Search

Since depression is a heterogeneous syndrome composing of different clinical symptoms, it is likely that its constituting symptoms are differentially related to various vulnerability factors of the disorder. However, much remains to be discovered about how individual symptoms of depression are related to such purported vulnerability factors. This study aims to learn more about the association between particular depressive symptoms and dampening, a way of responding to positive emotions that has been shown to be related to depression in general. To reach that purpose, two approaches will be adopted. Analyses will be run on existing cross-sectional and longitudinal data from previous studies of our lab, mainly collected in community samples consisting of adolescents and adults. First, via network analysis (NA) on cross-sectional data, we will reveal the relations between different depressive symptoms and, especially how dampening relates to each particular symptom. The network will be further analysed by estimating the centrality of the symptoms. Second, through random forest machine learning (ML) on longitudinal data, we will identify which dampening thoughts serve as most important predictors of depressive symptoms. If possible, the learning models’ performance will be cross-validated by splitting datasets. Also, to verify findings, analyses will be run on several validated depression scales. For the longitudinal data, time intervals adopted in the different studies slightly differ. Nevertheless, the longitudinal data will be merged to obtain a sufficiently large data set to explore the predictive validity of the different dampening features/thoughts. Although emotion regulation features appear to differ in terms of temporal stability of their predictive validity, the predictive validity of the most potent emotion regulation features has been shown to be relatively stable over time (Everaert et al., 2022). So, we expect to be able to especially learn more about those dampening thoughts that have considerable predictive validity. After main analyses have been run, exploratory follow-up analyses will be conducted to detect the potential effect of age on the associations between depressive symptoms and dampening. In addition, if dampening appears to be especially associated with depressive symptoms reflecting anhedonia, similar analyses (ML and NA) will be conducted to shed light on which particular features of anhedonia (reflecting consummatory, anticipatory or motivational difficulties) are most closely related to dampening (NA) and which symptoms of dampening best predict symptoms of anhedonia (ML).

The ability to organize and segment continuous input of information to extract embedded patterns of regularity in the environment is critical to understanding what is happening around us. The process through which people extract ‘regularities’ or probabilistic relationships between recurrent events across space and time is referred to as statistical learning (Frost, Armstrong, & Christiansen, 2019; Frost, Armstrong, Siegelman, & Christiansen, 2015; Schapiro & Turk-Browne, 2015). Though research has widely studied statistical learning of regularities within flows of neutral information (for reviews, see Frost et al., 2019; Krogh, Vlach, & Johnson, 2013; Santolin & Saffran, 2018; Siegelman, Bogaerts, & Frost, 2017), researchers have only recently started to investigate whether people can learn and extract regularities within continuous input of emotional information (Everaert, Koster, & Joormann, 2018). In a recent study (Everaert, Koster, & Joormann, 2020; osf.io/tu9ps, "Experiment 1"), we found that people are able to detect co-occurrences among positive and negative verbal stimuli without instruction to detect patterns, intention to learn, or explicit awareness. Examining factors that may modulate statistical learning, we observed that naturally occurring positive mood state was positively associated with the extraction of statistical regularities within positive words. The present study aims to replicate these initial observations and extend them by exploring the role of trait happiness in modulating statistical learning of emotional information.