	Approxi-mate df	p
Full model
Intercept	5.30	0.52	10.19	66	< .001
SBSK	−0.19	0.44	−0.44	66	.666
Temperament	−0.07	0.47	−0.16	66	.877
SBSK × Temperament	0.14	0.45	0.31	66	.759
Condition	3.04	0.30	10.01	338	< .001
SBSK × Condition	−0.51	0.29	−1.78	338	.077
Temperament × Condition	−0.19	0.28	−0.68	338	.469
SBSK × Temperament × Condition	−0.48	0.32	−1.47	338	.142
Emotion	3.04	0.21	14.15	338	< .001
SBSK × Emotion	−0.25	0.16	−1.60	338	.111
Temperament × Emotion	−0.08	0.18	−0.47	338	.639
SBSK × Temperament × Emotion	0.03	0.12	0.24	338	.814
Condition × Emotion	−1.53	0.15	−10.31	338	< .001
SBSK × Condition × Emotion	0.28	0.13	2.17	338	.031
Temperament × Condition × Emotion	0.08	0.14	0.55	338	.585
SBSK × Temperament × Condition × Emotion	0.17	0.14	1.23	338	.218
Distressed-target-condition model
Intercept	2.27	0.67	3.37	66	.001
SBSK	0.32	0.52	0.61	66	.545
Temperament	0.12	0.55	0.21	66	.831
SBSK × Temperament	0.62	0.49	1.26	66	.211
Emotion	4.57	0.32	14.38	136	< .001
SBSK × Emotion	−0.54	0.25	−2.19	136	.030
Temperament × Emotion	−0.16	0.29	−0.56	136	.578
SBSK × Temperament × Emotion	−0.14	0.22	−0.65	136	.515
Distressed-versus-happy-target model
Intercept	15.67	0.66	23.61	66	< .001
SBSK	−1.14	0.57	−2.01	66	.048
Temperament	−0.20	0.75	−0.26	66	.795
SBSK × Temperament	−0.08	0.65	−0.12	66	.907
Target Type	1.13	0.25	4.53	66	< .001
SBSK × Target Type	−0.48	0.22	−2.15	66	.035
Temperament × Target Type	0.14	0.25	0.58	66	.567
SBSK × Temperament × Target Type	0.28	0.22	1.29	66	.201

Note: All models included secure base script knowledge (SBSK) and temperament (negative temperamental emotionality) as predictors. In addition, the full model included condition (distressed target, happy target) and emotion (distressed, happy, neutral) as predictors, the distressed-target-condition model included emotion as a predictor, and the distressed-versus-happy-target model, which examined only the data for when the target stimulus was presented in each condition, included target type (distressed, happy) as a predictor.

Is the association between mothers’ attachment and P3b amplitude specific to when attention was directed toward infant distress?

Focusing first on the distressed-target condition, as seen in Table 1, there was a significant main effect of emotion, and adding emotion to the model decreased Level 1 residual variance by 61%. Consistent with the well-documented three-stimulus oddball effect (Polich, 2007), distressed facial expressions (target stimuli) produced the largest P3b amplitude (M = 16.80 μV, SD = 6.69), followed by happy facial expressions (deviant stimuli; M = 9.74 μV, SD = 4.96) and then neutral facial expressions (standard stimuli; M = 7.66 μV, SD = 4.41). All main and interactive effects for infants’ negative temperament were not significant, and adding temperament to the model did not decrease Level 2 variance. Consistent with hypotheses, the main effect for emotion was qualified by a significant interaction between SBSK and emotion, and adding SBSK to the model decreased Level 2 variance by 2%. Following up this significant interaction, we ran a simple-slopes analysis (Aiken & West, 1991; Preacher, Curran, & Bauer, 2006), which indicated that, as expected, SBSK was associated with mothers’ P3b amplitude to their infants’ distressed facial expressions, b = –1.29, SE = 0.55, t(136) = –2.33, p = .021. As seen in Figure 2, mothers with lower SBSK exhibited a larger P3b amplitude—indicative of greater resource allocation—when attending to their infants’ distressed facial expressions, compared with mothers who had higher SBSK. As expected, mothers’ SBSK was not associated with P3b amplitude to their infants’ happy, b = –0.75, SE = 0.42, t(136) = –1.80, p = .074, or neutral, b = –0.22, SE = 0.40, t(136) = –0.54, p = .590, facial expressions.

Fig. 2.

Average event-related potentials for mothers who were 1 standard deviation above the mean (top) and 1 standard deviation below the mean (bottom) on secure base script knowledge for the distressed-target condition.

We probed these results by calculating BFs. Consistent with the findings presented above, in predicting mothers’ P3b amplitude to their infants’ distressed expressions in the distressed-target condition, findings favored the model including SBSK over the null model (BF = 1.49) and models including temperament (BF = 3.10) and the interaction between attachment and temperament (BF = 7.68). As expected, in predicting mothers’ P3b amplitude to their infants’ happy expressions in the distressed-target condition, findings favored the null model (BF null model vs. model with SBSK = 3.95, BF null model vs. model with SBSK and temperament = 4.75, BF null model vs. model with SBSK, temperament, and their interaction = 10.68). The null model was also favored when predicting mothers’ P3b amplitude to their infants’ neutral expressions in the distressed-target condition (BF null model vs. model with SBSK = 2.53, BF null model vs. model with SBSK and temperament = 6.88, BF null model vs. model with SBSK, temperament, and their interaction = 12.58).

In the happy-target condition, there was a significant main effect of emotion, b = 3.04, SE = 0.27, t(136) = 11.18, p < .001; happy target expressions elicited the largest P3b (M = 14.41 μV, SD = 5.14), followed by distressed deviant expressions (M = 11.38 μV, SD = 5.32) and neutral standard expressions (M = 8.34 μV, SD = 4.43). Adding emotion to the model decreased Level 1 residual variance by 53%. All main and interactive effects for infants’ negative temperament were not significant, and adding temperament to the model did not decrease Level 2 variance. As expected, the interaction between SBSK and emotion was not significant, b = –0.07, SE = 0.23, t(136) = –0.30, p = .767, and adding SBSK knowledge to the model did not decrease Level 2 variance.

Consistent with the findings presented above, in predicting mothers’ P3b amplitude to their infants’ distressed facial expressions in the happy-target condition, findings from BFs favored the null model (BF null model vs. model with SBSK = 2.86, BF null model vs. model with SBSK and temperament = 6.68, BF null model vs. model with SBSK, temperament, and their interaction = 14.40). The null model was also favored for happy expressions (BF null model vs. model with SBSK = 2.12, BF null model vs. model with SBSK and temperament = 6.14, BF null model vs. model with SBSK, temperament, and their interaction = 14.79) and for neutral expressions (BF null model vs. model with SBSK = 2.25, BF null model vs. model with SBSK and temperament = 5.33, BF null model vs. model with SBSK, temperament, and their interaction = 12.93).

Is maternal attachment associated with differential P3b amplitudes when attention is directed toward infants’ distressed compared with infants’ happy expressions?

As seen in Table 1, in the model comparing mothers’ P3b amplitude when they were instructed to attend to target images of their infants’ distressed versus happy facial expressions, there was a significant main effect of target type, and adding target type to the model decreased Level 1 residual variance by 20%. Distressed targets elicited a larger P3b response (M = 16.80 μV, SD = 6.69) than happy targets (M = 14.41 μV, SD = 5.14). All main and interactive effects for infants’ negative temperament were not significant, and adding temperament to the model did not decrease Level 2 variance. Consistent with hypotheses, there was a significant interaction between SBSK and target type, and adding SBSK to the model decreased Level 2 variance by 2%. As seen in Figure 3, a simple-slopes analysis revealed that mothers who have lower SBSK exhibited a significantly larger P3b amplitude to distressed expressions than to happy expressions, b = 1.65, SE = 0.38, t(66) = 4.40, p < .001. In contrast, mothers with higher SBSK did not exhibit significant differences in their P3b amplitude in response to distressed expressions compared with happy expressions, b = 0.60, SE = 0.35, t(66) = 1.72, p = .090.

Fig. 3.

Consistent with the findings presented above, in predicting mothers’ P3b amplitude to their infants’ distressed versus happy expressions, findings from BFs favored the model including SBSK over the null model (BF = 1.08) and the models including temperament (BF = 2.79) and the interaction between attachment and temperament (BF = 4.85).

Mothers’ attachment and performance data

As seen in Table 2, in the model predicting mothers’ total number of missed responses to infants’ emotional cues, although the effect for condition was not significant, findings revealed a significant interaction between mothers’ SBSK and condition. Adding condition to the model contributed to a 3% decrease in Level 1 residual variance, and adding SBSK to the model contributed to a 1% decrease in Level 2 variance. Because the interaction was predicted a priori, we proceeded with probing the interaction. Simple-slopes analysis revealed that when attending to their infants’ distressed facial expression, mothers with lower SBSK had a greater total number of missed responses than mothers with higher SBSK, b = –0.09, SE = 0.04, t(66) = –2.22, p = .030. Thus, contrary to expectations, mothers with lower SBSK were less accurate in identifying their infants’ distress cues. As expected, no differences were found for mothers’ accuracy in identifying happy targets as a function of SBSK, b = 0.03, SE = 0.05, t(66) = 0.56, p = .576. Similar to ERP findings, all main and interactive effects for infants’ negative temperament were not significant. The model for reaction time did not reveal any significant main or interactive effects.

Table 2.

Results of the Hierarchical Linear Model Predicting Mothers’ Number of Missed Responses to Infants’ Distressed and Happy Facial Expressions

Predictor	b	SE	t ratio	Approxi-mate df	p
Intercept	0.23	0.05	4.24	66	< .001
SBSK	−0.03	0.04	−0.82	66	.412
Temperament	−0.01	0.05	−0.28	66	.784
SBSK × Temperament	−0.02	0.04	−0.41	66	.683
Condition	−0.07	0.04	−1.85	66	.069
SBSK × Condition	−0.06	0.03	−2.19	66	.032
Temperament × Condition	0.01	0.04	0.36	66	.718
SBSK × Temperament × Condition	0.03	0.03	1.41	66	.258

Note: The predictors in this analysis were secure base script knowledge (SBSK), temperament (negative temperamental emotionality), and condition (distressed vs. happy expression).

Consistent with the findings presented above, in predicting mothers’ total number of missed responses to their infants’ distressed expressions, findings from BFs favored the model including SBSK over the null model (BF = 2.05) and the models including temperament (BF = 3.16) and the interaction between attachment and temperament (BF = 7.28). As expected, in predicting mothers’ total number of missed responses to their infants’ happy expressions, findings favored the null model (BF null model vs. model with SBSK = 3.83, BF null model vs. model with SBSK and temperament = 10.16, BF null model vs. model with SBSK, temperament, and their interaction = 20.90). Consistent with the findings presented above, in predicting mothers’ reaction time to their infants’ distressed expressions, findings favored the null model (BF null model vs. model with SBSK = 3.83, BF null model vs. model with SBSK and temperament = 2.73, BF null model vs. model with SBSK, temperament, and their interaction = 6.86). Findings also favored the null model in predicting mothers’ reaction time to their infants’ happy expressions (BF null model vs. model with SBSK = 4.07, BF null model vs. model with SBSK and temperament = 11.34, BF null model vs. model with SBSK, temperament, and their interaction = 23.21).

Discussion

We investigated mothers’ early-stage processing of their infants’ facial expressions to examine the significance of mothers’ attachment representations for attentional biases in processing infant distress. Findings revealed that mothers lower (vs. higher) on SBSK exhibited a larger P3b amplitude—indicative of greater allocation of cognitive resources (Polich, 2007)—when attention was directed toward their infants’ distressed expressions. As expected, no differences in P3b amplitude were found according to SBSK when attention was directed toward happy expressions, providing new evidence that attachment-related differences in responding to infant cues are specific to infant distress. Prior research indicates that P3b amplitude is sensitive to stimulus meaning for individuals (Johnson, 1993). Thus, our findings suggest that infant distress is an especially salient emotional cue for mothers with lower SBSK. Emotion systems are theorized to operate as associative networks whereby emotional stimuli that correspond to significant mental representations activate memory systems (Lang, 1994). Thus, the heightened sensitivity of mothers with lower SBSK to their infants’ distress might reflect a match between the emotional cue and mothers’ representations of attachment-relevant experiences. Indeed, longitudinal evidence indicates that lower SBSK in adulthood is predicted by experiences of less sensitive caregiving in childhood (Schoenmaker et al., 2015; Steele et al., 2014). Thus, a history of insensitive responding to attachment needs might heighten the personal significance of negative attachment-relevant cues because they activate painful representations of early negative caregiving.

Attachment-related differences in P3b amplitude were not found during the happy-target condition in which infant distress was the deviant stimulus, suggesting that attachment-related differences in mothers’ neural responding to infant distress were not stimulus driven. In this experiment, deviant stimuli were not task relevant. Thus, mothers needed to process the expression only so far as to withhold, rather than initiate, a response. In contrast, because target stimuli were task relevant, they elicited the greatest amount of processing capacity. Thus, our findings suggest that processing demands are important in eliciting attachment-relevant cognitive biases. When demands were low, mothers allocated limited cognitive resources to processing their infants’ distress regardless of attachment. When demands were high, such that mothers had to attend and respond to their infants’ distressed expressions, mothers’ attachment representations played a role in the allocation of cognitive resources to processing their infants’ distress. Such evidence that task condition, not facial stimuli alone, was important in eliciting differences in mothers’ neural processing of infants’ facial expressions highlights the importance of context and the deployment of attentional resources in understanding attachment-relevant differences in mothers’ processing of infant distress. Research might further explore the role of context and attention in attachment-related attentional biases by manipulating processing demands, for example, by employing oddball tasks in which mothers either passively view or attend to infant expressions.

Experiences of secure base support, in which caregivers sensitively respond to intense emotions, promote individuals’ flexible emotion processing. Thus, individuals with access to a well-elaborated secure base script are expected to process positive and negative attachment-relevant cues in a flexible, nonbiased manner, whereas individuals without access to a secure base script are expected to exhibit biased processing of attachment-relevant information, especially negative emotional cues (Bowlby, 1980; Dykas & Cassidy, 2011). Our findings support this theorizing. For mothers with higher SBSK, P3b responses to distressed and happy targets were comparable, suggesting comparable allocation of cognitive resources to processing of infants’ negative and positive emotions. In contrast, mothers with lower SBSK exhibited larger P3b responses when instructed to attend to their infants’ distressed (vs. happy) expressions, suggesting a disproportional allocation of resources to processing infant distress. Such evidence suggests that prior research focused on mothers’ responding to infants within either distressing or nondistressing caregiving contexts might miss important variation in mothers’ attachment-relevant responding and that greater insight might be gained by examining not only within-context but also between-context variation.

Contrary to hypotheses, our results showed that mothers’ reaction time did not differ as a function of SBSK, and mothers with lower (vs. higher) SBSK were less accurate in identifying infants’ distress cues. Given that mothers with lower SBSK exhibited neural activity indicative of greater allocation of cognitive resources when attending to infant distress, it is unlikely that lower accuracy is due to attentional deficits. Instead, findings might suggest that heightened allocation of cognitive resources when attending to infant distress leaves mothers with limited resources for organizing a behavioral response. Such evidence might help elucidate underlying processes accounting for attachment-caregiving linkages (Verhage et al., 2016). Specifically, mothers with limited access to a secure base script might respond less sensitively to infant distress because such cues overwhelm their cognitive resources, undermining their ability to organize sensitive behavioral responses.

These findings raise the question of why, in prior research, mothers’ attachment representations were not associated with P3b amplitude when attending to infants’ emotional expressions (Fraedrich et al., 2010; Leyh et al., 2016). A key difference between this study and prior work concerns the use of images of mothers’ own versus unfamiliar infants. Although it could be argued that characteristics of mothers’ infants might explain our findings, evidence that mothers’ attachment-relevant processing of infant distress did not vary according to infants’ temperament provides little support for this interpretation. Alternatively, prior research indicates that mothers’ P3b amplitude is larger in response to their own (vs. other) infants (Grasso et al., 2009; Weisman et al., 2012). Similarly, in research documenting attachment-related differences in mothers’ responses to unfamiliar infants’ distress vocalizations, manipulations are employed that heighten the personal relevance of distress cues (e.g., instructing mothers to imagine how they would respond if the infant were their own; Groh & Roisman, 2009; see also Ablow, Marks, Feldman, & Huffman, 2013; Schoenmaker et al., 2015). Such evidence suggests that the personal significance of infant cues is important in eliciting attachment-relevant differences and heightened neural responding, which might explain differences between this study and prior research. Future research might test this by comparing the significance of mothers’ attachment representations for neural responding with their own versus unfamiliar infants’ distress cues.

This research elucidates underlying mechanisms placing mothers at risk for providing less sensitive care to their distressed infants as a function of attachment. Findings revealed that mothers’ attachment history plays a role in how the brain responds when attending to their infants’ distress, providing evidence that mothers with limited access to a secure base script exhibit neural responding reflective of biased processing of their infants’ distress. Findings also highlight the utility of incorporating ERPs into behavioral research on attachment-related cognitive biases in processing emotion cues.

Footnotes

Action Editor

Eddie Harmon-Jones served as action editor for this article.

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Supplemental Material

Additional supporting information can be found at

References

Ablow

J. C.

Marks

A. K.

Feldman

S. S.

Huffman

L. C.

(2013). Associations between first-time expectant women’s representations of attachment and their physiological reactivity to infant cry. Child Development, 84, 1373–1391. doi:10.1111/cdev.12135

Aiken

S. L.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: SAGE.

Ainsworth

M. D. S.

Blehar

M. C.

Waters

Wall

(1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale, NJ: Erlbaum.

Bowlby

(1973). Attachment and loss: Vol. 2. Separation. New York, NY: Basic Books.

Bowlby

(1980). Attachment and loss: Vol. 3. Loss. New York, NY: Basic Books.

Bowlby

(1982). Attachment and loss: Vol. 1. Attachment. New York, NY: Basic Books. (Original work published 1969)

Bryk

A. S.

Raudenbush

S. W.

(1992). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage.

Deoliveira

Moran

Pederson

D. R.

(2005). Understanding the link between maternal adult attachment classifications and thoughts and feelings about emotions. Attachment & Human Development, 7, 153–170. doi:10.1080/14616730500135032

Dykas

M. J.

Cassidy

(2011). Attachment and the processing of social information across the life span: Theory and evidence. Psychological Bulletin, 137, 19–46. doi:10.1037/a0021367

10.

Fraedrich

E. M.

Lakatos

Spangler

(2010). Brain activity during emotion perception: The role of attachment representation. Attachment & Human Development, 12, 231–248. doi:10.1080/14616731003759724

11.

Gartstein

M. A.

Rothbart

M. K.

(2003). Studying infant temperament via the Revised Infant Behavior Questionnaire. Infant Behavior & Development, 26, 64–86. doi:10.1016/S0163-6383(02)00169-8

12.

George

Kaplan

Main

(1984). Adult Attachment Interview protocol. Unpublished manuscript, University of California, Berkeley.

13.

Grasso

D. J.

Moser

J. S.

Dozier

Simons

(2009). ERP correlates of attention allocation in mothers’ processing faces of their children. Biological Psychology, 81, 95–102. doi:10.1016/j.biopsycho.2009.03.001

14.

Groh

A. M.

Roisman

G. I.

(2009). Adults’ autonomic and subjective emotional responses to infant vocalizations: The role of secure base script knowledge. Developmental Psychology, 45, 889–893. doi:10.1037/a0014943

15.

Groh

A. M.

Roisman

G. I.

Haydon

K. C.

Bost

McElwain

Garcia

Hester

(2015). Mothers’ electrophysiological, subjective, and observed emotional responding to infant crying: The role of secure base script knowledge. Development and Psychopathology, 27, 1237–1250. doi:10.1017/S0954579414000881

16.

Johnson

(1993). On the neural generators of the P300 component of the event-related potential. Psychophysiology, 30, 90–97. doi:10.1111/j.1469-8986.1993.tb03208.x

17.

Kagan

(1982). Psychological research on the human infant: An evaluative summary. New York, NY: W. T. Grant Foundation.

18.

Lang

Paulo

Molina

Clyde

M. A.

Berger

J. O.

(2008). Mixtures of g priors for Bayesian variable selection. Journal of the American Statistical Association, 103, 410–423. doi:10.1198/016214507000001337

19.

Lang

P. J.

(1994). The varieties of emotional experience: A meditation on James-Lange theory. Psychological Review, 101, 211–221. doi:10.1037/0033-295X.101.2.211

20.

Leerkes

E. M.

(2011). Maternal sensitivity during distressing tasks: A unique predictor of attachment security. Infant Behavior & Development, 34, 443–446. doi:10.1016/j.infbeh.2011.04.006

21.

Leerkes

E. M.

Blankson

A. N.

O’Brien

(2009). Differential effects of sensitivity to infant distress and non-distress on social-emotional functioning. Child Development, 80, 762–775. doi:10.1111/j.1467-8624.2009.01296.x

22.

Leerkes

E. M.

Supple

A. J.

O’Brien

Calkins

S. D.

Haltigan

J. D.

Wong

M. S.

Fortuna

(2015). Antecedents of maternal sensitivity during distressing tasks: Integrating attachment, social information processing, and psychobiological perspectives. Child Development, 86, 94–111. doi:10.1111/cdev.12288

23.

Leyh

Heinisch

Behringer

Reiner

Spangler

(2016). Maternal attachment representation and neurophysiological processing during the perception of infants’ emotional expressions. PLOS ONE, 11(2), Article e0147294. doi:10.1371/journal.pone.0147294

24.

Maier

M. A.

Bernier

Pekrun

Zimmermann

Strasser

Grossmann

K. E.

(2005). Attachment state of mind and perceptual processing of emotional stimuli. Attachment & Human Development, 7, 67–81. doi:10.1080/14616730500039606

25.

Main

Kaplan

Cassidy

(1985). Security in infancy, childhood, and adulthood: A move to the level of representation. Monographs of the Society for Research in Child Development, 50, 66–104. doi:10.2307/3333827

26.

McElwain

N. L.

Booth-LaForce

(2006). Maternal sensitivity to infant distress and nondistress as predictors of infant-mother attachment security. Journal of Family Psychology, 20, 247–255. doi:10.1037/0893-3200.20.2.247

27.

Polich

(2007). Updating P300: An integrative theory of P3a and P3b. Clinical Neurophysiology, 118, 2128–2148. doi:10.1016/j.clinph.2007.04.019

28.

Preacher

K. J.

Curran

P. J.

Bauer

D. J.

(2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437–448. doi:10.3102/10769986031004437

29.

Rouder

J. N.

Morey

R. D.

(2012). Default Bayes factors for model selection in regression. Multivariate Behavioral Research, 47, 877–903. doi:10.1080/00273171.2012.734737

30.

Rouder

J. N.

Morey

R. D.

Wagenmakers

(2016). The interplay between subjectivity, statistical practice, and psychological science. Collabra, 2, Article 6. doi:10.1525/collabra.28

31.

Schoenmaker

Juffer

van IJzendoorn

M. H.

Linting

van der Voort

Bakermans-Kranenburg

M. J.

(2015). From maternal sensitivity in infancy to adult attachment representations: A longitudinal adoption study with secure base scripts. Attachment & Human Development, 17, 241–256. doi:10.1080/14616734.2015.1037315

32.

Steele

R. D.

Waters

T. E. A.

Bost

K. K.

Vaughn

B. E.

Truitt

Waters

H. S.

. . . Roisman

G. I.

(2014). Caregiving antecedents of secure base script knowledge: A comparative analysis of young adult attachment representations. Developmental Psychology, 50, 2526–2538. doi:10.1037/a0037992

33.

Tabachnik

B. G.

Fidell

L. S.

(2001). Using multivariate statistics (4th ed.). Boston, MA: Allyn & Bacon.

34.

Verhage

M. L.

Schuengel

Madigan

Fearon

R. M. P.

Oosterman

Cassibba

. . . van IJzendoorn

M. H.

(2016). Narrowing the transmission gap: A synthesis of three decades of research on intergenerational transmission of attachment. Psychological Bulletin, 142, 337–366. doi:10.1037/bul0000038

35.

Warren

S. L.

Bost

K. K.

Roisman

G. I.

Levin Silton

Spielberg

J. M.

Engels

A. S.

. . . Heller

(2010). Effects of adult attachment and emotional distractors on brain mechanisms of cognitive control. Psychological Science, 21, 1818–1826. doi:10.1177/0956797610388809

36.

Waters

H. S.

Rodrigues-Doolabh

(2004). Manual for decoding secure base narratives. Unpublished manuscript, Department of Psychology, State University of New York at Stony Brook.

37.

Waters

H. S.

Waters

(2006). The attachment working models concept: Among other things, we build script-like representations of secure base experiences. Attachment & Human Development, 8, 185–198. doi:10.1080/14616730600856016

38.

Weisman

Feldman

Goldstein

(2012). Parental and romantic attachment shape brain processing of infant cues. Biological Psychology, 89, 533–538. doi:10.1016/j.biopsycho.2011.11.008

39.

Zeijlmans van Emmichoven

I. A.

van IJzendoorn

M. H.

de Ruiter

Brosschot

J. F.

(2003). Selective processing of threatening information: Effects of attachment representation and anxiety disorder on attachment and memory. Development and Psychopathology, 15, 219–237. doi:10.1017/S0954579403000129

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB