The influence of childhood emotional neglect experience on brain dynamic functional connectivity in young adults

ABSTRACT

Background: Childhood emotional neglect (CEN) confers a great risk for developing multiple psychiatric disorders; however, the neural basis for this association remains unknown. Using a dynamic functional connectivity approach, this study aimed to examine the effects of CEN experience on functional brain networks in young adults.

Method: In total, 21 healthy young adults with CEN experience and 26 without childhood trauma experience were recruited. The childhood trauma experience was assessed using the childhood trauma questionnaire (CTQ), and eligible participants underwent resting-state functional MRI. Sliding windows and k-means clustering were used to identify temporal features of large-scale functional connectivity states (frequency, mean dwell time, and transition numbers).

Result: Dynamic analysis revealed two separate connection states: state 1 was more frequent and characterized by extensive weak connections between the brain regions. State 2 was relatively infrequent and characterized by extensive strong connections between the brain regions. Compared to the control group, the CEN group had a longer mean dwell time in state 1 and significantly decreased transition numbers between states 1 and 2.

Conclusion: The CEN experience affects the temporal properties of young adults’ functional brain connectivity. Young adults with CEN experience tend to be stable in state 1 (extensive weak connections between the brain regions), reducing transitions between states, and reflecting impaired metastability or functional network flexibility.

HIGHLIGHTS

• We focus solely on the childhood emotional neglect experience and its long-term effects on brain function.

• Eligible participants with and without childhood emotional neglect were identified through a large-scale screen among young adults.

• The results found childhood emotional neglect experiences have a long-term impact on brain flexibility.

Antecedentes: La negligencia emocional infantil (CEN por sus siglas en inglés) confiere un alto riesgo de desarrollar múltiples trastornos psiquiátricos; sin embargo, aún se desconoce la base neuronal de esta asociación. Utilizando un enfoque de conectividad funcional dinámica, este estudio tuvo como objetivo examinar los efectos de la experiencia de CEN en las redes cerebrales funcionales en adultos jóvenes.

Método: En total, se reclutaron 21 adultos jóvenes sanos con experiencias de CEN y 26 sin experiencias de trauma infantil. La experiencia del trauma infantil se evaluó mediante el cuestionario de trauma infantil (CTQ), y los participantes elegibles se sometieron a una resonancia magnética funcional en estado de reposo. Se utilizaron ventanas deslizantes y agrupaciones de k-means para identificar características temporales de estados de conectividad funcional a gran escala (frecuencia, tiempo medio de permanencia y números de transición).

Resultados: El análisis dinámico reveló dos estados de conexión separados: el estado 1 era más frecuente y se caracterizaba por conexiones débiles y extensas entre las regiones del cerebro. El estado 2 era relativamente poco frecuente y se caracterizaba por conexiones fuertes y extensas entre regiones del cerebro. En comparación con el grupo de control, el grupo CEN tuvo un tiempo medio de permanencia más largo en el estado 1 y una disminución significativa en el número de transiciones entre los estados 1 y 2.

Conclusión: La experiencia de CEN afecta las propiedades temporales de la conectividad cerebral funcional de los adultos jóvenes. Los adultos jóvenes con experiencia CEN tienden a ser estables en el estado 1 (conexiones débiles extensas entre las regiones del cerebro), reduciendo las transiciones entre estados, y reflejando una metaestabilidad o la flexibilidad funcional de la red deteriorada.

背景: 童年期情感忽视（CEN）对多种精神疾病的发展造成了巨大风险；然而，这种关联的神经基础仍然未知。本研究采用动态功能连接方法，旨在研究CEN经历对年轻成年人大脑功能网络的影响。

方法: 总共招募了21名有CEN经历和26名没有童年期创伤经历的健康年轻人。童年创伤经历使用童年创伤问卷(CTQ)进行评估，符合条件的参与者接受静息态功能磁共振检查。使用滑动窗和K-均值聚类法识别大规模功能连接状态的时间特征（频率、平均停留时间和状态转换次数）。

结果: 动态分析揭示了两个独立的连接状态：状态1比较频繁，以脑区间广泛的弱连接为特征。而状态2相对不频繁，其特点是脑区间广泛的强连接。与对照组相比，CEN组在状态1中的平均停留时间更长，且在状态1和状态2间的转换次数明显减少。

结论: CEN经历影响了年轻人大脑功能连接的时间属性。有CEN经历的年轻人往往稳定地停留在状态1（脑区间广泛的弱连接），减少了状态间的转换，反映出亚稳态性或功能网络的灵活性的降低。

KEYWORDS:

• Childhood emotional neglect
• Dynamic functional connectivity
• Brain flexibility
• Metastability
• Young adults

PALABRAS CLAVE:

• Negligencia emocional infantil
• conectividad funcional dinámica
• flexibilidad cerebral
• metaestabilidad
• adultos jóvenes

关键词:

• 童年期情感忽视
• 动态功能连接
• 大脑灵活性
• 亚稳态性
• 年轻成年人

1. Introduction

Childhood emotional neglect (CEN) is a form of childhood trauma (including physical/sexual/emotional abuse and physical/emotional neglect), referring to when a child's basic emotional needs are not met, distress is not treated sensitively, and social and emotional development is ignored (Teicher & Samson, 2013). Meta-analyses revealed that the global prevalence of CEN is very high, reaching nearly 18% (Stoltenborgh et al., 2013; Stoltenborgh et al., 2015), and CEN often occurs alone, up to 6.2% (Taillieu et al., 2016). Recent studies suggest that CEN has long-term effects on mental disorders and social function dysfunction. Even in adults, CEN experiences are associated with depression, anxiety, stress, substance abuse (Grummitt et al., 2022; Infurna et al., 2016; Salokangas et al., 2020), elevated social anxiety, poor interpersonal interactions, and reduced relationship quality (Derin et al., 2022; Haslam & Taylor, 2022; Müller et al., 2019; Rees, 2008). However, long-term effects of CEN in adults and its neural mechanisms have not been explored in detail. It may be crucial to further understand the pathogenesis of CEN-related psychological and behavioral issues.

Studies have demonstrated that examining of functional connectivity in resting-state fMRI data (RS-FC) is feasible, offering essential insights into the functional connections of specific brain regions and local brain networks and the overall organization of functional communication within brain networks (Smith et al., 2009; van den Heuvel & Hulshoff Pol, 2010). Using RS-FC, a growing corpus of research has investigated the relationship between childhood trauma and brain function, expanding our understanding of the influence of early adverse experiences on neurodevelopment (Barch et al., 2018; Boccadoro et al., 2019; Goltermann et al., 2023; Hakamata et al., 2021; Lu et al., 2017; Luo et al., 2022; Yu et al., 2019; Zhao et al., 2021). For example, childhood trauma severity was associated with decreased RS-FC between the hippocampus and medial prefrontal cortex (PFC). It increased RS-FC between the extrastriate cortex and lateral and anteromedial PFC (Hakamata et al., 2021). Moreover, although additional studies have attempted to distinguish the effects of its different subtypes on brain network development, they have not yet separated emotional neglect from neglect (including physical and emotional neglect) or emotional maltreatment (including emotional neglect and emotional abuse). For instance, in adults, childhood neglect (CN, including physical and emotional neglect) is associated with decreased RS-FC within the salience network. It increased RS-FC between the salience and default mode network (Fadel et al., 2021). CN is related to an enhanced RS-FC within-salience network in community adolescents (Rakesh et al., 2021). CN experiences in patients with major depressive disorder are related to reduced RS-FC in brain regions within the prefrontal-limbic-thalamic-cerebellar circuitry (Wang et al., 2014). Adults reporting childhood emotional maltreatment (including emotional abuse and emotional neglect) showed reduced RS-FC within the right amygdala-bilateral precuneus and dACC-precuneus/frontal regions of the brain (van der Werff et al., 2013). Despite these studies, to our knowledge, none of them has focused solely on CEN.

Previous studies have highlighted that different childhood trauma subtypes may have distinct neurobiological impacts (Edmiston et al., 2011; McLaughlin et al., 2014). Moreover, among subtypes of childhood trauma, only emotional neglect significantly predicts depression, anxiety, stress, resilience, alexithymia, and substance misuse disorders (Aust et al., 2013; Grummitt et al., 2022; Lee et al., 2018; Salokangas et al., 2020). These findings suggest that combining CEN and another type of childhood trauma into one exposure may obscure the effect. Consequently, it is necessary to further uncover the CEN experience effect on the brain network.

Dynamic functional connectivity (DFC), quantifying changes in functional connectivity metrics over time, is an essential development of traditional static functional connectivity and can provide a greater understanding of the fundamental properties of brain networks (Hutchison et al., 2013). Recent studies have revealed abnormal DFC patterns of childhood trauma with a mixture of all subtypes (Huang et al., 2021) and some mental disorders correlated with CEN, such as major depression (Demirtaş et al., 2016) and generalized anxiety (Chen et al., 2020). Therefore, DFC is a potential analytical method to probe large-scale brain network characteristics of young adults with CEN and is expected to enrich neuroimaging results.

The primary aim of this study was to investigate the differences in temporal properties of large-scale functional connectivity in the whole brain between healthy young adults with and without CEN. We hypothesized that alterations in DFC temporal properties (frequency, mean dwell time, and transition numbers) would be found in young adults with CEN compared to controls.

2. Materials and methods

2.1. Participants

A total of 21 young adults with CEN (CEN group, 20.2 ± 0.7 years) and 26 young adults without CEN experience (HC group, 19.9 ± 1.1 years) were recruited in this study. Participants were screened and recruited in the following steps: First, we preliminarily assessed childhood trauma online using the Childhood Trauma Questionnaire-Short Form (CTQ-SF) for 5010 undergraduates from four universities in Tianjin, China. Recruiting was conducted via campus flyers and internet advertisements, which were linked to online screening questions. The CTQ-SF is a widely used self-report questionnaire for assessing childhood traumatic experiences (Bernstein et al., 2003). Its Chinese version has good validity and reliability (Zhao et al., 2005). Next, potential participants qualifying for the initial screening were re-recruited for online interviews for CTQ-SF retesting and mental disorder diagnosis at least one month later. The CTQ retest was to minimize the influence of participants’ misreading questionnaires and their recall bias on the results. According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) Axis I Disorder criteria stereotyped clinical examination (SCID-I/P) (First et al., 2005), mental disorder diagnosis was assessed by two clinical experts independently. Only those who were diagnosed as asymptomatic by both experts were recruited for our formal experiment. Eligible participants were finally invited to participate in follow-up fMRI experiments. All participants participating in the experiment provided informed consent. Ethical approval was obtained from the local ethics committee (Ethics Committee of Tianjin Normal University).

The significance of each CTQ-SF factor was determined using the cut-off scores of moderate-to-severe childhood trauma as recommended by relevant studies; as long as the score is below cut-off score, it is correspondingly considered as a non-trauma experience (Kim et al., 2018; Lu et al., 2017; Peters et al., 2019; Wu et al., 2020). Therefore, participants who met the following CTQ-SF assessments criteria were thought to be primarily influenced by childhood emotional neglect and included in the CEN group: The scores of emotional neglect were ≥ 15, physical abuse < 9, emotional abuse < 12, sexual abuse < 7, and physical neglect < 9. Participants with the lowest score of 5 on CTQ-SF assessments for physical abuse, emotional abuse, sexual abuse, emotional neglect, and physical neglect scores were considered without any childhood trauma and included in the HC group. Additionally, participants with psychological disorders were excluded from the CEN and HC groups based on assessment results of SCID-I/P.

2.2. Data acquisition

Images were acquired on a 3.0-Tesla SIEMENS scanner using a 64-channel head coil in Tianjin Normal University. We acquired resting-state fMRI images with an echo-planar imaging sequence using the following settings: parallel imaging (acceleration factor PE = 2, reference lines PE = 26, acceleration factor slice = 3), slice number = 75, repetition time (TR) = 2,000 ms, echo time (TE) = 30 ms, flip angle (FA) = 90°, field of view (Fov) = 224 mm × 224 mm, matrix = 112 × 112, and slice thickness = 2 mm. We collected 240 resting-state fMRI brain volumes for each participant. In the scanner, the participants were instructed to remain still with their eyes closed, but not to fall asleep. In order to ensure that participants were indeed closed eyes but not asleep during the scanning, the experimenter monitored whether a participant was asleep through the eye images transmitted by a camera equipped on the coil during the scanning. And further, after the scanning, each participant was asked if they had fallen asleep during the scan. The following parameters were used to acquire T1-weighted structural MRI images: inversion time = 1100 ms, slice number = 192, repetition time (TR) = 2,530 ms, echo time (TE) = 2.98 ms, flip angle (FA) = 7°, field of view (Fov) = 256 mm × 256 mm, matrix = 256 × 256, and slice thickness = 1 mm.

2.3. Data preprocessing

Data preprocessing was performed using RESTplus V1.2 (http://www.restfmri.net). We discarded the first ten functional images due to magnetic field instability, resulting in 230 remaining images. Slice timing correction and spatial realignment were applied to adjust for interleaved slice acquisition and eliminate head motion. One CEN group participant was excluded from subsequent analyses because of excessive head movement (> 2.5 mm translation or 2.5°rotation). Functional images were normalized to the Montreal Neurological Institute (MNI) space by T1 image unified segmentation. Furthermore, spatial smoothing was performed using a 6 mm FWHM Gaussian kernel. After a linear detrend of the signal, nuisance signals (white matter signals, cerebrospinal fluid signals, and Friston-24 head motion parameters) were extracted and regressed from the data. Subsequently, Data were bandpass filtered between 0.01 and 0.1 Hz. Finally, in order to reduce the impact of head motion on fMRI data, we performed data scrubbing. Specifically, interpolation was performed with data from two previous and one subsequent time point for frame-wise displacement (FD) greater than 0.5 (Chen et al., 2020; Guo et al., 2020). Based on our examination, only two participants’ resting-state data were scrubbed, with an average scrubbing rate of less than 1%.

2.4. Dynamic functional connectivity analysis

A sliding window method was employed to examine dynamic functional connectivity characteristics in the Dynamic Brain Connectome (DynamicBC) toolbox (V2.2 http://restfmri.net/forum/DynamicBC). Regions of interest (ROIs) were chosen from the Automated Anatomical Labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002), which segmented the brain into 116 cortical, subcortical, and cerebellar regions and is widely applied in neuroscience. We then calculated the correlations between the time courses of each pair of ROIs within RS-fMRI scan windows to determine whole-brain functional connectivity. In line with previous research (J. Li et al., 2019; Zhang et al., 2022), the resting state data of 230 TRs were divided into 37 windows at the setting of 50 TRs (100 s) as the window length and 5 TRs (10 s) as the step size. Therefore, the time series generated 37 functional connectivity matrices of 114 × 114 for each participant. This segment length has been demonstrated to optimize the balance between capturing a rapidly shifting dynamic relationship and the quality of correlation matrix estimation (Chen et al., 2020; C. Li et al., 2019; J. Li et al., 2019; Pang et al., 2018). Temporal variability was then estimated by computing the variance of each pair of connections in the 37 functional connectivity matrix windows generated for each participant, representing the overall variability of whole-brain functional connectivity during the scan (Chen et al., 2020).

2.5. Clustering analysis

We used a k-means clustering algorithm on the windowed functional connectivity matrices to estimate recurring functional connectivity patterns (states). The similarity between windowed functional connectivity matrices was calculated using the default distance measures (sqEuclidean distance). It has been proven to be an effective measure for high-dimensional data (Lin et al., 2021; Xu et al., 2022). Furthermore, we performed a cluster validity analysis to evaluate the optimal number of states based on three criteria (Silhouette, Davies-Bouldin values, and Calinski – Harabasz), and the maximum number of states allowed to estimate was set at 10. The temporal properties of each participant's functional connectivity state were explored by calculating each state's frequency, average dwell time, and transition numbers between states. Specifically, the ‘frequency’ refers to the ratio of the number of connection matrices belonging to a specific state to the total number of connection matrices of the cluster, which reflects the temporal fraction of a specific state in the scan sequence. The ‘mean dwell time’ is the average number of consecutive matrix windows belonging to a state. The ‘transition numbers’ is the total number of transitions between states, reflecting each state's reliability and stability (Fiorenzato et al., 2019). Consistent with previous studies, the fact that some participants did not enter a given state during the scan would lead to inaccurate estimates of mean dwell time. In contrast, the frequency and transitions number are unaffected because they are still informative to the state's total occurrence (Snyder et al., 2021). Therefore, thirteen participants (5 CEN group, 8 HC group) who did not enter given states were excluded from the analysis of mean dwell time.

Referring to previous studies (Fiorenzato et al., 2019; Zhang et al., 2022), we next performed a cluster analysis for the CEN and HC groups separately to explore whether the two groups have a different optimal number of states.

2.6. Statistical analysis

Chi-square tests were performed for other demographic variables (p < .05), including gender, education levels of parents, work status of parents, student origin distribution, and whether only child. Independent samples t-tests were performed for age, CTQ subscale scores, and total scores of both groups (p < .05).

The differences in temporal variability between the CEN and HC groups were tested using the independent samples t-test, with gender, age, and mean FD as covariates. Multiple comparison correction was then performed using false discovery rate (FDR) corrections (Benjamini & Hochberg, 1995).

The group differences in temporal properties of DFC states between the CEN and HC groups were tested using the Bootstrap 2-sample t-test (with 1000 bootstrap replicates) between the CEN and HC groups. In cases when there is no sampling distribution formula or if the assumptions of existing formulas are inappropriate (e.g. small sample size, non-normal distribution), bootstrapping provides an effective way to accurately estimate standard errors (Efron & Tibshirani, 1994; Schambra et al., 2015; Toy et al., 2021). This study conducted a relatively high number of bootstrap analyses with 1,000 replicates to generate accurate estimates of standard errors.

3. Results

3.1. Group differences in demographics and CTQ questionnaire

The statistical results of demographics showed no significant difference between the two groups regarding gender, age, education levels of parents, work status of parents, student origin distribution, and whether only child (p > .05).

The statistical results of the CTQ questionnaire showed that the CEN group exhibited significantly higher scores in all CTQ subscales and the overall score compared to the HC group (p < .001). However, as mentioned previously, the other trauma subtypes in the CEN group were well below the diagnostic criteria (cut-off scores), except for emotional neglect, which was above the diagnostic criteria. Referring to a previous study (Wang et al., 2014), if a childhood trauma score does not reach the diagnostic threshold, it could be considered to be absent; that is, the experience has not accumulated to a state that may cause qualitative change. We, therefore, inferred that the CEN group in the current study was primarily influenced by emotional neglect (see Appendix 1 for details).

3.2. Group differences in temporal variability

After FDR correction, the analysis of temporal variability revealed no statistically significant group differences in the variance of each pair of connections across all sliding windows.

3.3. Group differences in temporal properties of DFC states

According to the estimation of the DynamicBC toolbox, all three criteria consistently indicate that the optimal number of clusters is 2. Consequently, we identified two states of functional connectivity patterns recurring throughout scans across participants. According to the percentage of state occurrence, state 1 (67.27%) is far more frequent than state 2 (32.73%). Moreover, state 1 is characterized by extensive weak functional connections between brain regions, as evidenced by numerous blue areas and little orange regions in the heat map of Figure 1 (left). The limited strong connections within state 1 usually occur proximate to the diagonal of the heat map, corresponding to the left and right parts of the same brain structure according to the segmentation of the AAL template. Such as, the brain regions 43–52 (namely Calcarine_L, Calcarine_R, Cuneus_L, Cuneus_R, Lingual_L, Lingual_R, Occipital_Sup_L, Occipital_Sup_R, Occipital_Mid_L, and Occipital_Mid_R), are the most conspicuous in state 1. In contrast, state 2 is characterized by widespread and strong functional connections among brain regions, as demonstrated by the numerous orange regions and little blue regions illustrated in the heat map of Figure 1 (right). Within State 2, almost all brain regions have more or less strong connections with other brain regions, except for the brain region 96 (Cerebelum_3_R), 110 (Vermis_3), and 113 (Vermis_7).

Figure 1. Centroids of clusters. Each state is represented by the median of each cluster and shows each state's total number and frequency of occurrence. Both horizontal and vertical coordinates indicate the 116 brain regions segmented by the AAL atlas. The heat map represents the average functional connectivity values of the corresponding brain regions’ time series, which were calculated using Pearson's r.

Furthermore, we calculated the state transitions for each participant in consecutive time windows (Figure 2). Then, the states’ temporal properties (frequency, dwell time, and transitions number) were compared between groups. Overall, for the frequency of state 1, there was no significant difference between the CEN (67.16 ± 7.59%) and HC groups (67.36 ± 6.52%), and the same for state 2 (P = .99); see Figure 3A. For state 1, we found a significant difference in the mean dwell time between the CEN and HC groups. Specifically, the CEN group showed a longer mean dwell time of state 1 than the HC group (CEN group: 19.13 ± 11.57, HC group: 10.73 ± 9.37; P = .039; Figure 3B). However, there was no significant group difference in mean dwell time for state 2 (CEN group: 7.9 ± 6.47, HC group: 8.13 ± 9.37; P = .95; Figure 3B). Moreover, there were significant differences in transition numbers between the CEN and HC groups. As displayed in Figure 3C, the transition numbers between states 1 and 2 were significantly lower in the CEN group than in the HC group (CEN group: 1.45 ± 0.99, HC group: 2.65 ± 2.56; P = .043; Figure 3C).

Figure 2. State transition pattern for each participant in 37 matrix windows

Figure 3. Temporal properties of DFC states for CEN and HC groups. (A) The percentage of each state, (B) the mean dwell time of each state, and (C) the transition numbers between states during the scan for both groups. The solid and dashed lines of the violin diagram represent the mean and quartiles, respectively. CEN = childhood emotional neglect group, HC = control group.

3.4. Cluster analysis across groups

As demonstrated in Figure 4, the optimal number of clusters for both groups is 2, validating the rationality of clustering into two classes when spanning all participants. Moreover, we observed that state transitions were more sporadic in the HC group than in the CEN group, which means that the CEN group preferred to stay in a certain state, while the HC group switched states more frequently. This is consistent with our overall clustering analysis that the HC group has a higher number of state transitions.

Figure 4. Cluster analysis (state types and transition patterns) for CEN and HC groups separately

4. Discussion

The present study was designed to compare the temporal properties of functional connectivity between young adults with and without childhood emotional neglect experience. We observed two distinct connectivity patterns across all participants – a frequent, extensive weak connectivity dominant (state 1) and an infrequent, extensive strong connectivity dominant (state 2). The results indicate that the CEN group dwelled longer in state 1 than the HC group. However, the CEN group showed fewer state transitions during the scan than the HC group.

A theoretical framework, namely metastability, can describe these dynamic results observed in this study. The metastability mechanism argues that the tendencies for brain regions to couple and coordinate globally for multiple functions (integration) coexist with tendencies to express their autonomy and specialized functions (segregation) (Shanahan, 2010; Tognoli & Kelso, 2014). Furthermore, metastability theory suggests that brain coordination regions occur flexibly, with ensembles of various sizes constantly assembling and dissolving in repetitive cycles. This process realizes a dynamic balance between the global integration of information and the local separation of functions, thus optimizing information transfer and processing capabilities (Braun et al., 2015; Li et al., 2017; Tognoli & Kelso, 2014). According to this theory, integrated and separated brain interaction patterns correspond to the two states observed in our study. The CEN group might have a poor metastability capacity, reducing functional network flexibility. Therefore, the CEN group tended to consistently have an integrated functional connectivity pattern (state 1), reducing the probability of transition between states. From the following perspectives, the decreased flexibility of the functional brain network in the CEN group might be understood.

First, this decreased metastability, characterized by reduced state transitions and altered state dwell time, may be a potential neural mechanism for reduced cognitive flexibility at the behavioral level in adults experiencing CEN. Previous studies indirectly indicate that a decreased metastability level was associated with reduced cognitive flexibility and information processing speed (Braun et al., 2015; Hellyer et al., 2015). DFC studies have revealed that individuals with cognitive impairment, such as Parkinson's disease (Fiorenzato et al., 2019) and subjective cognitive decline (Chen et al., 2021), are associated with a decreased transition number between states. Furthermore, behavioral research has shown that teenagers who experienced a single type of childhood maltreatment performed worse than teenagers without maltreatment or with multitype maltreatment on cognitive flexibility tasks and visual processing speed (Mothes et al., 2015). Adolescents with early life stress experiences revealed significantly impaired cognitive flexibility. Although they could learn to match stimuli to positive and negative outcomes, when these connections changed abruptly [when the reward corresponding to the stimulus suddenly changed to punishment], they were less able to renew these pairings than the HC group (Harms et al., 2018). The decreased metastability observed in adults experiencing CEN may be linked to reduced cognitive flexibility. Nevertheless, future research should examine potential links between dynamic functional connectivity properties and cognitive flexibility among individuals with CEN.

Second, we speculate that metastable neural dynamics may be disrupted by negative affectivity in young adults with CEN. Preliminary evidence suggests that emotional states positively correlate with brain flexibility (Betzel et al., 2016, 2017). Studies such as Betzel et al. (2017) divided the RS-fMRI time series into several non-overlapping windows and then calculated the functional connectivity of each window separately. Then, each brain region is assigned to a community in each window using a community detection algorithm. An inflexible brain network is characterized by brain regions consistently assigned to the same community, indicating segregated sub-networks. Their correlation analysis showed that positive mood was positively correlated with flexibility. Moreover, individuals with CEN, even adults, still have more negative affectivity (Jin et al., 2018; Schimmenti et al., 2015). For example, Schimmenti et al. (2015) examined the relationships between CEN, current psychiatric symptoms, and negative affectivity in community adults. Their findings showed a positive correlation between CEN experiences and the severity of current mental symptoms, which is mediated by levels of negative affectivity. Thus, the reduced brain flexibility observed in the current study (as reflected by reduced state transitions and altered state dwell time) may be linked to negative affect resulting from the CEN experience. Further research is necessary to account for these variables. More specifically, future research could incorporate behavioral measures of state and trait affect and correlate these results with dynamic connectivity properties.

Notably, there are no group differences in the variance of functional connectivity across the scan for specific brain regions. This result suggests that dynamic functional connectivity abnormalities in the CEN group appear to be primarily related to temporal properties rather than global variation in the strength of functional connectivity. Additionally, when we exploratively performed cluster analysis in each group, the optimal number of clusters was 2. This is consistent with cluster analysis results across two groups, suggesting that both states were consistently observed.

Nevertheless, some limitations are inevitable in this study and can be considered in future studies. First, the undergraduate sample selected for the current study had a long-term school life, which may make their emotional experience somewhat different from that of young adults with short school life, limiting the generalizability of the current results. Second, given that the sample is all undergraduate students, the ability to generalize to other age groups is limited, especially considering that brain developmental processes in adolescence may alter results. Finally, we chose the sliding window method with a fixed window length to construct a dynamic functional connectivity matrix, which may affect the flexibility of detecting abrupt changes in functional connectivity (Lindquist et al., 2014).

5. Conclusion

Early emotional neglect experience alters the temporal features of the brain's functional connections, according to our research. Specifically, compared to young people without CEN experience, those with CEN experience exhibited longer dwell time in the separated connection pattern and fewer transitions between the separated and integrated brain connectivity states. These findings indicate a decreased metastability or flexibility of the functional network in the CEN group, which may be related to the decreased cognitive flexibility and increased negative emotions of CEN individuals. This study helps us better understand the long-term outcomes of childhood emotional neglect on the brain's functional networks. It also provides a reference for preventing and intervening in psychological and social problems secondary to CEN.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Appendix 1.

Groups differences in demographic and CTQ-questionnaire.

Table

Variables		CEN Group (n = 21)	HC Group (n = 26)	t / χ^2	p
Age		19.19 ± 0.68	19.04 ± 0.87	t = 0.66	p = .52
Gender				χ^2= 0.026	p = .872
	Male	10	13
	Female	11	13
Education Levels of Parents				χ^2= 3.594/2.280	p = .309/.516
(father/mother)	None	0/0	0/0
	Primary School	1/6	1/3
	Junior Middle School	12/6	8/10
	High School/Technical Secondary School	6/6	12/8
	University / Junior College	2/3	5/5
	Master Degree or Above	0/0	0/0
Work Status of Parents				χ^2= 3.460/1.918	p = .749/0.927
(father/mother)	None	2/5	1/5
	Education System	0/1	1/1
	Health System	0/1	1/1
	State Owned Enterprises	2/1	5/1
	Government Departments	1/3	2/2
	Private Enterprises	3/1	3/4
	Others	13/9	13/12
Student Origin Distribution				χ^2= 2.316	p = .510
	Northeast	0	1
	East	9	15
	Middle	5	3
	West	5	6
Whether Only Child				χ^2= 0.016	p = .900
	Yes	8	11
	No	11	14
CTQ_ questionnaire
	emotional neglect	16.00 ± 1.22	5.00 ± 0.00	t = 45.918	p < .001
	emotional abuse	8.10 ± 2.43	5.00 ± 0.00	t = 6.520	p < .001
	physical abuse	6.05 ± 1.40	5.00 ± 0.00	t = 3.838	p < .001
	sexual abuse	5.52 ± 0.81	5.00 ± 0.00	t = 3.292	P = .002
	physical neglect	7.76 ± 1.34	5.00 ± 0.00	t = 10.553	p < .001
	total score	43.43 ± 4.03	25.00 ± 0.00	t = 20.460	p < .001

Note: After excluding the three participants with excessive head movement, the significance of the differences between the two groups did not change.