Background: Childhood maltreatment is associated with reduced facial emotion recognition (FER), which may impair the interpretation of emotional states during social interactions. However, the underlying mechanisms of this association remain unclear. This study investigates whether depressive symptoms, aggression, and psychological inflexibility mediate the association between childhood maltreatment and FER accuracy in youth. Additionally, we examine differential associations of threat- and deprivation-related childhood maltreatment with FER accuracy.

Method: Cross-sectional data were collected from 246 adolescents and young adults aged 14 to 24 years (M = 20.59 years, SD = 2.2 years). Childhood maltreatment was assessed using the Childhood Trauma Questionnaire (CTQ-SF) separated into its threat (abuse) and deprivation (neglect) dimensions. The possible mediators depressive symptoms, aggression, and psychological inflexibility were measured using validated self-report questionnaires. FER accuracy was assessed using the Penn Emotion Recognition Test (ER-40), yielding overall and emotion-specific accuracy scores. Linear regression and mediation analyses were conducted using bootstrapping procedures, controlling for age and sex.

Results: Both threat- and deprivation-related childhood maltreatment was associated with lower overall FER accuracy. Only the threat dimension showed emotion-specific associations, particularly with reduced recognition of happiness, fear, and sadness. Aggression significantly partially mediated the relationship between both maltreatment dimensions and FER accuracy. Depressive symptoms and psychological inflexibility did not emerge as significant mediators.

Although parenting capacity plays a central role in legal decision-making and child welfare, there is no consensus on a definition or way of measuring the construct. This scoping review examines definitions, indicators, and assessment of parenting capacity within the contexts of child custody and child protection evaluations. Following PRISMA-ScR guidelines, we conducted a systematic literature search across five databases, resulting in the inclusion of 19 studies. We systematically synthesized the terminology, definitions, indicators, and assessment approaches used in these studies. In total, 115 indicators across twelve overarching indicator domains were identified. The findings reveal a shared conceptual core alongside substantial variability. Based on our analysis, we propose a theoretically grounded definition and conceptualization of parenting capacity. We discuss the available assessment approaches and instruments and their limitations, as well as the need for using clearer theoretical and operational frameworks in research and practice.

Objectives: Major depressive disorder (MDD) is a leading cause of disability worldwide, and accurate identification of individuals who will develop a first episode is a precondition for indicated prevention. A growing number of studies have developed clinical prediction models for the prospective onset of MDD, but this literature has not yet been synthesized. We systematically reviewed and meta-analyzed studies developing or evaluating multivariable clinical prediction models for the prospective first onset of MDD, appraised their risk of bias and applicability, and assessed the completeness of their reporting.

Methods: We searched PubMed, Scopus, Web of Science, PsycINFO, PSYNDEX and Embase from database inception to 27 October 2025, supplemented by forward and backward citation searching. Eligible studies were peer-reviewed, longitudinal, human studies in which a multivariable prediction model, validated at least internally, was used to predict the first onset of MDD or a major depressive episode, and in which at least one measure of discrimination was reported. Records were screened in two stages by two independent reviewers, and data were extracted in duplicate. Risk of bias and applicability were assessed with PROBAST+AI and reporting completeness with TRIPOD+AI. The unit of synthesis was the validation: each external validation cohort contributed a separate estimate, whereas studies reporting only internal validation contributed a single estimate from the model the original authors recommended. Logit-transformed AUROCs were pooled in a random-effects model with cluster-robust variance estimation at the study level.

Results: Nine studies published between 2014 and 2025 met the inclusion criteria, contributing 14 prediction models of which 18 validations from eight studies entered the meta-analysis; one study reported no AUROC and is reviewed narratively. Development samples ranged from 25 to 184,669 participants and outcome event counts from 10 to 3,964, with event rates between 1.5% and 40%. Seven studies ascertained the outcome with a structured or semi-structured diagnostic interview and two used routinely collected diagnostic codes. Reported AUROCs ranged from 0.59 to 0.84. The pooled AUROC was 0.696 (cluster-robust 95% CI 0.590 to 0.785; 95% prediction interval 0.459 to 0.861), with substantial heterogeneity (tau squared = 0.095 on the logit scale; I squared = 97.4%). Calibration was reported in five of nine studies, was poor wherever it was examined without recalibration, and was too sparsely and too heterogeneously reported to pool. A formal decision curve analysis was reported in one study. No study was rated at low risk of bias for its model evaluation; six were rated high and three unclear. Median adherence to TRIPOD+AI was 57.8% (range 44.4% to 65.2%), with universal non-reporting of protocol availability, registration, patient and public involvement, and model usability guidance.

Conclusions: Models for the first onset of MDD report discrimination in a range comparable to established prognostic models in other fields, but the prediction interval includes chance-level performance, and the evidence base is limited by small numbers of events, heterogeneous outcome definitions, a scarcity of external validation, and near-absent calibration and clinical utility evidence. No model can currently be recommended for use in indicated prevention.

Background: Adolescence is a decisive window for mental health; and timely care uptake is crucial to prevent negative long-term consequences. Yet adolescents access formal care less than any other age group. During informal help-seeking (i.e., interactions with non-mental-health professionals), caregivers are typically the first point of contact and act as gatekeepers to formal services, shaping whether and how further care is sought. Yet the daily interactional dynamics through which adolescents and caregivers communicate about mental health problems remain poorly understood and neglected in current help-seeking models.

Objective: At BEACON, the protocol of this mixed-methods project will be presented. It aims to examine the temporal dynamics of informal help-seeking interactions between help-seeking adolescents not yet engaged in formal care and their caregivers, and to contextualize these dynamics within lived experience.

Methods: Seventy adolescent-caregiver dyads will complete baseline and follow-up questionnaires and a 28-day intensive longitudinal assessment (event-contingent post-interaction and end-of-day surveys) capturing interaction quality, interpersonal behavior, socio-cognitive processes, affect, and relationship functioning, analyzed via Longitudinal Actor-Partner Interdependence Models and Dynamic Structural Equation Modeling. A subsample of 15-20 dyads will complete dyadic interviews, and separate focus groups (5 adolescents and caregivers each) will be analyzed via reflexive thematic analysis.

Expected Outcomes: This project will provide novel insight into how adolescents and caregivers recognize, communicate, and respond to distress in daily life. Findings are expected to advance existing conceptual models of help-seeking and may point toward potentially modifiable targets, offering a starting point for interventions that support timely access to mental health care.

Background: Coercive measures have a range of harmful effects on patients and staff alike. Despite increasing evidence for these effects, coercion numbers stayed comparably stable in Swiss psychiatry over the last years. Coercion prevention research may profit from the shift towards personalized treatments, risk identification and early prevention in precision psychiatry. Machine learning-based prediction could open up opportunities for early coercion prevention by identifying patients at risk of receiving a coercive measure.

Aims: To predict the application of coercive measures during inpatient stays using routine data from over 40 Swiss psychiatric hospitals available at admission. To this end, we will compare the predictive performance of different machine learning models with that of classical logisticregression.

How a society weighs the promised benefits of a new technology against its risks shapes whether the technology is adopted. In democracies–and especially in direct democracies such as Switzerland, where citizens vote on the rules governing technology–that weighing is largely done by non-experts, informed by the risks and benefits made salient in public discourse. Genome editing, spanning agriculture and medicine, is a case in point.

Three projects map this discourse and its effects: the first mines topics and arguments from decades of Swiss newspaper coverage across language regions; the second elicits mental representations and media use in a population survey to derive audience segments; and the third experimentally tests which argument types and formats—text, video, AI chat—engage which segments.

In the first project, from an archive of over 28 million Swiss newspaper articles, we identified over 26,000 articles from the past 35 years containing genome-editing keywords. Open-weight large language models extracted the applications discussed in each article; embedding and clustering these labels yields two meta-clusters, agriculture and medicine. Sentiment differs between them: coverage of medical applications has been consistently rather positive, whereas agricultural applications have become markedly more positive in recent years than in earlier decades. In ongoing work, we compare risk and benefit attributions across applications, sources, and over time.

Our pipeline can track discourse on other contested topics at scale. Learning how a democratic society debates the risks and safety of technology—and the role traditional and new media play in it—illuminates how consequential adoption decisions are made.

Metacognition comprises two processes: monitoring one’s knowledge states (e.g., confidence judgments) and using these judgments to guide control (e.g., selecting items to restudy). Confidence judgments are related to restudy decisions, but existing theoretical accounts make conflicting predictions about this relation, leaving the underlying mechanism unclear. Moreover, restudying requires choosing among multiple items, suggesting a competitive process that has not yet been formally modeled. We formalize these competing theoretical accounts as alternative uncertainty-to-value mappings within a unified computational framework that shares a common decision mechanism. Within this value-based choice framework, we operationalize discrepancy reduction, criterion-based accounts, region-of-proximal learning, and context-dependent evaluation as four computational models. We first compared the predictions of these models in simulations and subsequently tested them in two paired-associate learning studies. In Study 1 (N = 78), participants freely chose how many items to restudy (1–10); in Study 2 (N = 364), they selected at least 10 of 30 items. Model comparisons and participant-level leave-one-out cross-validation (LOOCV) revealed that the best-fitting model differed across decision environments: the Threshold-like Value Model provided the best fit in Study 1, whereas the Relative Value Model provided the best fit in Study 2. The Quadratic Value Model performed substantially worse in both studies. Taken together, these findings offer a computational account of metacognitive control and provide first insights into how principles from prominent metacognitive theories, including region-of-proximal-learning and discrepancy-reduction, can be implemented within a common value-based choice framework.

Explaining differences in learning outcomes is a central goal of educational research. While fluid intelligence (FI) and working memory (WM) are well-established predictors, educational developments emphasize self-regulatory skills. This includes metacognitive control (MC), the ability to make strategic learning decisions (e.g., restudy) based on monitoring. Despite theoretical links, the interrelations among WM, FI, MC, and learning outcomes remain unclear.

The present study addresses (1) whether FI and WM explain differences in MC and (2) to what extent FI, WM, and MC uniquely predict learning outcomes. 352 adolescents (Mean age = 13.7) completed three WM tasks, a matrix FI task, and a concept learning task with self-testing, monitoring, and restudy decisions. Structural equation modelling tested direct and indirect contributions to task performance and grades.

Results indicate that WM (β = .17) and intelligence (β = .16) both significantly predict MC, explaining 8% of its variance. All three predictors are significantly associated with task performance: intelligence (β = .16), WM (β = .42,), and MC (β = .38), together explaining 51% of its variance. Intelligence (β = .21) and WM (β = .24) also significantly predict GPA, whereas metacognitive control does not uniquely predict GPA (β = .11). The model explains 19% of variance in GPA.

Overall, these results support interrelations between MC, WM, and FI. All three uniquely contribute to learning task performance, whereas only WM and FI uniquely contribute to GPA. This pattern suggests a more proximal contribution of MC, whereas FI and WM contribute to broader learning outcomes.

International assessments report a concerning decline in students’ academic performance, particularly in reading. Because successful learning depends on extracting and integrating information from texts, understanding the metacognitive processes involved is crucial. Reading-based learning requires monitoring and regulation of comprehension and might be especially challenging for students with reading difficulties and multilingual students whose home language differs from the language of instruction. We investigate students’ metacognitive monitoring, regulation, and comprehension during reading-based learning, focusing on individual differences in reading ability and multilingual background.

We report preliminary findings from an ongoing study with elementary school students (n = 79). After reading short texts, students completed a comprehension test with item-level confidence judgments, selected items for restudy, and completed a final comprehension test. Multilevel models examined associations between confidence and initial performance, restudy decisions, and final performance, as well as moderation by reading ability and multilingual background.

Confidence was positively related to initial accuracy (β = 1.16, p < .05), indicating accurate metacognitive monitoring. Confidence also guided regulation: students with lower confidence were more likely to select items for restudy (β = −0.33, p < .05). However, restudying was not associated with higher final performance (β = 0.13, p > .05). The confidence–restudy association was weaker among students with higher reading ability (β = 0.14, p < .05) and multilingual students (β = 0.40, p < .01). These findings suggest that accurate monitoring does not necessarily translate into effective regulation yet and highlight the importance of individual differences when designing targeted support.

Perceived societal threats are key drivers of political extremism. Previous research has shown the impact of both socio-cultural and economic threats, but it remains unclear whether these different types of threat fuel extremism differently. We examined this across four experiments (total N = 1751). In Experiment 1, participants were exposed to both threat types in a within-subjects design. Experiment 2 replicated this design without framing. Experiments 3 & 4 used a 2 (Threat type: socio-cultural vs. economic) x 2 (Framing: present vs. absent) between-subjects design. In Experiment 1, socio-cultural threats elicited stronger extremist intentions than economic threats (B = .21, p <.001). This difference did not replicate in Experiment 2. In Experiment 3 we found that socio-cultural threats heightened extremist intentions only among left-wing participants (B = .10, p <.009). Crucially, in Experiment 4 we found out that the differences from Study 1 replicated and socio-cultural threats led to higher extremism than economic threats in general, through topic moralisation and an increased moral obligation to act upon the threat, supporting a full serial mediation (ab = 0.074, p = .002, SE = 0.023, 95% CI [0.032, 0.125]) and thus identifying a new explanatory mechanism. In sum, although not consistently, socio-cultural threats seem to be more radicalising than economic threats. This difference is accounted by moral mechanisms, such as moralisation and moral obligation to confront the threat.

Introduction. Irritability, defined as proneness to anger that may reach an impairing extent, is associated with a wide range of emotional and behavioral outcomes. However, challenges remain regarding its conceptualization and measurement. Discrepancies between caregiver and youth self-report in irritability ratings represent a stable phenomenon, widely observed across constructs in developmental psychopathology, that contributes to increased complexity in assessment. These discrepancies are systematically associated with factors such as diagnosis, age, and symptom severity. Moreover, missing measurement invariance suggests that caregiver and youth ratings do not necessarily reflect the same underlying construct but rather differ in their sensitivity to specific aspects of irritability. Youth may provide more accurate reports of tonic irritability (persistently angry or irritable mood), given their direct access to internal emotional states. In contrast, parents may be more sensitive to phasic irritability (short-lived temper outbursts), as they primarily observe its behavioral manifestations.

Objectives. This review aims to (1) quantify the magnitude and direction of discrepancies between youth and caregiver reports of irritability across instruments, and (2) examine whether these discrepancies differ between tonic and phasic irritability.

Methods. A systematic literature search will be conducted in PsycINFO, PubMed, Web of Science, and Scopus. Studies including both caregiver and youth self-report of irritability in samples aged 5–18 years will be included. Effect sizes reflecting directional discrepancies will be calculated as standardized mean differences and synthesized using random-effects models. Irritability items will be theoretically classified as reflecting tonic or phasic irritability to derive component-specific discrepancy estimates.

Results. We are currently in the process of preregistering the study and expect to present results at the time of the conference.

Conclusion. This study aims to clarify how and why reports of irritability differ across informants and assessment approaches, providing a foundation for refinement of measurement strategies and for more consistent guidelines for interpreting and integrating multi-informant data, ultimately contributing to more precise measurement of irritability.

Contamination-related obsessive-compulsive disorder (C-OCD) is characterized by contamination fears and associated excessive cleaning and washing behaviours that are elicited by a wide range of triggers. Exposure with Response Prevention (ERP) is the first-line psychological treatment for OCD and is effective for many patients. However, treatment response is heterogeneous, and symptoms may return after initially successful symptom reduction; potentially reflecting limited generalization of treatment effects across triggers and contexts. A better understanding of how such triggers are mentally represented may therefore provide new insights into processes relevant to exposure, particularly the generalizations of treatment gains.

The present study investigates the mental representations of contamination-related triggers using a similarity-judgement approach. We employ an empirically developed stimulus set comprising 300 images specifically selected to represent contamination-related triggers. Using the Spatial Arrangement Method, 325 participants provide a total of 44,850 pairwise similarity judgments across these stimuli. The resulting similarity structure will be analysed using multidimensional scaling to identify its underlying dimensions.

By mapping this psychological similarity space, this study aims to advance our understanding of how contamination cues are organized and to provide a basis for investigating how this knowledge may ultimately inform the optimization of ERP.