Assessing Temporal Emotion Dynamics Using Networks

Network analysis of pandemic fatigue symptoms in samples from five South American countries

BACKGROUND

Pandemic fatigue generates low motivation or the ability to comply with protective behaviors to mitigate the spread of COVID-19.

AIMS

This study aimed to analyze the symptoms of pandemic fatigue through network analysis in individuals from five South American countries.

METHOD

A total of 1,444 individuals from Argentina, Bolivia, Paraguay, Peru, and Uruguay participated and were evaluated using the Pandemic Fatigue Scale. The networks were estimated using the ggmModSelect estimation method and a polychoric correlation matrix was used. Stability assessment of the five networks was performed using the nonparametric resampling method based on the case bootstrap type. For the estimation of network centrality, a metric based on node strength was used, whereas network comparison was performed using a permutation-based approach.

RESULTS

The results showed that the relationships between pandemic fatigue symptoms were strongest in the demotivation dimension. Variability in the centrality of pandemic fatigue symptoms was observed among participating countries. Finally, symptom networks were invariant and almost identical across participating countries.

CONCLUSIONS

This study is the first to provide information on how pandemic fatigue symptoms were related during the COVID-19 pandemic.

Using cross-validation methods to select time series models: Promises and pitfalls

Vector autoregressive (VAR) modelling is widely employed in psychology for time series analyses of dynamic processes. However, the typically short time series in psychological studies can lead to overfitting of VAR models, impairing their predictive ability on unseen samples. Cross-validation (CV) methods are commonly recommended for assessing the predictive ability of statistical models. However, it is unclear how the performance of CV is affected by characteristics of time series data and the fitted models. In this simulation study, we examine the ability of two CV methods, namely,10-fold CV and blocked CV, in estimating the prediction errors of three time series models with increasing complexity (person-mean, AR, and VAR), and evaluate how their performance is affected by data characteristics. We then compare these CV methods to the traditional methods using the Akaike (AIC) and Bayesian (BIC) information criteria in their accuracy of selecting the most predictive models. We find that CV methods tend to underestimate prediction errors of simpler models, but overestimate prediction errors of VAR models, particularly when the number of observations is small. Nonetheless, CV methods, especially blocked CV, generally outperform the AIC and BIC. We conclude our study with a discussion on the implications of the findings and provide helpful guidelines for practice.

Staging models in eating disorders: A systematic scoping review of the literature

Eating Disorders (ED) are characterized by low remission rates, treatment drop-out, and residual symptoms. To improve assessment and treatment of ED, the staging approach has been proposed. This systematic scoping review is aimed at mapping the existing staging models that explicitly propose stages of the progression of ED. A systematic search of PubMed, PsycINFO, Scopus was conducted with the terms staging, anorexia nervosa, bulimia nervosa, binge-eating disorders, eating disorders. Eleven studies met inclusion criteria presenting nine ED staging models, mostly for anorexia nervosa. Three were empirically tested, one of which was through an objective measure specifically developed to differentiate between stages. Most staging models featured early stages in which the exacerbation of EDs unfolds and acute phases are followed by chronic stages. Intermediate stages were not limited to acute stages, but also residual phases, remission, relapse, and recovery. The criteria for stage differentiation encompassed behavioral, psychological, cognitive, and physical features including body mass index and illness duration. One study recommended stage-oriented interventions. The current review underscores the need to empirically test the available staging models and to develop and test new proposals of staging models for other ED populations. The inclusion of criteria based on medical features and biomarkers is recommended. Staging models can potentially guide assessment and interventions in daily clinical settings.

A Network Study of Family Affect Systems in Daily Life

Adolescence is a time period characterized by extremes in affect and increasing prevalence of mental health problems. Prior studies have illustrated how affect states of adolescents are related to interactions with parents. However, it remains unclear how affect states among family triads, that is adolescents and their parents, are related in daily life. This study investigated affect state dynamics (happy, sad, relaxed, and irritated) of 60 family triads, including 60 adolescents (Mage = 15.92, 63.3% females), fathers and mothers (Mage = 49.16). The families participated in the RE-PAIR study, where they reported their affect states in four ecological momentary assessments per day for 14 days. First, we used multilevel vector-autoregressive network models to estimate affect dynamics across all families, and for each family individually. Resulting models elucidated how family affect states were related at the same moment, and over time. We identified relations from parents to adolescents and vice versa, while considering family variation in these relations. Second, we evaluated the statistical performance of the network model via a simulation study, varying the percentage missing data, the number of families, and the number of time points. We conclude with substantive and statistical recommendations for future research on family affect dynamics.

Skewness and Staging: Does the Floor Effect Induce Bias in Multilevel AR(1) Models?

Multilevel autoregressive models are popular choices for the analysis of intensive longitudinal data in psychology. Empirical studies have found a positive correlation between autoregressive parameters of affective time series and the between-person measures of psychopathology, a phenomenon known as the staging effect. However, it has been argued that such findings may represent a statistical artifact: Although common models assume normal error distributions, empirical data (for instance, measurements of negative affect among healthy individuals) often exhibit the floor effect, that is response distributions with high skewness, low mean, and low variability. In this paper, we investigated whether-and to what extent-the floor effect leads to erroneous conclusions by means of a simulation study. We describe three dynamic models which have meaningful substantive interpretations and can produce floor-effect data. We simulate multilevel data from these models, varying skewness independent of individuals' autoregressive parameters, while also varying the number of time points and cases. Analyzing these data with the standard multilevel AR(1) model we found that positive bias only occurs when modeling with random residual variance, whereas modeling with fixed residual variance leads to negative bias. We discuss the implications of our study for data collection and modeling choices.

Assessment of fit of the time-varying dynamic partial credit model using the posterior predictive model checking method

Several new models based on item response theory have recently been suggested to analyse intensive longitudinal data. One of these new models is the time-varying dynamic partial credit model (TV-DPCM; Castro-Alvarez et al., Multivariate Behavioral Research, 2023, 1), which is a combination of the partial credit model and the time-varying autoregressive model. The model allows the study of the psychometric properties of the items and the modelling of nonlinear trends at the latent state level. However, there is a severe lack of tools to assess the fit of the TV-DPCM. In this paper, we propose and develop several test statistics and discrepancy measures based on the posterior predictive model checking (PPMC) method (PPMC; Rubin, The Annals of Statistics, 1984, 12, 1151) to assess the fit of the TV-DPCM. Simulated and empirical data are used to study the performance of and illustrate the effectiveness of the PPMC method.

Emotions under control? Better cognitive control is associated with reduced negative emotionality but increased negative emotional reactivity within individuals

Associations between impaired cognitive control and maladaptive emotion regulation have been extensively studied between individuals. However, it remains unclear if this relationship holds within individuals. In this study, we tested the assumption that momentary within-person fluctuation in cognitive control (working memory updating and response inhibition) is associated with emotional reactivity in everyday life. We conducted an experience sampling study (eight two-hourly prompts daily) where participants repeatedly performed short 2-back and Go/no-go tasks in daily life. We assessed negative and positive affective states, and unpleasantness of a recent event to capture emotional reactivity. We analyzed two overlapping samples: a Go/no-go and a 2-back dataset (N = 161/158). Our results showed that better momentary working memory updating was associated with decreased negative affect if the recent event was on average unpleasant for the given individual. However, better-than-average working memory updating in interaction with higher event-unpleasantness predicted higher negative affect levels (i.e., higher negative emotional reactivity). These findings may challenge the account of better cognitive control being universally related to adaptive emotion regulation. Although it is unlikely that emotional reactivity boosts working memory, future studies should establish the direction of causality.

Avoidance and fear day by day in social anxiety disorder

OBJECTIVE

Theories assert that avoidance maintains maladaptive anxiety over time, yet a clear prospective test of this effect in the day-by-day lives of people with social anxiety disorder (SAD) is lacking.

METHOD

We used intensive longitudinal data to test prospective relationships between social fear and social avoidance in 32 participants with SAD who reported on a total of 4256 time points.

RESULTS

Results suggested that avoidance strongly predicted future anxiety, but only in a minority of people with SAD. Relationships between anxiety and avoidance varied considerably across individuals. Pre-registered tests found that the strength of autocorrelation for social fear is a good target for future testing of prediction of exposure response. Participants with lower autocorrelations were less likely to show between-session habituation.

CONCLUSIONS

Overall, results suggest avoidance maintains fear in SAD for at least some individuals, but also indicates considerable variability. Further intensive longitudinal data is needed to examine individuals with SAD across varying time courses.

Temporal associations between salivary cortisol and emotions in clinically depressed individuals and matched controls: A dynamic time warp analysis

Depression can be understood as a complex dynamic system where depressive symptoms interact with one another. Cortisol is suggested to play a major role in the pathophysiology of depression, but knowledge on the temporal interplay between cortisol and depressive symptoms is scarce. We aimed to analyze the temporal connectivity between salivary cortisol and momentary affective states in depressed individuals and controls. Thirty pair-matched depressed and non-depressed participants completed questionnaires on momentary positive (PA) and negative (NA) affect and collected saliva three times a day for 30 days. The association between cortisol and affect was analyzed by dynamic time warp (DTW) analyses. These analyses involved lag-1 backward to lag-1 forward undirected analyses and lag-0 and lag-1 forward directed analyses. Large inter- and intra-individual variability in the networks were found. At the group level, with undirected analysis PA and NA were connected in the networks in depressed individuals and in controls. Directed analyses indicated that increases in cortisol preceded specific NA items in controls, but tended to follow upon specific affect items increase in depressed individuals. To conclude, at group level, changes in cortisol levels in individuals diagnosed with a depression may be a result of changes in affect, rather than a cause.

Balancing media selections over time: Emotional valence, informational content, and time intervals of use

The sequencing of information in media can influence processing of content via mechanisms like framing, mood management, and emotion regulation. This study examined three kinds of media sequences on smartphones: (1) balancing positive and negative emotional content; (2) balancing emotional content with informational content; and (3) balancing time spent on and off the media device. Actual media use was measured in natural settings using the Screenomics framework which gathers screenshots from smartphones every 5 s when devices are on. Time-series analyses of 223,531 smartphone sessions recorded from 94 participants showed that emotionally positive content was more likely to follow negative content, and that emotionally negative content was more likely to follow positive content; emotional content was more likely to follow informational content, and informational content was more likely to follow emotional content; and longer smartphone sessions were more likely to follow longer periods of non-use.

The computational psychopathology of emotion

Mood and anxiety disorders involve recurring, maladaptive patterns of distinct emotions and moods. Here, we argue that understanding these maladaptive patterns first requires understanding how emotions and moods guide adaptive behavior. We thus review recent progress in computational accounts of emotion that aims to explain the adaptive role of distinct emotions and mood. We then highlight how this emerging approach could be used to explain maladaptive emotions in various psychopathologies. In particular, we identify three computational factors that may be responsible for excessive emotions and moods of different types: self-intensifying affective biases, misestimations of predictability, and misestimations of controllability. Finally, we outline how the psychopathological roles of these factors can be tested, and how they may be used to improve psychotherapeutic and psychopharmacological interventions.

Experimental evidence that browsing for activewear lowers explicit body image attitudes and implicit self-esteem in women

Online apparel shopping is popular amongst women and offers salient visual information for making body image and self-worth judgements. Apparel segments which emphasize the value of women's bodies are particularly effective for eliciting low body image and self-worth. Across two studies, we investigated the association between self-reported and experimental online activewear exposure on women's self-worth, body image, appearance attitudes, mood and gaze behavior. In Study 1, participants (N = 399) completed a survey collecting their online apparel shopping habits, body appreciation, self-esteem, appearance comparison tendencies and self-objectification attitudes. Activewear was the second-most popular apparel segment amongst women (after casualwear) and weekly activewear browse time was positively correlated with appearance comparison tendencies, desires to be muscular/athletic and body shame. In Study 2, participants (N = 126) were randomly allocated to browse an activewear, casualwear or homewares website and completed pre and post measures of mood, body image, implicit self-esteem and body gaze behavior. In the activewear condition, there was a significant reduction in positive body image and implicit self-esteem scores. There were no experimental effects for body gaze behavior. These findings illustrate that apparel choices have value for understanding the aetiology of maladaptive body image attitudes and low self-esteem in women.

Network analysis of ecological momentary assessment identifies frustration as a central node in irritability

BACKGROUND

Irritability presents transdiagnostically, commonly occurring with anxiety and other mood symptoms. However, little is known about the temporal and dynamic interplay among irritability-related clinical phenomena. Using a novel network analytic approach with smartphone-based ecological momentary assessment (EMA), we examined how irritability and other anxiety and mood symptoms were connected.

METHODS

Sample included 152 youth ages 8-18 years (M ± SD = 12.28 ± 2.53; 69.74% male; 65.79% White) across several diagnostic groups enriched for irritability including disruptive mood dysregulation disorder (n = 34), oppositional defiant disorder (n = 9), attention-deficit/hyperactivity disorder (n = 47), anxiety disorder (n = 29), and healthy comparisons (n = 33). Participants completed EMA on irritability-related constructs and other mood and anxiety symptoms three times a day for 7 days. EMA probed symptoms on two timescales: "since the last prompt" (between-prompt) versus "at the time of the prompt" (momentary). Irritability was also assessed using parent-, child- and clinician-reports (Affective Reactivity Index; ARI), following EMA. Multilevel vector autoregressive (mlVAR) models estimated a temporal, a contemporaneous within-subject and a between-subject network of symptoms, separately for between-prompt and momentary symptoms.

RESULTS

For between-prompt symptoms, frustration emerged as the most central node in both within- and between-subject networks and predicted more mood changes at the next timepoint in the temporal network. For momentary symptoms, sadness and anger emerged as the most central node in the within- and between-subject network, respectively. While anger was positively related to sadness within individuals and measurement occasions, anger was more broadly positively related to sadness, mood lability, and worry between/across individuals. Finally, mean levels, not variability, of EMA-indexed irritability were strongly related to ARI scores.

CONCLUSIONS

This study advances current understanding of symptom-level and temporal dynamics of irritability. Results suggest frustration as a potential clinically relevant treatment target. Future experimental work and clinical trials that systematically manipulate irritability-related features (e.g. frustration, unfairness) will elucidate the causal relations among clinical variables.

Markov chain analysis indicates that positive and negative emotions have abnormal temporal interactions during daily life in schizophrenia

Abnormalities in positive and negative emotional experience have been identified in laboratory-based studies in schizophrenia (SZ) and associated with poorer clinical outcomes. However, emotions are not static in daily life-they are dynamic processes that unfold across time and are characterized by temporal interactions. Whether these temporal interactions are abnormal in SZ and associated with clinical outcomes is unclear (i.e., whether the experience of positive/negative emotions at time t increases or decreases the intensity of positive/negative emotions at time t+1). In the current study, participants with SZ (n = 48) and healthy controls (CN; n = 52) completed 6 days of ecological momentary assessment (EMA) surveys that sampled state emotional experience and symptoms. The EMA emotional experience data was submitted to Markov chain analysis to evaluate transitions among combined positive and negative affective states from time t to t+1. Results indicated that: (1) In SZ, the emotion system is more likely to stay in moderate or high negative affect states, regardless of positive affect level; (2) SZ transition to co-activated emotional states more than CN, and once emotional co-activation occurs, the range of emotional states SZ transition to is more variable than CN; (3) Maladaptive transitions among emotional states were significantly correlated with greater positive symptoms and poorer functional outcome in SZ. Collectively, these findings clarify how emotional co-activation occurs in SZ and its effects on the emotion system across time, as well as how negative emotions dampen the ability to sustain positive emotions across time. Treatment implications are discussed.

Blind Subgrouping of Task-based fMRI

Significant heterogeneity in network structures reflecting individuals' dynamic processes can exist within subgroups of people (e.g., diagnostic category, gender). This makes it difficult to make inferences regarding these predefined subgroups. For this reason, researchers sometimes wish to identify subsets of individuals who have similarities in their dynamic processes regardless of any predefined category. This requires unsupervised classification of individuals based on similarities in their dynamic processes, or equivalently, in this case, similarities in their network structures of edges. The present paper tests a recently developed algorithm, S-GIMME, that takes into account heterogeneity across individuals with the aim of providing subgroup membership and precise information about the specific network structures that differentiate subgroups. The algorithm has previously provided robust and accurate classification when evaluated with large-scale simulation studies but has not yet been validated on empirical data. Here, we investigate S-GIMME's ability to differentiate, in a purely data-driven manner, between brain states explicitly induced through different tasks in a new fMRI dataset. The results provide new evidence that the algorithm was able to resolve, in an unsupervised data-driven manner, the differences between different active brain states in empirical fMRI data to segregate individuals and arrive at subgroup-specific network structures of edges. The ability to arrive at subgroups that correspond to empirically designed fMRI task conditions, with no biasing or priors, suggests this data-driven approach can be a powerful addition to existing methods for unsupervised classification of individuals based on their dynamic processes.

Bayesian Uncertainty Estimation for Gaussian Graphical Models and Centrality Indices

In the network approach to psychopathology, psychological constructs are conceptualized as networks of interacting components (e.g., the symptoms of a disorder). In this network view, interest is on the degree to which symptoms influence each other, both directly and indirectly. These direct and indirect influences are often captured with centrality indices, however, the estimation method often used with these networks, the frequentist graphical LASSO (GLASSO), has difficulty estimating (uncertainty in) these measures. Bayesian estimation might provide a solution, as it is better suited to deal with bias in the sampling distribution of centrality indices. This study therefore compares estimation of symptom networks with Bayesian GLASSO- and Horseshoe priors to estimation using the frequentist GLASSO using extensive simulations. Results showed that the Bayesian GLASSO performed better than the Horseshoe, and that the Bayesian GLASSO outperformed the frequentist GLASSO with respect to bias in edge weights, centrality measures, correlation between estimated and true partial correlations, and specificity. Sensitivity was better for the frequentist GLASSO, but performance of the Bayesian GLASSO is usually close. With respect to uncertainty in the centrality measures, the Bayesian GLASSO shows good coverage for strength and closeness centrality, but uncertainty in betweenness centrality is estimated less well.

The roles of caregivers and friends in adolescent daily emotion dynamics

Daily emotion dynamics provide valuable information about individuals' emotion processes as they go about their lives. Emotion dynamics such as emotion levels (mean), emotion variability (degree of fluctuation), and emotion network density (strength of temporal connections among emotions) are associated with risks for various psychopathology in youth and adults. Prior work has shown that caregivers and friends play crucial socializing roles in adolescent emotional well-being, but less is known about their roles in daily emotion dynamics. This study examined whether caregiver emotion coaching, caregiver-adolescent closeness, and friendship quality were associated with adolescents' emotion levels, emotion variability, and emotion network density. Further, we examined whether caregiver-adolescent closeness moderated the associations between coaching and emotion dynamics. Participants were 150 adolescents (61% girls; Mage = 14.75) and one of their caregivers (95% female; Mage = 43.35) who completed a baseline survey and 21 daily surveys. Results showed that caregiver emotion coaching interacted with caregiver-adolescent closeness in predicting emotion levels and variability. Specifically, when closeness was higher, emotion coaching was significantly associated with lower sadness and anger levels, higher happiness levels, and lower happiness variability. Caregiver emotion coaching, independent of closeness, was also associated with lower anxiety levels, lower sadness variability, and lower emotion network density. Friendship quality was significantly associated with lower levels of sadness, anxiety, and anger, higher levels of happiness, and lower variability in anxiety and anger. These findings suggest that caregivers and friends are central to everyday emotion levels and variability and a more flexible emotion system in adolescents.

Auditing the research practices and statistical analyses of the group-level temporal network approach to psychological constructs: A systematic scoping review

Network analyses have become increasingly common within the field of psychology, and temporal network analyses in particular are quickly gaining traction, with many of the initial articles earning substantial interest. However, substantial heterogeneity exists within the study designs and methodology, rendering it difficult to form a comprehensive view of its application in psychology research. Since the field is quickly growing and since there have been many study-to-study variations in terms of choices made by researchers when collecting, processing, and analyzing data, we saw the need to audit this field and formulate a comprehensive view of current temporal network analyses. To systematically chart researchers' practices when conducting temporal network analyses, we reviewed articles conducting temporal network analyses on psychological variables (published until March 2021) in the framework of a scoping review. We identified 43 articles and present the detailed results of how researchers are currently conducting temporal network analyses. A commonality across results concerns the wide variety of data collection and analytical practices, along with a lack of consistency between articles about what is reported. We use these results, along with relevant literature from the fields of ecological momentary assessment and network analysis, to formulate recommendations on what type of data is suited for temporal network analyses as well as optimal methods to preprocess and analyze data. As the field is new, we also discuss key future steps to help usher the field's progress forward and offer a reporting checklist to help researchers navigate conducting and reporting temporal network analyses.

The relationship between individual symptom connectivity and global eating disorder symptom severity

BACKGROUND

The network approach has emerged as a useful framework for conceptualizing and investigating psychopathology, including eating disorders. Network connectivity, that is, the density of the connections among network nodes, has been somewhat neglected despite its theoretical relevance. As predicted by network theory, symptom connectivity would be distinct but related to symptom severity and may be a useful clinical indicator of psychopathology as stronger and/or more diffuse connections among symptoms offer more avenues for symptom activation. This study aimed to investigate the relationship between moment-by-moment individual-level symptom connectivity and global levels of symptom severity in the context of eating disorder symptoms and experiences.

METHODS

A sample of 58 female undergraduate college students, mean (SD) age = 20.5 (3.1) provided data on eating disorder symptoms eight times a day over the course of 10 days. Network analyses were used to calculate the eating disorder symptoms network connectivity for each participant. In addition, participants completed survey of self-report measures of eating disorder symptom severity and trait mindfulness and body image flexibility.

RESULTS

Analyses revealed a moderate, positive relationship between individual network connectivity and eating disorder symptom severity. In addition, symptom connectivity predicted unique variance of symptom severity even after controlling for other clinically-relevant variables.

CONCLUSIONS

Individual-level network connectivity may be an important dimension of psychopathology and further work exploring the role of network connectivity is warranted.

PUBLIC SIGNIFICANCE

These findings suggest that symptom severity and the extent to which different eating disorder symptoms are connected are related but different dimensions. Investigating how these different dimensions play a role in eating disorder pathology could help to better understand and treat these disorders.

Peritraumatic physical symptoms and the clinical trajectory of PTSD after a terrorist attack: a network model approach

Introduction: Following a mass casualty event, such as the Paris terrorist attacks of 13 November 2015, first responders need to identify individuals at risk of PTSD. Physical peritraumatic symptoms involving the autonomic nervous system may be useful in this task.Objective: We sought to determine the trajectory of physical response intensity in individuals exposed to the Paris terrorist attacks using repeated measures, and to examine its associations with PTSD. Using network modelling, we examined whether peritraumatic physical symptom associations differed by PTSD status.Methods: Physical reactions were assessed using the Subjective Physical Reactions Scale at three time points: peritraumatic by retrospective recall, then current at one year (8-18 months) and three years (30-42 months) after the attacks. Interaction networks between peritraumatic physical reactions were compared according to PTSD status.Results: On the one hand, the reported intensity of physical reactions was significantly higher in the PTSD group at all time points. On the other hand, using the dynamic approach, more robust positive interactions between peritraumatic physical reactions were found in the PTSD group one and three years after the attacks. Negative interactions were found in the no-PTSD group at one year. Peritraumatic physical numbness was found to be the most central network symptom in the PTSD group, whereas it was least central in the no-PTSD group.Discussion: Network analysis of the interaction between peritraumatic physical subjective responses, particularly physical numbness, may provide insight into the clinical course of PTSD. Our knowledge of the brain regions involved in dissociation supports the hypothesis that the periaqueductal grey may contribute to the process leading to physical numbing.Conclusions: Our findings highlight the role of peritraumatic somatic symptoms in the course of PTSD. Peritraumatic physical numbness appears to be a key marker of PTSD and its identification may help to improve early triage.

ConNEcT: An R package to build contingency measure-based networks on binary time series

Dynamic networks are valuable tools to depict and investigate the concurrent and temporal interdependencies of various variables across time. Although several software packages for computing and drawing dynamic networks have been developed, software that allows investigating the pairwise associations between a set of binary intensive longitudinal variables is still missing. To fill this gap, this paper introduces an R package that yields contingency measure-based networks (ConNEcT). ConNEcT implements different contingency measures: proportion of agreement, corrected and classic Jaccard index, phi correlation coefficient, Cohen's kappa, odds ratio, and log odds ratio. Moreover, users can easily add alternative measures, if needed. Importantly, ConNEcT also allows conducting non-parametric significance tests on the obtained contingency values that correct for the inherent serial dependence in the time series, through a permutation approach or model-based simulation. In this paper, we provide an overview of all available ConNEcT features and showcase their usage. Addressing a major question that users are likely to have, we also discuss similarities and differences of the included contingency measures.

Analysis of dynamic psychological processes to understand and promote physical activity behaviour using intensive longitudinal methods: a primer

Physical activity behaviour displays temporal variability, and is influenced by a range of dynamic psychological processes (e.g., affect) and shaped by various co-occurring events (e.g., social/environmental factors, interpersonal dynamics). Yet, most physical activity research tends not to examine the dynamic psychological processes implicated in adopting and maintaining physical activity. Intensive longitudinal methods (ILM) represent one particularly salient means of studying the complex psychological dynamics that underlie and result from physical activity behaviour. With the increased recent interest in using intensive longitudinal data to understand specific dynamic psychological processes, the field of exercise and health psychology is well-positioned to draw from state-of-the-art measurement and statistical approaches that have been developed and operationalised in other fields of enquiry. The purpose of this review is to provide an overview of some of the fundamental dynamic measurement and modelling approaches applicable to the study of physical activity behaviour change, as well as the dynamic psychological processes that contribute to such change.

A Network Approach to Assessing the Relationship between Discrimination and Daily Emotion Dynamics

Discrimination-health research has been critiqued for neglecting the endogeneity of reports of discrimination to negative affect and the multidimensionality of mental health. To address these challenges, we model discrimination’s relationship to multiple psychological variables without directional constraints. Using time-dense data to identify associational network structures allows for joint testing of the social stress hypothesis, prominent in discrimination-health literature, and the negativity bias hypothesis, an endogeneity critique rooted in social psychology. Our results show discrimination predicts negative emotions from day-to-day but not vice versa, indicating that racial discrimination is a risk factor and not symptom of negative emotion. Furthermore, we identify sadness, guilt, hostility, and fear as a locus of interrelated emotions sensitive to racism-related stressors that emerges over time. Thus, we find support for what race scholars have argued for 120+ years in a model without a priori directional restrictions and then build on this work by empirically identifying cascading mental health consequences of discrimination.

Recovering Within-Person Dynamics from Psychological Time Series

Idiographic modeling is rapidly gaining popularity, promising to tap into the within-person dynamics underlying psychological phenomena. To gain theoretical understanding of these dynamics, we need to make inferences from time series models about the underlying system. Such inferences are subject to two challenges: first, time series models will arguably always be misspecified, meaning it is unclear how to make inferences to the underlying system; and second, the sampling frequency must be sufficient to capture the dynamics of interest. We discuss both problems with the following approach: we specify a toy model for emotion dynamics as the true system, generate time series data from it, and then try to recover that system with the most popular time series analysis tools. We show that making straightforward inferences from time series models about an underlying system is difficult. We also show that if the sampling frequency is insufficient, the dynamics of interest cannot be recovered. However, we also show that global characteristics of the system can be recovered reliably. We conclude by discussing the consequences of our findings for idiographic modeling and suggest a modeling methodology that goes beyond fitting time series models alone and puts formal theories at the center of theory development.

Back to Basics: The Importance of Conceptual Clarification in Psychological Science

Although the lack of conceptual clarity has been observed to be a widespread and fundamental problem in psychology, conceptual clarification plays a mostly marginal role in psychological research. In this article, we argue that better conceptualization of psychological phenomena is needed to move psychology forward as a science. We first show how conceptual unclarity seeps through all aspects of psychological research, from everyday concepts to statistical measures. We then turn to recommendations on how to improve conceptual clarity in psychology, emphasizing the importance of seeing research as an iterative process in which it is necessary to revisit the phenomena that are the foundations of theories and models, as well as how they are conceptualized and measured.

A Systematic Study into the Factors that Affect the Predictive Accuracy of Multilevel VAR(1) Models

The use of multilevel VAR(1) models to unravel within-individual process dynamics is gaining momentum in psychological research. These models accommodate the structure of intensive longitudinal datasets in which repeated measurements are nested within individuals. They estimate within-individual auto- and cross-regressive relationships while incorporating and using information about the distributions of these effects across individuals. An important quality feature of the obtained estimates pertains to how well they generalize to unseen data. Bulteel and colleagues (Psychol Methods 23(4):740-756, 2018a) showed that this feature can be assessed through a cross-validation approach, yielding a predictive accuracy measure. In this article, we follow up on their results, by performing three simulation studies that allow to systematically study five factors that likely affect the predictive accuracy of multilevel VAR(1) models: (i) the number of measurement occasions per person, (ii) the number of persons, (iii) the number of variables, (iv) the contemporaneous collinearity between the variables, and (v) the distributional shape of the individual differences in the VAR(1) parameters (i.e., normal versus multimodal distributions). Simulation results show that pooling information across individuals and using multilevel techniques prevent overfitting. Also, we show that when variables are expected to show strong contemporaneous correlations, performing multilevel VAR(1) in a reduced variable space can be useful. Furthermore, results reveal that multilevel VAR(1) models with random effects have a better predictive performance than person-specific VAR(1) models when the sample includes groups of individuals that share similar dynamics.

Taking Stock and Moving Forward: A Personalized Perspective on Mixed Emotions

Research on mixed emotions is flourishing but fractured. Several psychological subfields are working in parallel and separately from other disciplines also studying mixed emotions, which has led to a disorganized literature. In this article, we provide an overview of the literature on mixed emotions and discuss factors contributing to the lack of integration within and between fields. We present an organizing framework for the literature of mixed emotions on the basis of two distinct goals: solving the bipolar-bivariate debate and understanding the subjective experience of mixed emotions. We also present a personalized perspective that can be used when studying the subjective experience of mixed emotions. We emphasize the importance of assessing both state and trait emotions (e.g., momentary emotions, general levels of affect) alongside state and trait context (e.g., physical location, culture). We discuss three methodological approaches that we believe will be valuable in building a new mixed-emotions literature-inductive research methods, idiographic models of emotional experiences, and empirical assessment of emotion-eliciting contexts. We include recommendations throughout on applying these methods to research on mixed emotions, and we conclude with avenues for future interdisciplinary research. We hope that this perspective will foster research that results in the organized accumulation of knowledge about mixed emotions.

Examining the associations between PTSD symptoms and aspects of emotion dysregulation through network analysis

Despite the clearly established link between posttraumatic stress disorder (PTSD) and emotion dysregulation, little is known about how individual symptoms of PTSD and aspects of emotion dysregulation interrelate. The network approach to mental health disorders provides a novel framework for conceptualizing the association between PTSD and emotion dysregulation as a system of interacting nodes. In this study, we estimated the structural relations among PTSD symptoms and aspects of emotion dysregulation within a large sample of women who participated in a multi-site study of sexual revictimization (N = 463). We estimated expected influence to reveal differential associations among PTSD symptoms and aspects of emotion dysregulation. Further, we estimated bridge expected influence to identify influential nodes connecting PTSD symptoms and aspects of emotion dysregulation. Results highlighted the key role of concentration difficulties in expected influence and bridge expected influence. Findings highlight several PTSD symptoms and aspects of emotion dysregulation that may be targets for future intervention.

Changed dynamic symptom networks after a self-compassion training in patients with somatic symptom disorder: A multiple single-case pilot project

OBJECTIVE

Pre-to-post mean group differences of intermittently assessed generic outcome variables may not capture all relevant treatment-related changes in individual patients with somatic symptom disorder (SSD). Aim of this multiple single-case observational pilot project was to find out whether the Experience Sampling Method (ESM) and dynamic symptom networks may offer new opportunities in evaluating treatment outcomes for individual patients with SSD.

METHODS

Patients with SSD (N = 6 in study 1, N = 7 in study 2) received a self-compassion training in a tertiary care mental health expert center. Using a single-case pre-post treatment observational design, intensive longitudinal data were collected with ESM. A brief questionnaire was presented via the patient's smartphone three times per day for 16 weeks before, during and after the training in study 1, and for 5 weeks before and 5 weeks after the training in study 2. Eleven questions comprised somatic symptoms, functional disability, stress, self-compassion, and acceptance of affect; three personalized questions comprised self-chosen affects and an additional symptom.

RESULTS

Sufficient observations for means and network comparison were obtained for 11 and 10 patients, respectively. After the training, self-compassion was significantly increased in 10 patients, functional disability, stress and affect improved in 6 patients, and (although not a treatment goal) somatic symptoms decreased in 6 patients. Dynamic symptom networks significantly changed in 5 patients.

CONCLUSION

Patient-specific changes in means and dynamic symptom networks were observed after self-compassion training. In future clinical trials, single-case ESM may offer new opportunities to evaluate treatment outcomes in patients with SSD.

Why are you (un)conscientious? The dynamic interplay of goals, states, and traits in everyday life

Personality involves both trait and state components, personal goals serving a crucial regulatory function for the expression of personality states. The present study investigates the dynamic interplay between conscientiousness-related goals, conscientious personality states, and trait conscientiousness. A sample of 244 community participants responded to a baseline survey (T1), a 5-times-a-day Ecological Momentary Assessment (EMA) for 15 days, and a post-EMA survey (T2). Pre-registered multilevel analyses indicated significant contemporaneous positive and negative associations between momentary conscientious and unconscientious goals and state conscientiousness, respectively. Cross-lagged associations also emerged, with goals predicting future states of conscientiousness. A latent growth model was fitted on a subsample of participants (N = 159). Results indicated that change in trait conscientiousness from T1 to T2 was explained by growth in conscientiousness-related goals during the EMA phase, with a mediating effect of growth in state conscientiousness. Overall, the results corroborate the importance of goals for modeling contemporaneous and cross-lagged personality dynamics, both in short and longer timeframes.

Psychopathological networks: Theory, methods and practice

In recent years, network approaches to psychopathology have sparked much debate and have had a significant impact on how mental disorders are perceived in the field of clinical psychology. However, there are many important challenges in moving from theory to empirical research and clinical practice and vice versa. Therefore, in this article, we bring together different points of view on psychological networks by methodologists and clinicians to give a critical overview on these challenges, and to present an agenda for addressing these challenges. In contrast to previous reviews, we especially focus on methodological issues related to temporal networks. This includes topics such as selecting and assessing the quality of the nodes in the network, distinguishing between- and within-person effects in networks, relating items that are measured at different time scales, and dealing with changes in network structures. These issues are not only important for researchers using network models on empirical data, but also for clinicians, who are increasingly likely to encounter (person-specific) networks in the consulting room.

Dynamic Mixture Modeling with dynr

Mixture modeling is commonly used to model sample heterogeneity by identifying unobserved classes of individuals with similar characteristics. Despite abundance of evidence in the literature suggesting that individuals are often characterized by different dynamic processes underlying their physiological, cognitive, psychological, and behavioral states, applications of dynamic mixture modeling are surprisingly lacking. We present here a proof-of-concept example of dynamic mixture modeling, where latent groups of individuals were identified based on different dynamic patterns in their time series. Our sample consists of 192 men who were in a heterosexual relationship. They were asked to complete a daily questionnaire involving emotions related to their relationship. Two latent groups were identified based on the strength of association between positive (e.g., loving) and negative (e.g., doubtful) affect. Men in the group characterized by a strong negative association (β=-.67) tended to be younger and had higher levels of anxiety toward their relationship than men in the other group, which was characterized by a weaker negative association (β=-.31). We illustrate the specification and estimation of dynamic mixture model using "dynr," an R package capable of handling a broad class of linear and nonlinear discrete- and continuous-time models with regime-switching properties.

Emotion network density in burnout

Background

Health care workers are often affected by burnout, resulting in reduced personal well-being and professional functioning. Although emotional exhaustion is considered a core component of burnout, little is known about the dynamics of emotions and their relation to burnout. We used network analysis to investigate the correlation between the density of a negative emotion network, a marker for emotional rigidity in person-specific networks, and burnout severity.

Methods

Using an ecological momentary assessment design, the intensity of negative emotions of forty-three health care workers and medical students was assessed five times per day (between 6 am and 8 pm) for 17 days. Burnout symptoms were assessed at the end of the study period with the Maslach Burnout Inventory. Multilevel vector autoregressive models were computed to calculate network density of subject-specific temporal networks. The one-sided correlation between network density and burnout severity was assessed. The study protocol and analytic plan were registered prior to the data collection.

Results

We found a medium-sized correlation between the negative emotion network density and burnout severity at the end of the study period r(45) = .32, 95% CI = .09–1.0, p = .014).

Conclusions

The strength of the temporal interplay of negative emotions is associated with burnout, highlighting the importance of emotions and emotional exhaustion in reaction to occupational-related distress in health care workers. Moreover, our findings align with previous investigations of emotion network density and impaired psychological functioning, demonstrating the utility of conceptualizing the dynamics of emotions as a network.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-021-00670-y.

Emotion networks across self-reported depression levels during the COVID-19 pandemic

ABSTRACTDuring stressful circumstances, such as the COVID-19 pandemic, disturbances in emotional experiences can occur. These emotional disturbances, if not relieved or regulated, can be associated with feelings of depression. Currently, little is known about which emotional experiences (positive and negative) are associated with feelings of depression during COVID-19. This study aimed to estimate and compare mixed, positive and negative valence emotion networks during COVID-19 for low, moderate and high levels of self-reported depression. Across 26,034 participants, central emotional experiences included gratitude, sadness, fear, anxiety, compassion, and being moved for all self-reported depression levels; love for low levels of depression, and confusion for high levels of depression. The strongest edges included fear-anxiety, loneliness-boredom, anger-disgust, determination-hope, and compassion-being moved for all self-reported depression levels; calm-relief, and sadness-frustration for high levels of self-reported depression; and admiration-being moved for low and moderate self-reported depression levels. Network comparison tests showed mixed, positive and negative emotion networks significantly differed in structure across all self-reported depression levels. Network connectivity was also significantly stronger for low self-reported depression within positive and negative emotion networks. These networks provide key information on emotional experiences associated with depression during COVID-19.

Little evidence for sex or ovarian hormone influences on affective variability

Women were historically excluded from research participation partly due to the assumption that ovarian hormone fluctuations lead to variation, especially in emotion, that could not be experimentally controlled. Although challenged in principle and practice, relevant empirical data are limited by single measurement occasions. The current paper fills this knowledge gap using data from a 75-day intensive longitudinal study. Three indices of daily affective variability-volatility, emotional inertia, and cyclicity-were evaluated using Bayesian inferential methods in 142 men, naturally cycling women, and women using three different oral contraceptive formulations (that "stabilize" hormone fluctuations). Results provided more evidence for similarities between men and women-and between naturally cycling women and oral contraceptive users-than for differences. Even if differences exist, effects are likely small. Thus, there is little indication that ovarian hormones influence affective variability in women to a greater extent than the biopsychosocial factors that influence daily emotion in men.

Visual Expression of Emotion in Dynamic 3D Painting System Based on Emotion Synthesis Model

Emotion is a unique ability possessed by human beings as advanced creatures. Emotions give people a unique physical and mental experience. Assigning emotions to computer systems is one of the latest topics in artificial intelligence research. The purpose is to allow machines to achieve natural coordination between humans and computers. This article focuses on the visual expression of emotion in the dynamic three-dimensional painting system, creating an intelligent painting system and realizing a good user experience. In this paper, the discrete method is used to qualitatively analyze emotions, and the continuous method is used to quantify basic emotions, and emotional modeling and emotional quantitative analysis are proposed to realize quantitative analysis of emotions. Combining these two methods, a comprehensive method is proposed, which uses a continuous method to quantify the basic emotions of each discrete dimension, and finally superimposes them into a comprehensive emotional synthesis model. Emotion modeling is the basis of emotion visualization. Borrowing the relationship between emotion synthesis model and visual emotion elements, this article puts forward the concept of qualitative and quantitative visual emotion elements, and expounds that the multidimensional superposition of visual emotion elements makes dynamic three-dimensional painting system emotions. The experimental results in this article show that the emotional visualization scheme of 100 samples is tested by quantitative statistical methods to demonstrate its effectiveness. Starting from 5 points of concern, the emotion visualization method discussed in this article can indeed convey or suggest a certain positive emotion (the average value of experience, transitivity, and infectiousness > 2.5, and the variance is close to 0), but we also found this recognition at the same time The degree is not high enough, and individual differences are large (mean value < 2.5, variance close to 1). This can indicate that different subjects have different feelings and evaluations of this emotional visualization. As long as the difference is within a reasonable range, this emotional visualization also has practical value, and has the ability to convey or suggest emotions.

Clustering Individuals on Limited Features of a Vector Autoregressive Model

Dynamical interplays in emotions have been investigated using vector autoregressive (VAR) models, whose estimates can be used to cluster participants into unknown groups. The present study evaluated a clustering algorithm, the alternating least square (ALS) algorithm, for accuracy in predicting individual group membership. We systematically manipulated (a) the number of variables in a model, (b) the size of group differences in regression coefficients, and (c) the number of regression coefficients that vary across the groups (i.e., effective features). The ALS algorithm works reliably when there are at least 5 effective features with very large group differences in a 5-variable model; and 9 effective features with very large group differences in a 10-variable model. These findings suggest that the ALS algorithm is sensitive to group differences that are present only in several coefficients of a VAR model, but that the group differences have to be large. We also found that the ALS algorithm outperforms another clustering method, Gaussian mixture modeling. The ALS algorithm was further evaluated with unbalanced sample sizes between groups and with a greater number of groups in data (Study 2). A real data application was provided to illustrate how to interpret the detected group differences (Study 3).

Assessing psychological symptom networks related to HIV-positive duration among people living with HIV: a network analysis

This study aims to explore and visualize relationships among multiple psychological symptoms among people living with HIV (PLWH) with different HIV-positive durations and to compare centrality indices and densities of psychological symptom networks. We used subsets of data collected from five designated HIV/AIDS hospitals in China. Networks were constructed among 16 psychological symptoms. Centrality properties, including strength and closeness, were adopted to describe relationships among symptoms. The results showed that PLWH with longer HIV-positive durations had denser emotional networks, which indicated that they had more emotional neuroticism than their newly diagnosed counterparts. Sadness, self-abasement, and self-loathing were the most central psychological symptoms across different HIV-positive durations. Our study suggests the need to provide psychosocial support services targeting PLWH according to changing symptom severity and neuroticism trajectories. Interventions should focus on increasing empathy for PLWH and enhancing the ability to consider the situation from different perspectives to avoid the development of neuroticism in long-term survivors.

Emotion network density is a potential clinical marker for anxiety and depression: Comparison of ecological momentary assessment and daily diary

OBJECTIVES

Using two intensive longitudinal data sets with different timescales (90 minutes, daily), we examined emotion network density, a metric of emotional inflexibility, as a predictor of clinical-level anxiety and depression.

DESIGN

Mobile-based intensive longitudinal assessments.

METHODS

119 participants (61 anxious and depressed, 58 healthy controls) completed ecological momentary assessment (EMA) to rate a variety of negative (NE) and positive emotions (PE) 9 times per day for 8 days using a mobile phone application. 169 participants (97 anxious and depressed and 72 healthy controls) completed an online daily diary on their NE and PE for 50 days. Multilevel vector autoregressive models were run to compute NE and PE network densities in each data set.

RESULTS

In the EMA data set, both NE and PE network densities significantly predicted participants' diagnostic status above and beyond demographics and the mean and standard deviation of NE and PE. Greater NE network density and lower PE network density were associated with anxiety and depression diagnoses. In the daily diary data set, NE and PE network densities did not significantly predict the diagnostic status.

CONCLUSIONS

Greater inflexibility of NE and lower inflexibility of PE, indexed by emotion network density, are potential clinical markers of anxiety and depressive disorders when assessed at intra-daily levels as opposed to daily levels. Considering emotion network density, as well as the mean level and variability of emotions in daily life, may contribute to diagnostic prediction of anxiety and depressive disorders.

PRACTITIONER POINTS

Emotion network density, or the degree to which prior emotions predict and influence current emotions, indicates an inflexible or change-resistant emotion system. Emotional inflexibility or change resistance over a few hours, but not daily, may characterize anxiety and depressive disorders. Inflexible negative emotion systems are associated with anxiety and depressive disorders, whereas inflexible positive emotion systems may indicate psychological health. Considering emotional inflexibility within days may provide additional information beyond demographics and mean level and variability of emotions in daily life for detecting anxiety and depressive disorders. ​.

The Mental Health Ecosystem: Extending Symptom Networks With Risk and Protective Factors

Inspired by modeling approaches from the ecosystems literature, in this paper, we expand the network approach to psychopathology with risk and protective factors to arrive at an integrated analysis of resilience. We take a complexity approach to investigate the multifactorial nature of resilience and present a system in which a network of interacting psychiatric symptoms is targeted by risk and protective factors. These risk and protective factors influence symptom development patterns and thereby increase or decrease the probability that the symptom network is pulled toward a healthy or disorder state. In this way, risk and protective factors influence the resilience of the network. We take a step forward in formalizing the proposed system by implementing it in a statistical model and translating different influences from risk and protective factors to specific targets on the node and edge parameters of the symptom network. To analyze the behavior of the system under different targets, we present two novel network resilience metrics: Expected Symptom Activity (ESA, which indicates how many symptoms are active or inactive) and Symptom Activity Stability (SAS, which indicates how stable the symptom activity patterns are). These metrics follow standard practices in the resilience literature, combined with ideas from ecology and physics, and characterize resilience in terms of the stability of the system's healthy state. By discussing the advantages and limitations of our proposed system and metrics, we provide concrete suggestions for the further development of a comprehensive modeling approach to study the complex relationship between risk and protective factors and resilience.

A comparison of continuous and discrete time modeling of affective processes in terms of predictive accuracy

Intra-individual processes are thought to continuously unfold across time. For equally spaced time intervals, the discrete-time lag-1 vector autoregressive (VAR(1)) model and the continuous-time Ornstein–Uhlenbeck (OU) model are equivalent. It is expected that by taking into account the unequal spacings of the time intervals in real data between observations will lead to an advantage for the OU in terms of predictive accuracy. In this paper, this is claim is being investigated by comparing the predictive accuracy of the OU model to that of the VAR(1) model on typical ESM data obtained in the context of affect research. It is shown that the VAR(1) model outperforms the OU model for the majority of the time series, even though time intervals in the data are unequally spaced. Accounting for measurement error does not change the result. Deleting large abrupt changes on short time intervals (that may be caused by externally driven events) does however lead to a significant improvement for the OU model. This suggests that processes in psychology may be continuously evolving, but that there are factors, like external events, which can disrupt the continuous flow.

Network dynamics of momentary affect states and future course of psychopathology in adolescents

BACKGROUND

Recent theories argue that an interplay between (i.e., network of) experiences, thoughts and affect in daily life may underlie the development of psychopathology.

OBJECTIVE

To prospectively examine whether network dynamics of everyday affect states are associated with a future course of psychopathology in adolescents at an increased risk of mental disorders.

METHODS

159 adolescents from the East-Flanders Prospective Twin Study cohort participated in the study. At baseline, their momentary affect states were assessed using the Experience Sampling Method (ESM). The course of psychopathology was operationalized as the change in the Symptom Checklist-90 sum score after 1 year. Two groups were defined: one with a stable level (n = 81) and one with an increasing level (n = 78) of SCL-symptom severity. Group-level network dynamics of momentary positive and negative affect states were compared between groups.

RESULTS

The group with increasing symptoms showed a stronger connections between negative affect states and their higher influence on positive states, as well as higher proneness to form 'vicious cycles', compared to the stable group. Based on permutation tests, these differences were not statistically significant.

CONCLUSION

Although not statistically significant, some qualitative differences were observed between the networks of the two groups. More studies are needed to determine the value of momentary affect networks for predicting the course of psychopathology.

Emotions as Overlapping Causal Networks of Emotion Components: Implications and Methodological Approaches

A widespread perspective describes emotions as distinct categories bridged by fuzzy boundaries, indicating that emotions are distinct and dimensional at the same time. Theoretical and methodological approaches to this perspective still need further development. We conceptualize emotions as overlapping networks of causal relationships between emotion components—networks representing distinct emotions share components with and relate to each other. To investigate this conceptualization, we introduce network analysis to emotion research and apply it to the reanalysis of a data set on multiple positive emotions. Specifically, we describe the estimation of networks from data, and the detection of overlapping communities of nodes in these networks. The network perspective has implications for the understanding of distinct emotions, their co-occurrence, and their measurement.

[Network model of mental disorders: Application and interest in post-stroke depression]

In contrast to the classic models in psychopathology, the network model considers that the temporal interactions between symptoms are the causes of their occurrence. This model could also be particularly suitable for understanding the processes involved in post-stroke depression. The aim of this paper is to perform a network analysis in order to describe the temporal dynamic of the links existing between depression symptoms during the acute phase after stroke. Twenty-five patients (64% male, mean age 58.1±14.9 years old) hospitalized for a minor stroke (no neurocognitive or motor impairment) were involved in an Ecological Momentary Assessment methodology-based study. They used a smartphone application in order to complete four brief questionnaires each day during the week after hospital discharge. The questionnaire included 7-point Likert scales to measure the severity of the following depressive symptoms: sadness, anhedonia, fatigue, diminished concentration ability, negative thoughts on oneself, pessimism. We used Multilevel Vector Autoregressive analysis to describe the temporal links between those symptoms. We used the software R 3.6.0 with the mlVAR package. The p-value was set at .05. The results show two independent symptoms networks. The first one involves the anhedonia, fatigue, negative thoughts on oneself and sadness. It shows that: anhedonia predicts the activation of later fatigue (β=0.135, P=0.037) and later negative thoughts (β=0.152, P=0.019); negative thoughts predict later negative thoughts (β=0.143, P=0.028) and later sadness (β=0.171, P=0.021); fatigue predicts later fatigue (β=0.261, P<0.000). Pessimism and diminished concentration ability compose the second network, and the results show that pessimism predicts later pessimism (β=0.215, P=0.012) and later diminished concentration ability (β=0.178, P=0.045). On the one hand, anhedonia thus plays an important role in the initial and progressive activation of the other symptoms of its network. On the other hand, the cognitive symptoms (negative thoughts and pessimism) cause the deterioration of the mood and the deficit of attentional abilities. Using behavioral and cognitive strategies to support patients after hospital discharge would reduce the risk of depressive complications after a stroke. This study provides convincing empirical elements for the interest of the network model for research in psychopathology and the clinical implications and perspectives allowed by network analysis.