The t-test showed that there was a significant difference for the joint attention category, both in terms of occurrence and duration. Groups did not differ with respect to the categories “the mother looks at the baby”, “the baby looks at the mother” and dyadic attention (Table 1).

Table 1

Category	Non-PPD	PPD	p-value
Mother look at baby n=42
Occurrence	47.65 (11.17)	51.2 (13.34)	0.35
Duration	357.6 (72.63)	348.14 (100.11)	0.73
Baby look at mother n=40
Occurrence	4.85 (3.88)	5.85 (4.74)	0.47
Duration	8.25 (6.59)	9.2 (7.48)	0.67
Dyadic attention n=42
Occurrence	22.05 (9.8)	20.77 (11.4)	0.70
Duration	105.8 (60.51)	75.77 (46.74)	0.08
Joint attention n=19
Occurrence	1.25 (1.41)	0.55 (0.8)	0.05
Duration	20.27 (14.82)	7.38 (2.77)	0.02

The category “the mother looks at the baby” had the highest average of occurrences and longer duration for both groups, followed by dyadic attention, “the baby looks at the mother” and joint attention. Although no differences were found, it is worth noting that PPD mothers and their babies looked more at each other and for a longer time than the non-depressed dyads.

Given that joint attention was the category that had a significant difference between the two groups, in addition to its importance in child development, a specific analysis of this behavior was performed. Out of the 42 dyads, 19 (8 PPD; 11 non-PPD) engaged in joint attention. The non-PPD group had more events of joint attention (JAT), with longer duration. There was no significant difference between the ages of the babies that engaged in joint attention (JAT) and babies that had no episodes [B_Non-JAT = 18.15 weeks (SD = 2.78); B_JAT= 18.08 weeks (SD= 2.58)]. The dyads that performed joint attention had more episodes of dyadic attention (t (40) = - 3.05; p = 0.00) [M_non-JAT = 17.26 (SD=8.72); M_JAT = 26.37 (SD= 10.63)], with longer duration [ t (40) = - 2.68; p = 0.04) (SD=8.72) (M_non-JAT =70.7 seconds (SD= 51.36); M_JAT = 113.53 seconds (SD= 51.52)]. They also had on average more years of education (t (40) = - 2.17; p = 0.03) [M_JAT =10.1 (SD = 1.24); M_non-JAT = 8.61 (SD = 2.77)].

Three non-PPD mothers had each one non-codable episode (NCE) with a maximum duration of three seconds. Ten PPD mothers had non-codable episodes, with an average of three episodes per dyad, and duration ranging from one to 36 seconds. The EPDS average score of mothers who had non-codable episodes was higher (t(40) = -5.33; p = 0.00) and statistically significant than of mothers that positioned herself and the baby in a way that their faces were visible throughout the entire video [(M_NCE = 15.23 (SD= 6.11); M_non-NCE = 9.9 (SD=7.28)]. There was no significant difference between the groups in terms of years of education.

There was an association between the occurrence of joint attention and the child’s gender (X ² = 4.709; p = 0.03), in this case, the establishment of joint attention was associated with boys. Among the babies who had joint attention, 13 were boys and six were girls. Although no significance test was applied, due to the sample size, some differences that called our attention could be observed. Boys who engaged in joint attention were on average (M = 17.7 weeks; SD = 2.57) younger than girls (M = 18.8 weeks; SD = 2.65), and their mothers had higher EPDS scores [B_{EPDS scores =} 10.6 (SD = 7.05); G_{EPDS scores =} 7.67 (SD = 5.46)]. Nevertheless, the girls looked more at their mothers (M = 12.5; SD = 8.01) and for longer periods (M = 18 seconds; SD = 12.61) than the boys [M = 4.46 (SD = 2.63); M = 7.69 seconds; (SD = 4.28)].

The main objective of this study was to investigate if PPD would influence mother-child visual interaction, especially the joint attention. Our results are similar to those found by Boyd et al. (2006) and by Striano et al. (2002) in which depressed and non-depressed groups did not differ significantly in mother-child gazing behavior. Regarding dyadic attention, the groups had similar patterns both in frequency and duration. Mutual gazing was not impaired due to the maternal depression. These results imply that the less complex behaviors had similar frequency and duration for both groups. This means that depressed mothers and their babies looked at each other and sought visual interaction. However, the findings regarding the effect of PPD on mother-infant visual interactions are inconsistent. Studies state that one of the alterations in the parental function caused by PPD is the face-to-face interaction, resulting in impaired engagement in attention-sharing activities (Goldsmith & Rogoff, 1997). Righetti-Veltema et al. (2002) argue that even milder maternal depression can affect the baby. The infant would perceive the minimal deficiencies of the interaction with his mother, resulting in impaired visual interaction.

Consistent with our hypothesis, non-PPD dyads engaged more in joint attention and their episodes lasted longer. Even at a preliminary stage, this behavior requires more complex operations to happen, therefore it is expected that maternal depression is associated with differences in the ability to establish and maintain joint attention. Goldsmith and Rogoff (1997) also reported that dyads with depressed mothers spent less time engaged in joint attention. Nevertheless, the data in the literature are inconclusive. For example, in the study by Henderson and Jennings (2003), depressed and non-depressed mothers performed equally well on joint attention with their infants.

The mothers who engaged in joint attention had more years of schooling. Studies have already associated low-educated PPD mothers with impairments in some aspect of the interaction with the baby (Norcross et al., 2017; Santos et al., 2020; Slomian et al., 2019). Reilly et al. (2021) found that infants of well-resourced families respond better to cues for joint attention. On the other hand, in a study with mothers from a low socio-economic status, maternal depression didn’t have an effect on joint attention (Gueron-Sela et al., 2018). Studies from our longitudinal project, as well as other studies conducted in similar contexts, show that maternal schooling can be a protective factor for the child (Alvarenga et al., 2018; Defelipe et al., 2017; Morais et al., 2015; Santos et al., 2020; Santos et al., 2021). Greater access to schooling can enable women to have greater cognitive and social support to deal with the difficulties during the postpartum, reducing the probability of being excessively vulnerable to the adverse conditions that they may face.

Depressive symptoms during pregnancy were associated with shorter gestational age at delivery. Previous studies have also found this association (Felder et al., 2017; Jarde et al., 2016; Liu et al., 2016). It is likely that, together with socioeconomic disadvantage, antenatal depression influences the duration of pregnancy. Mothers who had non-codable episodes scored higher on the EPDS. It is possible that depressed mothers feel uncomfortable with the video cameras and try somehow not to be visible.

There was an association between mother-boy dyads and the frequency of joint attention. In contrast to our result, Hirose and Barnard (1997) found that joint attention was more frequent for girls in the context of PPD, but no difference was detected in the non-PPD group. It is likely that gender differences become more evident in the context of maternal depression.

The literature indicates that male infants seem to suffer more from the consequences of postpartum depression and are more neglected by their mothers (Murray et al., 1993; Reck, et al., 2018). Gender differences are widely discussed in research on mother-baby interaction in the context of postpartum depression. The results and the explanations are far from being conclusive. Murray (1992) reported that boys were more vulnerable to the effects of PPD. For example, the speech of depressed mothers directed to male babies was markedly more negative than that of mothers who weren’t depressed (Murray et al., 1993). This was not the case when the baby was female. Reck et al. (2018) described that boys of PPD mothers showed less positive engagement during play episodes compared to the non-PPD group. This effect was not found for girls.

A comparison between boys and girls that had episodes of joint attention showed that the girls looked more at their mothers and for a longer time. One explanation for this result may be linked to the fact that the girls, for not receiving enough attention, sought more diligently the mother’s gaze. PPD mothers had more and longer episodes of the non-codable category. This category was used when the mothers moved in a way that covered her face or the face of the infant, making it impossible to code the gaze direction. This finding may reflect that the depressed mothers moved in a way to hide themselves from the camera.

Our behavioral coding was based on gaze alternation as evidence of joint attention. However, Tomasello (1995) points out that this could lead to false positives, as gaze alternation can be just a sign of alternating attention or confirmation that the partner is paying attention, rather than joint attention. It was not possible to apply more accurate tests to confirm that the dyads were actually sharing the attention, as suggested by Carpenter and Call (2013). This is a limitation of this study, although the evidence obtained is plausible to consider the episodes as joint attention. Even though there are few cases and no general conclusions can be drawn, the results regarding the relationship between episodes of joint attention and the sex of the baby deserve attention and may indicate avenues for future research. Furthermore, studies are needed to examine how the impact of PPD may manifest differently on mother-infant visual interaction and joint attention, depending on the social and cultural context of the families.

In conclusion, this study highlights the potentially disruptive effects of PPD, specifically in a high-risk and low-class sample. The observation of the early joint attention may provide indications of a possible impairment between the dyad. Our results also highlight the differential performance between boys and girls, a research topic that has been discussed for decades, and more recently in the context of PPD.

Our findings show that the influence of PPD on mother-infant interaction may be mediated by the association with sociodemographic and contextual factors. Unfavorable socioeconomic conditions and lower levels of education have the potential to affect maternal care and investment in her children, and can have a negative impact on their development. Pinheiro et al. (2021, p. 21) state that “less schooling alone should be considered a possible factor that indicates the need for special attention in the pregnancy-puerperal cycle, even in those women who do not present cumulative risk factors”. In low income contexts, the health services should regularly evaluate women for PPD, giving them the benefit of health prevention programs and treatment for depression. Because low-income mothers usually have difficulty seeking and finding support, home-based interventions focused on mother-child interactions may be very beneficial in poorer areas.

Our behavioral coding was based on gaze alternation as evidence of joint attention. However, Tomasello (1995) points out that this could lead to false positives, as gaze alternation can be just a sign of alternating attention or confirmation that the partner is paying attention, rather than joint attention. It was not possible to apply more accurate tests to confirm that the dyads were actually sharing the attention, as suggested by Carpenter and Call (2013). This is a limitation of this study, although the evidence obtained is plausible to be considered as episodes of joint attention. The second limitation is the small sample size from only one city in Brazil. Besides that, mothers can interact with their babies differently in a laboratory setting, therefore it is important not to generalize these results. Even though there are few cases and no general conclusions can be drawn, the results regarding the relationship between episodes of joint attention and the sex of the baby deserve attention and may indicate avenues for future research. This study brings contributions to the understanding of postpartum depression and mother-infant interaction in high-risk contexts. Furthermore, analyses associating joint attention and postpartum depression add information to the literature that seeks to understand how socio-affective stressors can influence the early interactional process and the child development. Future studies are needed to examine how the impact of PPD may manifest differently on mother-infant visual interaction and joint attention, depending on the social and cultural context of the families.