Maternal anxiety during pregnancy predicts infant attention to affective faces

Abstract

Background:

Prenatal maternal anxiety is a known influence on offspring development. General anxiety and pregnancy-related anxiety (a distinct type of anxiety encompassing fears associated with pregnancy) are associated with offspring socioemotional development, with potential consequences for later emotional and behavioral problems. This study examines whether maternal pregnancy-related and general anxiety relate to infant attention to affective faces, a process which plays an integral role in early socioemotional development.

Methods:

Participants included 86 mothers and their 6-month-old infants (56.3% female). Mothers completed measures of pregnancy-related and general anxiety three times through gestation. Infants’ attention to affective faces was assessed with an eye-tracking task during which a series of face pairs were presented (happy, angry, or sad face paired with a neutral face). Overall attention measures included attention-holding (total looking time) and attention-orienting (latency to faces); affect-biased attention measures included proportion of total looking time to emotional faces and latency difference score.

Results:

Higher maternal pregnancy-related anxiety across gestation predicted decreased infant attention-holding to affective faces [F(1,80) = 7.232, p = .009, partial η² = .083]. No differences were found in infant attention-orienting or affect-biased attention.

Limitations:

Reliance on a correlational study design precludes the ability to make causal inferences.

Conclusions:

Maternal pregnancy-related anxiety is an important predictor of child outcomes. We provide novel evidence that pregnancy-related anxiety predicts infant attention to emotional faces, behaviors which have important implications for socioemotional development. Providers may consider pregnancy-related anxiety as a target for screening and treatment that may benefit both pregnant individual and offspring.

Introduction

Fetal development progresses incredibly rapidly, far exceeding the rate at which development occurs during any other period of the lifespan. Nearly 200 billion neurons are produced prior to the 3^rd trimester of pregnancy (Bourgeois, 1997), and there is a 21-fold increase in grey matter volume from the 2^nd trimester to term (Andescavage et al., 2017). The fetal brain is highly susceptible to perturbations in the uterine environment (Monk et al., 2019; Demers et al., 2021; Nevarez-Brewster et al., 2022). The Developmental Origins of Health and Disease (DOHaD) hypothesis suggests that the prenatal period is a sensitive window during which environmental factors can significantly influence the development of fetal systems with lasting consequences for postnatal outcomes (Gluckman & Hansen, 2004; Barker, 1990; Barker, 2004). Fetal exposure to prenatal maternal anxiety has emerged as an influential pathway through which offspring development may be sculpted (Davis & Sandman, 2012; Dunkel Schetter & Tanner, 2012), with a growing body of literature highlighting the importance of prenatal maternal anxiety for later offspring socioemotional outcomes (for a review see Madigan et al., 2018; Rees et al., 2019). However, few studies have investigated how prenatal exposure to maternal anxiety shapes specific behavioral components of socioemotional development in the first year of life, including infant attention (Peltola et al., 2018; Chawarska et al., 2016). Investigating how prenatal anxiety relates to early attentional behaviors is important because attention filters how infants perceive and interact with their surrounding environment, including faces and other emotionally salient stimuli, and is a predictor of later emotional and behavioral outcomes. The current study thus examines the relation between different forms of prenatal maternal anxiety and attention to affective faces during infancy.

Prenatal maternal anxiety and infant socioemotional development

The perinatal period is a sensitive time for both the pregnant individual and the fetus during which poor maternal mental health may alter offspring socioemotional development (Davis & Narayan, 2020). Anxiety during pregnancy is relatively common (Perzow, Hennessey et al., 2021), with the prevalence of self-reported general anxiety symptoms, such as recent feelings of general apprehension and nervousness (Spielberger, 1983), ranging from 18–24% across gestation (Dennis et al., 2017). Pregnancy-related anxiety, a distinct type of anxiety specific to the prenatal period (Dunkel-Schetter, 2009; Rini et al., 1999) broadly encompasses fears associated with pregnancy (e.g., concerns about labor and delivery, changes in physical appearance, health of the developing child, experience with the healthcare system, and social/financial issues surrounding pregnancy; Bayrampour et al., 2016; Dunkel Schetter & Ponting, 2021; Rini et al., 1999). Though sharing some variance with general anxiety (Huizink et al., 2014; Dunkel-Schetter, 1998), pregnancy-related anxiety is a distinct phenomenon that captures unique fears and concerns experienced by the pregnant individual across gestation.

Anxiety in the prenatal period has been broadly linked to offspring socioemotional outcomes (Madigan et al., 2018; López-Morales et al., 2023; Rees et al., 2019; Korja et al., 2017; Mahrer et al., 2020; Thomas et al., 2017). Specifically, general anxiety symptoms during pregnancy are associated with lower positive affect (Coplan et al., 2005), increased negative affect (Davis et al., 2007; Peltola et al., 2017), and greater socioemotional problems (e.g., externalizing, internalizing, and dysregulated behaviors; Porter et al., 2019) in infancy. Consistent with parent report measures, behavioral observation in infancy reveal that prenatal anxiety symptoms predicted increased crying and fussing in response to new stimuli at 4-months (Davis et al., 2004) and during a separation from caregiver at 9-months (de Weerth et al., 2013). Although fewer studies have explored the intergenerational consequences of pregnancy-related anxiety (Leigh & Brunton, 2021; Sinesi et al., 2019), extant research shows that elevated pregnancy-related anxiety during gestation predicts greater infant negative affectivity (Blair et al., 2011; Nolvi et al., 2016) and child internalizing and externalizing symptoms (Ali et al., 2019; Acosta et al., 2019). Notably in these studies, pregnancy-related anxiety remains a predictor even after covarying for general anxiety symptoms in pregnancy. Thus, fetal exposure to general and pregnancy-related anxiety in-utero can initiate a developmental pathway leading to increased vulnerability for poor socioemotional outcomes (Cicchetti & Cohen, 2006; Monk et al., 2019). However, the pathways by which prenatal maternal anxiety may sculpt later socioemotional development remain poorly understood. Therefore, the current study examines whether pregnancy-related and general anxiety relate to infant attention to affective faces, with the goal of providing insight into how adverse fetal experiences may shape individual differences in socioemotional behavior.

Infant attention to faces

Attention to faces emerges early in life and plays a central role in how infants engage with the world and regulate their emotional states (Reynolds & Roth, 2018; Reid et al., 2017). Almost immediately after birth, infants demonstrate a predilection for faces (Farroni et al., 2005; Farroni et al., 2013), orienting more quickly to and looking longer at faces as compared with non-face stimuli (Simpson et al., 2020; Di Giorgio et al., 2012; Gliga et al., 2009; Gluckman & Johnson, 2013). This innate visual preference for faces is highly adaptive as it orients infants towards caregivers, eliciting positive parenting behaviors that facilitate attachment and promote survival (Peltola et al., 2009; Peltola et al., 2015). By 5- to 7-months of age, infants can reliably discriminate between different types of affective facial configurations, including happy, sad, angry, fearful, and neutral faces (Leppänen & Nelson, 2009; Morales et al., 2016; Aran et al., 2022; Ruba & Repacholi, 2019). Shortly after, patterns of preferential attention towards salient emotional cues begin to emerge, as infants become quicker to detect emotion faces and look longer at these stimuli in comparison to neutral faces (Burris et al., 2017; Pérez-Edgar et al., 2017). Increased visual attunement to emotionally salient faces is particularly important during the first year of life, as faces provide important contextual clues on the surrounding environment and offer a window into the emotional states of others (Frank et al., 2009).

An emerging literature suggests that individual differences in these attention biases also have important implications for socioemotional outcomes (Field & Lester, 2010; Morales et al., 2016; Todd et al., 2012; Fu & Pérez-Edgar, 2021). For example, over-attunement and difficulty disengaging from threat-cuing faces in infancy has been linked to behavioral inhibition and elevated risk for childhood anxiety disorders (Pérez-Edgar et al., 2010), whereas longer looking time to affective faces (Peltola et al., 2018; Peltola et al., 2015) and a greater attention bias for fearful faces (Eskola et al., 2023) predict positive socioemotional outcomes at later timepoints. Early measurement of variability in infant attention is critical as these biases can persist over time to shape the trajectory of child socioemotional development.

Maternal anxiety and infant attention to affective faces

Few studies have evaluated links between prenatal anxiety and offspring attention to faces. However, studies of postnatal influences suggest that postnatal maternal anxiety is linked to infants’ attention to affective faces (Vallorani et al., 2021; Leppänen et al., 2018; Morales et al., 2017; Vallorani et al., 2023). Further, two studies examining links between the prenatal environment and infant attention to faces using eye-tracking technology found that higher levels of maternal prenatal anxiety or depression was associated with lower infant disengagement from fearful faces (Kataja et al., 2019; Kataja et al., 2020). Given the importance of infant attention for later socioemotional development and evidence that the prenatal period plays an important role in shaping developmental trajectories, further research on the potential link between prenatal maternal and infant attention to affective faces may shed light on a pathway by which prenatal maternal mood influences subsequent child emotional development.

The Current Study

The current study examined associations between maternal pregnancy-related and general anxiety and infant attention to affective faces. Attention was assessed at 6-months because by this time, infants can reliably discriminate between different affective faces, process affective faces at the perceptual and attentional levels, and are just beginning to demonstrate individual differences in visual attending to affective faces (Aran et al., 2022; Burris et al., 2017; Field & Lester, 2010; Morales et al., 2016; Peltola et al., 2009; Pérez-Edgar et al., 2017). We used a free-viewing eye-tracking paradigm, during which stimuli were presented for an extended period for the infant to freely explore (Swales et al., 2023). Similar to the classic visual paired comparison (VPC) task, the free-viewing paradigm used in the current study consists of two face stimuli presented side by side, allowing participants to explore stimuli for a set period of time and disengage at their own volition. Consistent with the VPC task, free-viewing paradigms apply minimal constraints to infant looking behaviors. However, and in contrast to the VPC task, free-viewing paradigms provide the additional advantage of allowing for assessment of both attention-orienting (which supports detection and selection of relevant information among competing stimuli; Cohen, 1972) and attention-holding (which involves the maintenance of attention upon a stimulus to support more detailed processing; Cohen, 1972) within the same task (Armstrong & Olatunji, 2012; Bar-Haim et al., 2007). Metrics gathered from the current free-viewing task have been shown to detect infants’ ability to discriminate between different affective faces and relate to individual differences in temperament (Swales et al., 2023). Additionally, joint assessment of attention-orienting and attention-holding through a free-viewing paradigm may allow for evaluation of how different forms of prenatal anxiety relate to each attentional component. First, we evaluated the potential link between pregnancy-related and general anxiety and infant overall attention to affective faces. Next, we examined if maternal prenatal anxiety is differentially associated with infant attention to affective faces based on the type of emotion presented.

Next, we evaluated the impact of timing of exposure during gestation and sex differences. Evidence demonstrates that both timing of exposure to maternal distress during gestation and change in mood over gestation may be important for offspring outcomes (Glynn et al., 2001; Buss et al., 2010; Davis & Sandman, 2010; Rini et al., 1999; Glynn et al., 2018). Pregnancy is a dynamic period of rapid change. Profiles of pregnancy-related anxiety change over gestation, with those early in gestation more concerned about viability and those later in gestation more concerned about childbirth (Davis & Sandman, 2010; Buss et al., 2010; Dunkel Schetter & Ponting, 2021). Additionally, sensitive windows for fetal maturation may lead to differential consequences based on timing of gestation. For example, some studies have shown that exposure to prenatal anxiety early in gestation may be more consequential for development (Buss et al., 2010; Davis & Sandman, 2010), and though others have shown effects later in gestation (Rini et al., 1999). For this reason, we evaluated whether observed associations between maternal pregnancy-related and general anxiety and infant attention to affective faces differed based on timing of exposure. Additionally, as an extensive literature documents sex-specific responses to the in-utero environment (Sandman et al., 2013; Davis & Pfaff, 2015; Clifton, 2010), we assessed whether sex moderated any observed associations between pregnancy-related and general anxiety and infant attention to affective faces.

Methods

Study Overview

Pregnant individuals provided ratings of pregnancy-related and general anxiety at 3 timepoints (early-pregnancy: M_GA = 16.01 weeks, SD = 4.40; mid-pregnancy: M_GA = 28.28 weeks, SD = 4.10; late-pregnancy: M_GA = 35.81 weeks, SD = 1.50). At 6-months postpartum (M_age = 6.36 months, SD = 0.56) mothers provided ratings of general anxiety and demographic information, and infants participated in the eye-tracking task. All study procedures were reviewed and approved by the Institutional Review Board for the Protection of Human Subjects at the University of Denver and the University of Colorado Anschutz Medical Campus, and all pregnant individuals provided informed consent for themselves and their infant.

Participants

Participants included 86 mother-infant dyads. Pregnant individuals were recruited prenatally from obstetrics and gynecology clinics in the Denver Metro Area as part of the Care Project, a longitudinal study and randomized clinical trial examining the effect of interpersonal psychotherapy on prenatal depression and offspring outcomes (Davis et al., 2018; Hankin et al., 2023). Individuals randomized to the treatment group of the parent study were not included in current analyses. Inclusion criteria were less than 25 weeks gestational age, 18–45 years old, proficient in English, and carrying a singleton intrauterine pregnancy. Exclusion criteria included current illicit drug or methadone use, major health conditions requiring invasive treatments (e.g., dialysis, blood transfusions, chemotherapy), past or current symptoms of psychosis or mania based on the Structured Clinical Interview for the DSM-5, and/or current participation in cognitive behavioral therapy or interpersonal psychotherapy. See Table 1 for sample characteristics.

Table 1.

Demographic Characteristics

Maternal Characteristics	M (SD) or %
Age at delivery	31.59 (5.36)
Obstetric complications	.86 (.80)
Annual household income ($)*	84500 (68839)
Household INR*	3.7 (3.75)
Percent living below federal poverty line	27%
Cohabitating with partner*	82.6%
Education (highest degree earned)
High school	12.8%
Some college	14.0%
College degree	46.5%
Graduate degree	26.7%
Race and ethnicity
American Indian	5.8%
Asian	3.5%
Black	12.8%
Hispanic/Latinx	16.3%
Non-Latinx White	53.5%
Other	8.1%
Substance use during pregnancy
Illicit drugs	0%
Marijuana	2.3%
Alcohol	2.3%
Cigarettes	2.3%
Infant Characteristics
Age at eye-tracking assessment (months)	6.36 (.56)
Biological sex at birth (% female)	57%
Race and ethnicity
American Indian	4.7%
Asian	2.3%
Black	5.8%
Hispanic/Latinx	19.8%
Non-Latinx White	47.7%
Multiracial/Multiethnic	19.7%
Birth Outcomes
Gestational age at birth (weeks)	38.85 (1.84)
Birth weight percentile	42.36 (24.85)

Notes.

^*Median reported

Measures

Pregnancy-Related Anxiety.

Pregnancy-related anxiety was measured with the Pregnancy-Related Anxiety Scale (PRAS; Rini et al., 1999) which measures a pregnant individual’s feelings about their health during pregnancy, the health of their baby, and labor and delivery. Participants rated each item using a 4-point Likert Scale ranging from 1 (not at all) to 4 (very much) with scores ranging from 10–40. Examples of items include “I am concerned or worried about the health of my baby” and “I am concerned or worried about developing medical problems during my pregnancy.” Internal consistency in the current sample ranged from .79-.82. Pregnancy-related anxiety scores at each timepoint were highly intercorrelated (r’s ranging from .71-.79, p’s < .001), and scores decreased across gestation (Table 2), with levels early in gestation significantly higher than late in gestation (t(78) = 2.66, p = .01).

Table 2.

Descriptive Statistics

	M ± SD, Range
Maternal Anxiety
PRAS (averaged across all timepoints)	18.44 ± 4.46, 10.67–32.67
T1 PRAS	18.74 ± 4.84, 10.00–35.00
T2PRAS	18.10 ± 5.10, 10.00–34.00
T3 PRAS	17.49 ± 4.51, 10.00–35.00
STAI (averaged across all timepoints)	35.78 ± 12.40, 20.67–68.67
T1 STAI	36.91 ± 13.91, 20.00–77.00
T2 STAI	34.76 ± 13.29, 20.00–72.00
T3 STAI	32.84 ± 11.75, 20.00–69.00
Latency to Faces (in ms)
Happy Trials	491.9 ± 158.6; 302.6–1410.2
Angry Trials	490.4 ± 134.2; 294.5–1079.4
Sad Trials	505.7 ± 180.4; 329.2–1365.6
Total Looking Time to Faces (in ms)
Happy	2692.1 ± 736.2; 577.8–4207.4
Angry	2686.3 ± 678.2; 987.2–4085.6
Sad	2553.0 ± 764.1; 838.0–4141.4

General Anxiety.

Prenatal general anxiety was assessed using the 20-item State Anxiety subscale of the State Trait Anxiety Inventory (STAI; Spielberger, 1983). Participants were asked to rate their experiences of anxiety-related symptoms over the past week using a 4-point Likert Scale ranging from 1 (not at all) to 4 (very much so). Scores ranged from 20–80, with higher scores indicating more anxiety, and a cutoff score of 40 indicating probable clinical levels of anxiety (Emons et al., 2019). Internal consistency in the current sample ranged from .95-.96. General anxiety scores at each timepoint were highly intercorrelated (r’s ranging from .70-.85, p’s < .001), and scores decreased across gestation (Table 2) with levels early in gestation significantly higher than in late gestation (t(76) = 2.59, p = .01).

Demographic Variables.

Participants reported on their race, ethnicity, educational attainment, whether they were cohabitating with a partner, and income-to-needs ratio (INR). INR was calculated by dividing the total reported household income by the poverty threshold corresponding to the number of people living in the household for the year income was recorded (US Census Bureau, 2023).

Pregnancy and Birth Outcomes.

Medical records were used to gather information on prenatal obstetric complications, birth outcomes, and infant biological sex. Gestational age was determined using the American College of Obstetricians and Gynecologists (ACOG) guidelines by early ultrasound measures and date of last menstrual period. Birthweight percentile (accounting for GAB and biological sex) was calculated. Finally, a score for obstetric complications was derived to indicate the presence or absence or pregnancy-related complications (including prenatal infection, pregnancy-induced hypertension, gestational diabetes, and other complications; Hobel, 1982).

Attentional Processing.

Infants’ attentional processing was assessed using a free-viewing eye-tracking paradigm (Swales et al., 2023). This paradigm enabled us to jointly assess attention-holding and attention-orienting, as well as assess both overall attention (i.e., total looking time to faces, latency to faces) and affect-biased attention (i.e., proportion of total looking time to emotional faces; latency difference score). See Eye Tracking Metrics section below for more details.

Stimuli.

Stimuli consisted of 18 color photographs (20×24cm) of affective (happy, sad, angry, neutral) faces pulled from the NimStim set (Tottenham et al., 2002). A happy, sad, or angry facial stimulus was presented simultaneously beside a neutral facial stimulus from the same actor. Over the course of the experiment, six face pairs for each of the three emotional states were presented twice (36 experimental trials). Additional stimuli details are presented in Supplement.

Apparatus.

Stimuli were presented using Tobii Pro Lab software (Tobii Technology, Stockholm, Sweden), and infant gaze was measured using the Tobii Pro Spectrum eye-tracker (150 Hz sampling rate). Infants were seated in a highchair at a viewing distance of 60cm from the Tobii Pro Spectrum monitor (presentation resolution = 1920×1080). Infants’ view of the experimenter and caregiver was occluded with a black curtain to minimize distraction.

Calibration.

The eye tracker was calibrated using Tobii’s 5-point calibration program. During calibration, an infant-friendly attention-grabbing stimulus was presented at five distinct locations on the screen. The experimenter recalibrated any location that was unsuccessfully or poorly calibrated. The average estimated accuracy in the measurement of infant gaze was 0.49° (SD = 0.21°).

Procedure.

After calibration, infants viewed 36 experimental face pair trials, presented in a fixed randomized order across two blocks of 18 trials of unique face pairs. Stimuli were counterbalanced to account for side presentation and presented for 5 seconds. In between trials, attention getters were employed to draw infants’ attention to the center of the screen. The experimenter advanced to the next trial when infants attended to the fixation cue, which was monitored by live gaze tracking in the experimenter monitor. The task was ended if mother requested or if the infant cried for more than 20 seconds. See Figure 1 for a depiction of the eye-tracking task.

Figure 1.

Sample Task Trials

Areas of Interest (AOIs).

Areas of interest (AOIs) were drawn to encompass each facial stimulus. AOI size did not differ significantly between angry, happy, and sad trials, F(2,17) = 0.02, p = .966). Dimensions were held constant for the happy, angry, or sad face and neutral face in each face pair.

Data Processing & Cleaning.

The Tobii Fixation Filter classification algorithm was used to determine changes in infant fixation over the course of the task (minimum fixation duration was 60ms). Data were cleaned on a trial-by-trial basis. Participants provided data (defined as looking on screen at least once during a trial) for 96.2% of all trials. Inclusion criteria for further analyses were (1) fixation on screen within the first 500ms of the trial (excluding 2.4% of all trials) and (2) three or more (>25%) usable trials for a given emotion type (excluding one participant with an insufficient number of sad trials). On average, 94% of trials were usable in each condition (happy: median = 12 [IQR 11–12]; angry: median = 12 [IQR 11–12]; sad: median = 12 [IQR 11–12]).

Of the 102 infants who attempted the eye-tracking task, 8 were excluded due to eye-tracker malfunction, 7 were excluded for unsuccessful calibration, and 1 was excluded for insufficient eye-tracking data. Therefore, 86 mother-infant dyads were included in present analyses. Infants included were more likely to have higher INR than infants excluded (t(98) = −2.15, p < .05), but did not differ on any other covariates.

Eye-Tracking Metrics

Total Looking Time to Faces and Proportion of Total Looking Time.

Total looking time to faces and proportion of total looking time were used as measures of attention-holding. Total looking time to faces was used as a measure of overall attention-holding to faces and was defined as the sum of total looking time to both the emotional and neutral faces during each trial; this was averaged across trial type (i.e., separately for each of the three emotions; Table 2). Proportion of total looking time, a measure commonly used in classic visual paired comparison tasks (e.g., DeBolt & Oakes, 2022; Richmond et al., 2015; Segal & Moulson, 2020), was used as a measure of affect-biased attention. The proportion was computed for each trial type (happy, angry, sad) by dividing total looking time to the emotional face by the sum of total looking time to the emotional and neutral face during that trial (i.e., when infants have detected both faces, this value represents the proportion of time they spent attending to the emotional face as opposed to the neutral face it was paired with). Internal consistency for total looking time to faces across all trials was 0.95.

Latency to Faces and Latency Difference Score.

Latency to faces and latency difference score were used as indexes of attention-orienting (Wieser et al., 2009). Latency to faces was used as a measure of overall attention-orienting to faces and was generated by computing the time elapsed between stimulus presentation and first fixation on a facial stimulus (either emotional or neutral, excluding latencies <150ms) during each trial; this was averaged across trial type (Table 2). The latency difference score was used as a measure of affect-biased attention. Latency difference score was defined as the difference between the latency to initial fixation on neutral face and the latency to initial fixation on emotional face. A positive score indicates quicker fixation on the neutral face. The latency difference score was calculated using only trials in which infants fixated at least once on each face in the pair. Internal consistency for latency to faces across all trials was 0.77.

Analysis Plan

Assessment of Covariates

Maternal race and ethnicity, parity, obstetric complications, INR, maternal education and cohabitation status, GAB, birthweight percentile, and infant age at assessment were considered as potential covariates (Rini et al., 1999; Kataja et al., 2019; Blackmore et al., 2016). Variables associated with eye-tracking outcomes at p < .05 were included as covariates. Only maternal cohabitation status met covariate criteria and thus was included in all analyses. Maternal postnatal anxiety was included as an a priori covariate in all analyses to determine whether any prenatal associations were independent of postnatal maternal anxiety.

Data Processing & Analyses

Data were processed and analyses were conducted using SPSS version 28 (IBM, 2021). For both anxiety measures missingness was less than 5% and Little’s MCAR test was nonsignificant, indicating that missingness in pregnancy-related and general anxiety occurred at random. Thus, expectation maximization was used to impute missing pregnancy-related and general anxiety scores. Pregnancy-related and general anxiety were correlated across gestation at a level consistent with prior literature (r = .61; p < .001; Huizink et al., 2014; Dunkel-Schetter, 1998) and indicating 37% shared variance. To reduce the number of analyses, primary analyses were conducted with the average pregnancy-related and general anxiety score. See Table 3 for bivariate correlations between prenatal anxiety and infant eye-tracking outcomes. An a-priori power analysis was conducted using G*Power revealing a sample size of 73 is 95% powered to detect an .25 effect size at alpha <.05 based on prior literature (Faul et al., 2007).

Testing Study Aims

To evaluate whether pregnancy-related and general anxiety predicted total looking time to faces and latency to faces, a series of general linear multivariate models with Greenhouse-Geisser correction were fitted, allowing for a full model in every cell and for different slopes and intercepts within each cell. Pregnancy-related and general anxiety were first tested as predictors in separate models. Analyses were repeated with measures of infant affect-biased attention (proportion of total looking time and latency difference score). All analyses included maternal cohabitation status and postnatal anxiety as covariates. If either anxiety measure was associated with an infant outcome, an additional model was conducted including both anxiety measures to determine independent contributions to infant outcomes (Garcia et al., 2021).

If associations were found between either pregnancy-related or general anxiety and infant attention measures, supplementary analyses were performed to examine timing and sex differences. To determine whether associations between pregnancy-related anxiety and general anxiety and eye-tracking metrics varied over gestation, we fitted multilevel models using HLM software. First, linear and quadratic growth curves were included to assess for changes in anxiety across gestation. Next, eye-tracking metrics were added to examine their potential relation to anxiety trajectories across gestation. Finally, if an association was found in any of the above models, an infant biological sex by trial type interaction was evaluated to determine whether this association was moderated by sex.

Results

Attention to Faces

Total Looking Time to Faces

Higher maternal pregnancy-related anxiety predicted lower infant total looking time to faces [between-subjects effect: F(1,81) = 6.172, p = .015, partial η2 = .071; Figure 2A and B]. Notably, pregnancy-related anxiety remained a significant predictor of infant total looking time to faces even after additionally covarying for prenatal general anxiety [F(1,80) = 7.232, p = .009, partial η2 = .083]. Prenatal general anxiety did not significantly relate to infant total looking time to faces [between-subjects effect: F(1,81) = 0.011, p = .915, partial η² < .001]. No within-subjects interactive effects between either anxiety measure and trial type were observed; infants’ total looking time to faces did not differ by emotion type (all F’s < 0.711 and p’s > .492; see Supplement Table 4A).

Figure 2A-B.

Prenatal maternal pregnancy-related anxiety, but not general anxiety, predicted decreased infant total looking time to faces.

Latency to Faces

Neither pregnancy-related nor general anxiety significantly related to infant latency to faces [PRAS: F(1,81) = 2.112, p = .150, partial η2 = .025; STAI: F(1,81) = 2.852, p = .095, partial η2 = .034; Figure 3A and B]. No within-subjects interactive effects between either anxiety measure and trial type were observed; infants’ latency to faces did not differ by emotion type (all F’s < 3.052 and p’s > .05; see Supplement Table 4B).

Figure 3A-B.

Neither prenatal maternal pregnancy-related anxiety nor general anxiety significantly related to infant latency to faces.

Affect-Biased Attention

Neither maternal anxiety measure significantly related to proportion of infant total looking time to faces or latency difference score (see Supplement Tables 5A–B).

Timing of Exposure

HLM analyses used Full Information Maximum Likelihood (FIML) to ensure that participants with at least 1 time point of data would be included in our final model. Gestational age was centered at 8 weeks. As our results demonstrated no difference by emotion type, HLM analyses evaluated whether pregnancy-related anxiety was related to infant total looking time to any emotional face (happy, angry sad), collapsed across trial type.

Pregnancy-related anxiety changed significantly across gestation with deviance scores indicating that a linear slope provided a good fit for its trajectory (Δχ2(11) = 1368.1 – 1311.8 = 56.3, p < .005). PRAS scores decreased linearly across gestation (b = −.188, p = .017). As shown in Figure 4, maternal pregnancy-related anxiety was significantly associated with infant total looking time throughout gestation. Higher PRAS was associated with shorter total looking time at the intercept (b = −1.61, p = .01), though PRAS was not significantly associated with the slope (b = 0.16, p = .244). These results indicate that higher maternal PRAS throughout gestation was associated with decreased infant total looking time to faces, but that gestational timing did not moderate this association (See Supplement Tables 6A and 6B for analyses).

Figure 4.

Higher maternal pregnancy-related anxiety throughout gestation was associated with decreased infant total looking time to faces.

Note: Data were analyzed continuously. The top and bottom SD are graphed here for visualization purposes only.

Sex Differences

Sex at birth did not moderate the association between pregnancy-related anxiety and infant total looking time [F(1,80) = .027, p = .871, partial η2 < .001].

Discussion

Higher maternal pregnancy-related anxiety throughout gestation predicted decreased infant attention-holding to affective faces. This association persisted after consideration of covariates including postnatal anxiety. Infant attention to affective faces is an early emerging process important for socioemotional development and linked with positive socioemotional outcomes in childhood (Peltola et al., 2018; Eskola et al., 2023). Our findings point to the importance of pregnancy-related anxiety as a potent factor for fetal development with potential consequences for offspring attentional processes and highlight an underlying pathway through which prenatal maternal mental health may be linked with offspring socioemotional development.

Maternal pregnancy-related anxiety predicted infant attention-holding to affective faces. Pregnancy-related anxiety is a facet of maternal mental health that has thus far been understudied when compared with other measures of prenatal distress (Dryer & Brunton, 2021). However, pregnancy-related anxiety captures the specific concerns experienced by pregnant individuals as opposed to more generalized measures of anxiety or distress. Furthermore, and unlike other forms of anxiety for which temporary relief from symptoms may be achieved through avoidance of triggers, pregnancy-related anxiety is difficult to avoid as physical, social, and psychological changes may serve as constant reminders (Dryer & Brunton, 2021). Thus, pregnancy-related anxiety may be experienced with greater potency and have greater consequence for fetal development (Blair et al., 2011; Dunkel Schetter & Tanner, 2012). For example, some studies indicate that pregnancy-related anxiety may be a stronger predictor of offspring socioemotional development than general anxiety (Sinesi et al., 2019; Garcia et al., 2021; Blackmore et al., 2016). Interestingly, we did not observe any timing effects, as higher pregnancy-related anxiety throughout the entire course of gestation was associated with infant attention-holding to affective faces. Pregnancy is a dynamic period with trajectories of maternal mental health symptoms as well as fetal maturational processes changing across gestation. Prior work has both observed effects of timing of exposure and links between mood patterns over gestation and offspring outcomes (Davis & Sandman, 2010; Buss et al., 2010; Rini et al., 1999. Thus, it will be important for future research to consider timing of exposure when examining links between prenatal factors and infant socioemotional development. Overall, the current results align with extant literature that demonstrates that elevated pregnancy-related anxiety may increase offspring vulnerability to poor socioemotional development (Garcia et al., 2021, Buss et al., 2010, Davis & Sandman, 2010).

The current findings add to the small literature suggesting that the prenatal environment is associated with early infant attentional processes. Employing a different paradigm, Kataja et al. (2019) showed that prenatal maternal general anxiety was linked with a lower likelihood of disengaging from fearful faces at 8-months. As the current study paradigm did not include fearful faces, a direct comparison is not possible (Leppänen et al., 2018). Nevertheless, these studies indicate that prenatal anxiety may be associated with attention to affective faces and future research is necessary to examine this link. Our findings join a small literature of two other studies (Kataja et al., 2019; Kataja et al., 2020) examining prenatal factors and infant attention to affective faces. Extant literature shows that 7-month-old infants who demonstrated greater attention-holding to affective faces overall, rather than biases toward specific emotion types, showed positive socioemotional outcomes such as more frequent helping responses at 24-months and reduced callous-unemotional traits at 48-months (Peltola et al., 2018). Models explaining how infant attention relates to later socioemotional outcomes suggest that infants who are more responsive to faces may ultimately develop to be more responsive to others (Peltola et al., 2018; Bedford et al., 2015). Thus, it is possible that greater fetal exposure to pregnancy-related anxiety and subsequent decreased attention-holding to all affective faces may indicate an early deviation from typical developmental trajectories, potentially placing the offspring at greater risk for maladaptive socioemotional outcomes.

The mechanisms linking pregnancy-related anxiety and infant attention-holding are unknown. In infancy, regions including the medial prefrontal cortex and anterior cingulate cortex may underlie attention-holding (Petersen & Posner, 2012), and reductions in offspring grey matter volume in these regions have also been associated with greater maternal pregnancy-related anxiety (Buss et al., 2010). Maternal stress hormones and altered maternal immune activation may mediate these relations. Increases in these maternal stress hormones during gestation and altered maternal immune functioning have been associated with both pregnancy-related anxiety (Kane et al., 2014; Ramos et al., 2019; Mancuso et al., 2004; Karlsson et al., 2017) and development of offspring neural circuits (Sandman et al., 2018; Sandman et al., 2011; Buss et al., 2012, Li et al., 2011; Davis et al., 2017; Knuesel et al., 2014). Thus, it is plausible that fetal exposure to higher levels of maternal stress hormones or altered immune activation associated with increased pregnancy-related anxiety may shift neurodevelopment, resulting in alterations to attentional abilities and consequentially, poorer socioemotional outcomes. Finally, future research should investigate whether maternal nutrition during pregnancy may be a pathway by which prenatal maternal mental health alters trajectories of fetal development (Monk et al., 2013; Lindsay et al., 2017; Deer et al., 2023; Cortés-Albornoz et al., 2021).

In the current study, neither form of prenatal maternal anxiety was associated with infant attention-orienting to affective faces. This is consistent with the postnatal literature which has not identified a direct link between maternal anxiety and infant attention orienting (Vallorani et al., 2021). Finally, our findings did not differ by sex. Previous work suggests there may be differential effects of fetal programming by sex (Sandman et al., 2013; Clifton, 2010). Although we may be underpowered to detect sex differences, no other studies have examined potential sex differences in the relation between prenatal factors and infant attention to affective faces, and further research is warranted.

Strengths & Limitations

The current study employs a prospective and longitudinal design to examine the association between prenatal maternal anxiety and infant attention to affective faces in a racially, ethnically, and socioeconomically diverse sample. We examined pregnancy-related anxiety as well as general anxiety across multiple time points during gestation which allowed us to assess whether changes in pregnancy-related anxiety across gestation were related to infants’ attention to affective faces. Additionally, we used a free-viewing design which enabled us to examine both attention-orienting and attention-holding within a single task.

The current study also has several limitations. The stimuli used in the eye-tracking task only included white actors. While race-based biases in infants’ visual processing of faces are not reliably demonstrated at 6-months (Hunter & Markant, 2021; Liu et al., 2015), future studies should use stimuli representative of caregivers, as by 9-months infants show difficulty discriminating between other-race faces (Markant et al., 2016; Kelly et al., 2009). Secondly, the correlational nature of the data prevents us from making causal conclusions about the relation between pregnancy-related anxiety and infant attention to affective faces. As this study relies on natural variations in maternal anxiety rather than experimental manipulations, we cannot rule out the possibility that the current findings are due to shared genes. However, cross-species research and quasi-experimental designs examining stressors such as natural disasters document that there are effects of fetal exposure to prenatal distress that are not genetic in nature (Davis et al., 2017; Glynn et al., 2001; Lafortune et al., 2021; Glover et al., 2014). Finally, as this is one of the first studies to examine how different forms of prenatal maternal anxiety relate to infant attention (but see Kataja et al., 2019), it will be important for future research to replicate the current findings with a larger sample size.

Conclusions & Clinical Implications

Fetal exposure to maternal pregnancy-related anxiety may have long-term implications for offspring health and wellbeing. Our results indicate that infants born to mothers who experienced greater levels of pregnancy-related anxiety across gestation demonstrated decreased attention-holding to affective faces. As exploration into the links between prenatal factors and infant attention is just beginning, these findings should be interpreted cautiously. However, our results are promising as they identify a link between prenatal factors and offspring attention to faces, with potential consequences for later socioemotional development. Although pregnant individuals are not routinely screened for pregnancy-related anxiety (as opposed to general prenatal distress; Dryer & Brunton, 2021), pregnancy-related anxiety has been suggested as an excellent target for intervention because of its time-limited specificity and focus (Garcia et al., 2021). Ultimately, future studies should examine how treating pregnancy-related anxiety may benefit both the pregnant individual and early offspring socioemotional development.

Highlights.

• Maternal pregnancy-related anxiety has intergenerational effects

• Fetal exposure to maternal anxiety and depression predicts child emotion processing

• Infant attention to affective faces is linked with later socioemotional development

• Infants exposed to greater pregnancy-related anxiety look less at affective faces

• Reducing pregnancy-related anxiety may benefit offspring socioemotional development

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author, E.-M. P. H., upon reasonable request.