Accepted Proposals
Individual Participant Data Meta-Analysis of Anxiety Traits and Fear Learning: An Extension of Aggregated Meta-Analytic Findings
Individual differences, particularly anxiety-related traits, are central to understanding variability in fear conditioning responses. Numerous studies—including meta-analyses—have examined these associations, but inconsistencies in trait selection and outcome measures complicate synthesis and generalizability of findings. Therefore, there is a need for a comprehensive meta-analysis considering diverse anxiety-related traits (e.g. STAI, IUS, NEO) and outcome measures (e.g. SCR, FPS, ratings) to identify shared and distinct patterns, as well as potential moderators. Building on Bruntsch et al. (2024), who conducted an aggregated data meta-analysis (AD-MA) on this topic, the present study extends this work using an individual participant data meta-analysis (IPD-MA). The IPD-MA will leverage the fearbase dataset, to mitigate publication bias and potentially yield more accurate effect estimates. Additionally, IPD-MA facilitates more accurate moderator analyses by using individual participant data rather than aggregated values, thus enhancing statistical power and specificity. Where possible, we will directly compare results from AD-MA and IPD-MA to assess the impact of publication bias and methodological differences. Further, we are aiming to clarify the relationship between anxiety-related traits and fear conditioning measures, including analyses at the questionnaire item level. By utilizing item-level data from various anxiety questionnaires, we aim to identify specific predictors and latent factors underlying the association between anxiety-related traits and fear conditioning outcomes.
Maria Bruntsch* , Tina Lonsdorf , Mana Ehlers , Fritz Becker , Annalena Witte , Conrad Alting
Cross-Measure Comparisons in Human Fear Conditioning
Defensive responses induced in fear conditioning paradigms are assessed using a variety of outcome measures, including subjective ratings of valence, arousal, fear, or US expectancy and physiological measures such as skin conductance response, heart rate, or defensive reflexes such as fear-potentiated startle. Often, these measures are treated as broadly equivalent in both primary and meta-analytic work. In meta-analyses, effects for different outcome measures are typically pooled within outcome types (e.g. all ratings combined) or across all outcomes. However, there is evidence that different measures capture different sub-processes and temporal components of the defensive response. To date, it has not been comprehensively investigated to what extent outcome measures converge. This limits the interpretability and comparability of findings, and poses a challenge for meta-analytical approaches that rely on aggregating results from multiple studies. In this project, fearbase data will be used to systematically compare the most prevalent outcome measures in fear conditioning research. The analyses aim to quantify the correspondence between different measures over time and examine whether measures cluster into distinct groups. Findings from this project are intended to facilitate both a more targeted selection of measures prior to data collection and a more differentiated interpretation of experimental results. Moreover, they aim to inform the development of suitable procedures for cross-study data aggregation with different outcome measures.
Annalena Witte* , Fritz Becker , Maria Bruntsch , Conrad Alting , Mana Ehlers , Tina Lonsdorf
Temporal Dynamics in Human Fear Conditioning
The temporal dynamics of fear acquisition and extinction are underresearched, despite their key relevance to learning processes. In most studies, fear responses are averaged across subsets of trials (e.g., early or late extinction trials) or entire experimental phases to simplify statistical analyses. Although this practice helps with model convergence and requires fewer participants, it implies the assumption of equal strength of stimulus association across aggregated trials. This, however, obscures the dynamics inherent to learning processes and limits theoretical and explanatory insight into conditioning mechanisms. This project aims to address this gap by leveraging the fearbase’s pool of trial-level, individual participant data. The primary analytical approach will employ mixed-effects models to assess how defensive responses manifest and extinct over time, across individuals, and across studies. Additionally, the project will investigate the feasibility of cognitive modeling approaches given the constraints that the precise trial sequence of each participant and the exact reinforcement protocol are typically not included in publicly shared data in the field, and hence are not yet included in the fearbase. The project aims to utilize mixed models and cognitively motivated modeling strategies to understand the temporal dynamics of fear learning. Overall, the project’s goals are to clarify the information loss associated with temporal averaging, demonstrate the value of open, individual participant data for fear conditioning research, and provide methodological guidance for future studies seeking to model learning dynamics under imperfect data conditions.
Fritz Becker* , Tina Lonsdorf , Maria Bruntsch , Annalena Witte , Mana Ehlers