Results

For directional hypotheses, one‐sided tests were conducted; for non‐directional hypotheses, two‐sided tests were carried out (p < 0.05).

Affective and Behavioural Outcomes

The final sample size for the affective and behavioural analyses was 41 for both infants and caregivers, separately. The results of the inclusion and exclusion blocks are reported below for infants and caregivers, respectively. Exploratory analyses of the reinclusion block are reported in the Supplementary Materials 10.

Infants

To test our hypotheses that infants show less positive emotionality and more negative emotionality and social referencing in exclusion compared to inclusion conditions, two paired samples t‐tests and one Wilcoxon test were performed. In addition, Bayes Factors were computed. The result confirmed no significant difference in positive emotionality (Minclusion = 16.01, SDinclusion = 18.18, Mexclusion = 13.82, SDexclusion = 20.27), (t (40) = 0.73, p = 0.233, d = 0.11, BF = 3.16) and in social referencing (Minclusion = 0.62, SDinclusion = 1.72, Mexclusion = 0.55, SDexclusion = 1.61), (W = 52, p = 1, r = 0.06, BF = 0.17) between inclusion and exclusion conditions. However, infants showed more negative emotionality in exclusion compared to inclusion conditions (Minclusion = 17.03, SDinclusion = 22.41, Mexclusion = 41.37, SDexclusion = 34.38), (t (40) = ‐4.62, p < 0.001, d = ‐0.72, BF = 32296.27). To test the hypotheses that there is a difference in attention‐seeking behaviour and visual attention between inclusion and exclusion conditions, we ran one paired samples t‐test and one Wilcoxon test. The result showed no significant difference in attention‐seeking behaviours (Minclusion = 10.55, SDinclusion = 11.24, Mexclusion = 18.07, SDexclusion = 20.59), (W = 299, p = 0.137, r = 0.22, BF = 1.40), however, Bayes Factors showed anecdotal evidence for the alternative hypothesis. In terms of visual attention, the result revealed that infants showed more visual attention (Minclusion = 85.09, SDinclusion = 13.45, Mexclusion = 75.49, SDexclusion = 20.02), (t (40) = 3.18, p = 0.002, d = 0.50, BF = 12.26) in inclusion compared to exclusion conditions. The behavioural responses of infants are displayed in Figure 2.

FIGURE 2: Figures show the percentage of sum scores of attention‐seeking behaviour, negative emotionality, positive emotionality, social referencing and visual attention in infants in inclusion, exclusion and reinclusion blocks.

Caregivers

To assess whether caregivers exhibit less positive emotionality and more negative emotionality and social referencing in the exclusion compared to the inclusion conditions, one paired samples t‐test and two Wilcoxon tests were conducted. In addition, Bayes Factors were computed. Results showed that caregivers showed more positive emotion (Minclusion = 53.3, SDinclusion = 26.55, Mexclusion = 43.46, SDexclusion = 34.42), (t (40) = 2.32, p = 0.012, d = 0.36, BF = 66.12) in the inclusion compared to the exclusion condition. Moreover, they showed more negative emotionality (Minclusion = 5.99, SDinclusion = 6.33, Mexclusion = 16.9, SDexclusion = 16.29), (W = 78, p < 0.001, r = 0.69, BF = 112529) and social referencing (Minclusion = 5.71, SDinclusion = 6.32, Mexclusion = 16.48, SDexclusion = 15.88), (W = 69, p < 0.001, r = 0.71, BF = 2567.68) in the exclusion compared to the inclusion conditions. To test whether there was a difference in attention‐seeking behaviours and visual attention in the inclusion and exclusion conditions, we ran two Wilcoxon tests. There was no significant difference in attention‐seeking behaviours (Minclusion = 6.95, SDinclusion = 12.05, Mexclusion = 13.27, SDexclusion = 29.69), (W = 126, p = 0.213, r = 0.10, BF = 0.92) between exclusion and inclusion conditions. However, there was more visual attention (Minclusion = 89.63, SDinclusion = 8.63, Mexclusion = 73.68, SDexclusion = 18.82), (W = 811, p < 0.001, r = 0.77, BF = 68591.15) in the inclusion compared to exclusion conditions. The behavioural responses of caregivers are displayed in Figure 3.

FIGURE 3: The percentage of sum scores of attention‐seeking behaviour, negative emotionality, positive emotionality, social referencing and visual attention in caregivers in inclusion, exclusion and reinclusion blocks.

Correlation Analyses

The final sample size for the correlation analyses between infants’ and caregivers’ affective and behavioural responses was 41 infant‐caregiver dyads.

To investigate the correlation between caregiver and infant on positive emotionality, negative emotionality, attention‐seeking behaviour and social referencing, Spearman correlations were computed as the normal distribution was violated. Correlation analyses showed significant correlations of the positive emotionality (rs(39) = 0.26, p = 0.048) and negative emotionality (rs(39) = 0.33, p = 0.015) between caregivers and infants in the inclusion condition. However, there was no significant correlation in attention‐seeking behaviour (rs(39) = 0.14, p = 0.177) and visual attention (rs(39) = 0.08, p = 0.292) between caregivers and infants in the inclusion condition. In the exclusion condition, negative emotionality (rs(39) = 0.34, p = 0.013) and visual attention (rs(39) = 0.30, p = 0.027) significantly correlated. However, positive emotionality (rs(39) = ‐0.00, p = 0.516) and attention‐seeking behaviour (rs(39) = 0.13, p = 0.195) did not significantly correlate in the exclusion condition (Supplementary Materials 11). Regarding social referencing, as most of the infants did not show social referencing, we did not run a correlation analysis for social referencing. Moreover, the results of the correlation analyses for the reinclusion block are provided in Supplementary Materials 11.

Theta Power

Infants

Thirty‐four infants were included in the theta power analysis. The mean and SD of included trials for infants’ analysis and removed ICA components and interpolated channels can be found in Supplementary Materials 3.

As expected, paired‐samples t‐tests showed a higher theta power during the “exclusion” compared to the “not my turn” events in infants (Mnot my turn = 8.44, SDnot my turn = 0.98, Mexclusion = 8.94, SDexclusion = 1.00), (t (33) = ‐2.16, p = 0.018, d = 0.37, BF = 43.76). There was no significant difference in theta power during “exclusion” events compared to “inclusion” events (Minclusion = 8.63, SDinclusion = 1.00, Mexclusion = 8.94, SDexclusion = 1.00), (t (33) = ‐1.44, p = 0.078, d = 0.24, BF = 10.65). However, the effect size for the inclusion‐to‐exclusion comparison was smaller than for the exclusion‐to‐not my turn comparison, possibly due to movement confounds (Figure 4).

FIGURE 4: The theta power (purple background) in infants (a) and caregivers (b) during inclusion, exclusion and not my turn trials.

Caregivers

Thirty‐seven caregivers were included in the theta power analysis The mean and SD of included trials for caregivers’ analysis and removed ICA components and interpolated channels can be found in Supplementary Materials 3.

Theta power was analysed using paired samples t‐tests. The results revealed no difference between the “not my turn” (Mnot my turn = 5.99, SDnot my turn = 1.04, Mexclusion = 5.90, SDexclusion = 0.84), (t (36) = 0.46, p = 0.678, d = ‐0.07, BF = 0.48) and “inclusion” events (Minclusion = 5.96, SDinclusion = 0.72, Mexclusion = 5.90, SDexclusion = 0.84), (t (36) = 0.40, p = 0.657, d = ‐0.06, BF = 0.53) compared to the “exclusion” events in caregivers (Figure 4).

Inter‐brain Synchrony

A total of 34 infant–caregiver dyads were included in all inter‐brain synchronisation analyses. The mean and SD of included trials and removed ICA components and interpolated channels for all inter‐brain synchrony analyses are provided in Supplementary Material 4.

Phase‐locking Value (PLV)

We adopted a non‐parametric permutation test and paired samples t‐tests to investigate whether the observed PLV is higher compared to the PLV in permuted dyads and infant, caregiver and video control conditions, respectively. We first tested whether PLV significantly exceeded permuted PLV in inclusion, exclusion and not‐my‐turn trials. The results showed that observed PLV values did not significantly exceed permuted PLV (pinclusion > .05, pexclusion > .05, pnot my turn > .05, FDR‐corrected) in any of the trials (Figures 5 & 6). Moreover, we used paired samples t‐tests to compare the observed PLV with PLV in the control conditions. The result did not support higher observed PLV compared to all the control conditions, except in the video‐control condition for not my turn trial (Supplementary Materials 7). The results of power correlations and PDC are provided in Supplementary Material 8.

FIGURE 5: Results of permutation tests for inter‐brain synchrony (PLV) analysis. Red dashed lines indicate averaged observed PLV in Theta, black dashed lines indicate the threshold for p < 0.05.

FIGURE 6: Results of PLV for real and permuted dyads and uncorrected p‐values. The figure shows the theta band brain‐to‐brain PLVs matrices of all 8 channels for real dyad (right matrixes) and surrogated dyad (middle matrixes) for each condition, and the corresponding uncorrected‐p‐value matrixes (left matrixes) for comparing real PLV to surrogated PLV. The resulting p‐maps were thresholded at p< .05.

Parental Bonding and PLV

The final sample size for the correlation between parental bonding and PLV comprised 34 infant–caregiver dyads.

To test the relation of parental bonding score and PLV between caregiver‐infant dyads, we ran a Spearman correlation analysis as the normal distribution was violated. The result showed a significant correlation in the not my turn trial (rs(32) = ‐0.30, p = 0.038), but there was no significant correlation in inclusion (rs(32) = ‐0.15, p = 0.189) and exclusion trials (rs(32) = ‐0.05, p = 0.374) (Supplementary Materials 9).

Exploratory Analysis

Left‐lateralized Theta Power

As several studies have highlighted the importance of the left/medial frontal regions in theta power responses to ostracism (Crowley et al. 2010; Van Noordt et al. 2015a), we selected a cluster of left/medial electrode (F7, F3, Fz, and FC1). Theta power was extracted across these electrodes, and t‐tests were conducted to compare “exclusion” events with “inclusion” and “not my turn” events in both infants and caregivers. Infants showed higher theta power in “exclusion” compared to “not my turn” events (Mnot my turn = 8.53, SD not my turn = 1.05, Mexclusion = 9.01, SDexclusion = 0.96), (t (33) = ‐1.95, p = 0.029, d = 0.33, BF = 28.55) and not in comparison to “inclusion” events (Minclusion = 8.72, SD inclusion = 0.88, Mexclusion = 9.01, SDexclusion = 0.96), (t (33) = ‐1.40, p = 0.085, d = 0.24, BF = 9.82). However, caregivers did not show any significant differences between “exclusion” and “not my turn” events (Mnot my turn = 6.12, SD not my turn = 1.00, Mexclusion = 5.91, SDexclusion = 0.93), (t (36) = 0.98, p = 0.833, d = ‐1.16, BF = 0.21) and nor “inclusion” events (Minclusion = 6.04, SD inclusion = 0.77, Mexclusion = 5.91, SDexclusion = 0.93), (t (36) = 0.77, p = 0.779, d = ‐0.12, BF = 0.29).

Correlation Between Affective and Behavioural Responses and Neural Markers

In order to explore which affective and behavioural responses contribute most significantly to the observed neural activity, 10 categories of correlation analyses were carried out using Pearson correlations. When the assumptions were violated, Spearman correlations were computed instead. P‐values were corrected using FDR: (1) correlation between caregivers’ affective and behavioural responses and theta power; (2) correlation between infants’ affective and behavioural responses and theta power; (3) cross‐correlation between caregivers’ affective and behavioural responses and infants’ theta power; (4) cross‐correlation between infants’ affective and behavioural responses and caregivers’ theta power; (5) correlations between caregivers’ affective and behavioural responses and alpha power; (6) correlations between infants’ affective and behavioural responses and alpha power; (7) cross‐correlations between caregivers’ affective and behavioural responses and infants’ alpha power; (8) cross‐correlations between infants’ affective and behavioural responses and caregivers’ alpha power; (9) correlation between PLV and caregivers’ affective and behavioural responses; (10) correlation between PLV and infants’ affective and behavioural responses.

The results showed no significant correlations between infants’ and caregivers’ theta and alpha power and their respective affective and behavioural responses (Tables 2, 3, 8 and 9 in Supplementary Materials 12). In terms of cross‐correlation analyses, no significant correlations were also found between infants’ alpha or theta power and caregivers’ emotional and behavioural responses, nor between caregivers’ alpha and theta power and infants’ emotional and behavioural responses. We only found a significant correlation between infants’ attention‐seeking behaviours and PLV during exclusion (rs(35) = 0.25, PFDR = 0.047) (Table 7 in Supplementary Materials 12).