Different authors, such as Cassidy (1994) and Bridges and Grolnick (1995), argue that infants learn to express and regulate their emotions, particularly negative ones, in a manner that allows their attachment needs to be met by their caregivers. For the most part, infants with a secure attachment experience parental sensitivity in response to a broad range of emotional signals, which promotes open and flexible communication of both positive and negative affect and flexible regulation of emotion based on the demands of the situation (de Wolff and van Ijzendoorn, 1997; Fuertes et al., 2006, 2009, 2020; Lucassen et al., 2011; Barbosa et al., 2019, 2021). Many insecure-avoidant infants may experience parental intrusiveness, which promotes the defensive minimization of affect, and fewer caregiver-oriented (e.g., proximity seeking, looking toward the caregiver, asking the caregiver for assistance) and more self-oriented (e.g., self-soothing, self-distraction) regulation behaviors to prevent additional controlling behavior (Isabella and Belsky, 1991; Swanson et al., 2000; Cantero and Cerezo, 2001; Fuertes et al., 2006, 2009, 2020; Barbosa et al., 2019, 2021). Insecure-resistant infants, in contrast, may experience maternal passivity or unresponsivity, which promotes the maximization or heightening of affect and frequent use of caregiver-oriented regulation behaviors to gain the caregiver’s attention (Cantero and Cerezo, 2001; Fuertes et al., 2006, 2009, 2020; Barbosa et al., 2019, 2021). These patterns, which serve different functions in the context of the parent-infant relationship, are thought to become internalized and then generalized to other contexts in which they may be less adaptive (Bowlby, 1969/1982; Ainsworth et al., 1978; Goldsmith and Alansky, 1987; Bretherton, 1990; Sroufe, 1996; de Wolff and van Ijzendoorn, 1997; Cassidy and Shaver, 1999; Crittenden, 2000; Bakermans-Kranenburg et al., 2003; Bretherton and Munholland, 2008; Beebe et al., 2010; Verhage et al., 2016).

Models of attachment and social information processing suggest that the attention allocated to salient social information, such as facial expressions, might occur in schema-driven processing, according to infants’ attachment patterns (Field and Lester, 2010; Dykas and Cassidy, 2011; Johnson and Chen, 2011; Morales et al., 2016; Gonçalves et al., 2023). Previous studies have shown that recurrent attachment-related experiences shape representational, physiological, and behavioral responses to emotional information (Pollak et al., 2000, 2001; Pollak and Kistler, 2002; Pollak and Sinha, 2002; Dykas and Cassidy, 2011; Vrtička et al., 2012; Vrtička and Vuilleumier, 2012; Cooke et al., 2016, 2018). To our knowledge, Peltola et al. (2015) were the first authors to investigate infants’ attention to facial expressions in association with attachment security, while others have addressed the relationship between caregiver-infant attachment security and infants’ visual preferences using other social–emotional stimuli (see Johnson et al., 2007, 2010; Biro et al., 2015, 2017). They used a longitudinal design and studied infants’ normative developmental bias to fearful faces at 7 months (Peltola et al., 2008, 2009a,b, 2011, 2013; Forssman et al., 2014a) as an antecedent of later attachment quality at 14 months (Peltola et al., 2015). Consistent with the models reviewed above, the main finding was a smaller attentional bias to fearful faces in infants with an insecure attachment pattern vs. secure infants, and this effect was most clearly associated with attachment disorganization, suggesting altered sensitivity to threat-related cues in infancy as a testable trait linking attachment disorganization to later behavioral outcomes (for a review about the outcomes of threat-related attention bias in socio-emotional development please see Pérez-Edgar et al., 2017; Burris et al., 2019a,b; Fu and Pérez-Edgar, 2019). In the same vein, but using neurophysiological measures, Peltola et al. (2018a,b), measured event-related potentials (ERPs) to investigate whether cortical responses to facial expressions of fear were associated with the development of secure and insecure patterns of infant-caregiver attachment during the first year. Based on previous findings showing reduced attentional biases to fearful faces in infants with insecure and disorganized attachment, the authors hypothesized that insecure and disorganized attachment would be associated with reduced ERP differentiation of fearful from non-fearful faces. ERPs to facial expressions were measured at 7 months of age and attachment was assessed at 14 months of age with the Strange Situation Procedure. The results revealed that occipitotemporal face-sensitive ERP responses particularly in the time range of the N290 component were related to attachment security at 14 months. Furthermore, only securely attached infants showed age-typical cortical discrimination of fearful from non-fearful faces at 7 months, whereas a similar pattern of ERP responses was not observed in infants with insecure and disorganized attachment. These results add to previous findings by suggesting that patterns of secure and insecure infant attachment are related to early-emerging differences in the perceptual processing of facial emotions, which could have implications for the development of social competence. More recently, Forslund et al. (2019) found that disorganized children (6 to 7-year-old children) showed lowered attention to facial expressions, a diminished ability to discriminate facial expressions, and elevated emotional reactivity.

However, the specific developmental pathways that link attachment and socio-emotional development are still a topic of research (Belsky, 1997, 2005; Belsky and Pluess, 2009; Ellis et al., 2011; Cassidy et al., 2013; Paschall and Mastergeorge, 2016; Groh et al., 2017; Slagt et al., 2018; Behrendt et al., 2019). In the present work, we expected to contribute to the line of research investigating the relation between the quality of early social–emotional experiences and social information processing in infancy (e.g., Striano et al., 2002; de Haan et al., 2004; Taylor-Colls and Pasco Fearon, 2015; Kungl et al., 2017; Morales et al., 2017; Kataja et al., 2018; Juvrud et al., 2019; Kataja et al., 2019), specifically how infants visually explore facial expressions of emotion. Our primary goal was to test the hypothesis of potential differences in attention to facial expressions and emotional discrimination (sensitivity to emotional change) between infants classified as securely attached (B), insecure-avoidant (A), and insecure-resistant (C), and between infants grouped as securely (B) or insecurely attached (non-B).

Following the literature reviewed, we identified four major limitations/literature gaps which were addressed in the present study. First, the existing developmental literature has focused on normative attentional patterns according to age, namely, age-expected attentional bias (e.g., to fearful faces at 7-months of age) (e.g., Peltola et al., 2015). In this study, we go beyond the developmental normative attention bias according to age, to consider general visual preferences and their emotional correlates. Second, previous studies have failed to study attachment in a discriminating way, only considering the secure vs. insecure attachment pattern contrast (e.g., Peltola et al., 2018a,b). Here, we consider not only secure vs. insecure but consider A-B-C attachment patterns. We compare all three primary attachment groups other than combining the resistant and avoidant infants into one insecure group, which is critically important as there are different theoretically-derived predictions about the pattern of attention to emotional stimuli that insecure-avoidant and insecure-resistant infants may utilize. Third, several studies have used behavioral measures such as looking time, typically measured by recording cumulative gaze to a whole visual scene, which has been one of the most used behavioral indices of infants’ cognitive and perceptual capabilities (e.g., Johnson, 2004). Here, we used an eye-tracking system that allows for much more accurate measurement of how attention is allocated than general measures of looking time, through greater spatial and temporal precision in quantifying eye movements than video encoding by human scorers (see Papageorgiou et al., 2014; Leppänen et al., 2015; Crawford et al., 2016) and including the analysis of two different measures of infants visual preferences—mean FT and mean FC—, following the suggestion of LoBue et al. (2020) of using multiple outcome measures in infant research. Fourth, most developmental studies of emotion perception are not designed to distinguish whether infants discriminate visual affective features, such as valence (pleasant vs. unpleasant) and arousal (high vs. low) (Barrett et al., 2019). They tend to mix arousal and valence contrasts, or to present valence contrasts ignoring their arousal levels, into impure comparisons that need to be disentangled. For instance, previous studies have compared a pleasant instance of emotion (smiling in happiness) with one that is intended to depict an unpleasant instance of emotion (e.g., scowling in anger, or gasping in fear), ignoring their levels of arousal (e.g., Montague and Walker-Andrews, 2001; Leppänen et al., 2007, 2009). In this sense, we will try to disentangle this by selecting facial stimuli that differ in emotional valence (happy, angry) and the level of arousal (low, high). Therefore, we aimed to analyze if the impact of attachment security on infants’ visual preferences was independent of the emotional contrast presented, by including in our experimental paradigm three different emotional contrasts, namely, valence, arousal, and valence + arousal (following recent debates in the field of Facial Expressions of Emotion; see Almeida et al., 2016; Barrett et al., 2019; Pereira et al., 2019).

Based on the literature reviewed above (Isabella and Belsky, 1991; Swanson et al., 2000; Cantero and Cerezo, 2001; Fuertes et al., 2006, 2009, 2020; Barbosa et al., 2019, 2021), we hypothesized that: (a) since insecure-avoidant infants may experience consistent maternal intrusiveness, they may learn to attend away from feelings of distress and/or social-emotional stimuli to inhibit or prevent expression of distress when frightened, developing an avoidance attentional biases to social stimuli—irrespective of emotional valence—as a function of early interactions with the mother, and (b) contrarily, insecure-resistant infants are likely to learn to attend excessively to social and/or emotional stimuli, developing a hypervigilance attentional bias to social stimuli, due to the higher prevalence of passive or unresponsive maternal behaviors, feeling a higher need to monitor the environment. However, except for the studies developed by Peltola et al. (2015, 2018a,b), no previous study has examined the association between the processing of facial expressions and attachment security in infants. Therefore, our specific aims were to evaluate (1) 14-month-old infants general emotion discrimination abilities (using eye-tracking visual parameters of Fixation Time and Fixation Count, FT and FC, hereafter); (2) their emotion discrimination (FT and FC) as a function of attachment pattern (A, B, C); and (3) their emotion discrimination (FT and FC) as a function of attachment security (secure vs. insecure, B vs. non-B).

In the present study, we found significantly higher mean FT and FC for the new face when compared with the old face, evaluated in the 6 experimental VPC trials, regardless of emotional condition (valence, arousal, valence + arousal contrasts), indicating that 14-month-old infants can discriminate novel emotional expressions. This finding was expected according to the developmental literature regarding the ontogeny of facial emotion processing, where behavioral studies have shown that infants’ ability to visually discriminate emotional expressions emerges around 5–7 months of age (e.g., Nelson and Dolgin, 1985; Nelson, 1987; de Haan and Nelson, 1998; de Haan, 2001; Kotsoni et al., 2001; Frank et al., 2009; Leppänen and Nelson, 2009; Frank et al., 2014).

Models of attachment and socio-emotional information processing have indicated that infants’ security in the primary caregiver availability and responsivity might influence the attention allocated to socio-emotional stimuli, namely, facial expressions of emotion (Field and Lester, 2010; Dykas and Cassidy, 2011; Johnson and Chen, 2011; Morales et al., 2016). Indeed, we found a significant main effect of attachment pattern on infants’ mean fixation time toward facial expressions emotional change, independent of the nature of that facial expression change (in each emotional contrast), suggesting that attachment is related to social information processing. The present results converge to suggest that patterns of secure and insecure infant attachment are related to differences in perceiving and attending to facial emotions, as has also been suggested in studies with children (e.g., Laible and Thompson, 1998; Pollak and Kistler, 2002; Pollak and Sinha, 2002; Bosmans et al., 2007; Steele et al., 2008; Vandevivere et al., 2014; Peltola et al., 2015; Forslund et al., 2017; Meinz et al., 2017; Peltola et al., 2018a,b), adolescents (e.g., Cassidy et al., 2003), and adults (e.g., Mikulincer, 1998; Mikulincer et al., 2000, 2002; Dewitte et al., 2007; Dewitte and De Houwer, 2008; Edelstein and Gillath, 2008).

Furthermore, post hoc tests revealed that there were significant differences between infants insecure-avoidant (A) and insecure-resistant (C), and between secure infants (B) and insecure-resistant (C). Overall, differences [old face-new face] were larger for attachment pattern C compared to both A and B, indicating that insecure-resistant children remain hypervigilant toward emotional change. Insecure-resistant infants show a preference for the new face, through a more negative value for the mean difference in the contrast [old face-new face] for both FT and FC, revealing a higher sensitivity to the novelty effect (an increased sensitivity to faces and facial expressions of emotion). Insecure-resistant infants’ attention deployment can be characterized by an overall hypervigilance visual tendency, as they seem to be more aware or hypervigilant to the new stimulus, suggesting a higher facial emotional discrimination (sensitivity to emotional change). The group of resistant infants is the only group that, compared with their counterparts, always prefers to look at the new face compared with the old face (across the three emotional contrasts), showing heightened attention toward the new stimuli.

Contrary to the hypervigilant pattern of emotion processing that characterized the resistant attachment pattern, insecure-avoidant infants revealed a positive value for the mean difference [old face-new face] FT, meaning that they tend to look less to the new face in comparison with the habituation face, and thus revealing less sensitivity to emotional change. Possibly using disengaging-deactivating strategies, infants classified as insecure-avoidant did not show the same sensitivity to emotional change, in contrast with infants classified as insecure-resistant (C), as they gave priority to familiar stimuli, which is by both theoretical and empirical work with adults (Niedenthal et al., 2002; Fraley et al., 2006).

Moreover, our results concerning infants’ visual preferences across the specific attachment sub-groups (secure, resistant, and avoidant), are similar to the tendency found in the adult literature (e.g., Mikulincer, 1998; Mikulincer et al., 2000, 2002; Niedenthal et al., 2002; Fraley et al., 2006; Dewitte et al., 2007; Dewitte and De Houwer, 2008; Edelstein and Gillath, 2008). According to adult attachment theory, individual differences in attachment-related anxiety reflect variations in individuals’ vigilance to cues relevant to appraising and monitoring the availability and responsiveness of significant others.

However, despite this study’s contributions, the sample size limits the generalization of our findings. Furthermore, since the assessment of both variables (infants’ attachment and the attention deployed to facial expressions of emotions) happened concurrently, we were not able to make inferences about the direction of the association found, which would be interesting to address prospectively. We must also note that in our design, pleasant and unpleasant expressions were, respectively, collinear with the emotional categories of happiness and anger, which precludes a conclusive interpretation of this factor. However, we hope that our findings come to be complemented by future work on the effects of affective properties across a broader range of emotional categories.

In conclusion, our results are consistent with the notion that variation in perceptual vigilance, and particularly the heightened vigilance to changes in emotional expression, may underlie attachment-related anxiety. It is noteworthy that our findings were generally consistent regardless of the valence and/or arousal (or emotional change resulting from the combination of both) of the emotions being perceived. In other words, highly resistant infants were just as sensitive to changes in positive emotions, such as happiness, as they were to negative emotions, such as anger, as well as to more or less intense emotions (e.g., High Angry Face vs. Low Angry Face). The heightened sensitivity to emotional change in insecure-resistant infants independent of emotional contrast (positive vs. negative, more vs. less intense) is an important finding because it suggests that in appraising the availability and responsiveness of the attachment figure the attachment system does not differentially weigh signals that are indicative of happiness (acceptance, sensitivity) vs. angriness (rejection, control) (see also Zhang and Hazan, 2002).

In this work, we have attempted to understand how different attachment patterns are linked to different types of emotional information processing. Future studies should proceed in advancing this line of research, expanding our findings across other age groups, and integrating different stimuli (e.g., both attachment and non-attachment related) and/or using different experimental protocols or paradigms (the importance of using multiple outcome measures in infant research was recently addressed by LoBue et al., 2020). Multimodal assessment of attention in infants using different tasks or paradigms may facilitate the identification of early behavioral and neurocognitive markers of risk factors for adverse developmental outcomes such as socioemotional difficulties.

The raw data supporting the conclusions of this article will be made available by the authors, upon request, without undue reservation.

The studies involving humans were approved by Comissão de Ética para a Saúde do Centro Hospitalar de São João e Comissão de Ética da Unidade Local de Saúde de Matosinhos. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

JG: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Visualization, Writing – original draft, Writing – review & editing. MF: Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – review & editing. SS: Data curation, Formal analysis, Investigation, Methodology, Software, Writing – review & editing. PL-d-S: Conceptualization, Formal analysis, Investigation, Supervision, Writing – review & editing. FF-S: Conceptualization, Data curation, Formal analysis, Methodology, Supervision, Writing – review & editing.