Abstract
This dataset provides a comprehensive resource for exploring the relationship between the brain's intrinsic functional architecture and a wide array of human traits. By combining high-resolution multimodal MRI with an extensive battery of 31 questionnaires and 7 behavioral tasks, the researchers offer a "deep phenotyping" approach to 194 healthy participants. This resource, part of the MPI-Leipzig Mind-Brain-Body database, aims to address the reproducibility crisis in human connectomics by providing high-quality, open-access data.
Introduction
The field of human functional connectomics seeks to map the communication patterns between brain regions and understand how these patterns underpin individual differences in behavior and personality. However, progress has been hampered by the "reproducibility crisis," often stemming from small sample sizes or limited behavioral characterization. To move toward precise "phenotyping"—identifying neural signatures of specific traits—the community requires datasets that pair rigorous neuroimaging with exhaustive psychological measures. This paper introduces a significant contribution to the MPI-Leipzig Mind-Brain-Body database designed to meet this need.
Methods and Participant Demographics
The study recruited 194 native German-speaking participants, ranging in age from 20 to 75 years. The cohort was carefully selected to ensure a broad representation of the adult lifespan, providing a valuable cross-sectional look at brain-behavior relationships.
Data acquisition followed a rigorous 5-day protocol. The neuroimaging component, conducted at a 3-Tesla MRI facility, included a high-resolution structural scan using a 3D MP2RAGE sequence, which offers superior gray/white matter contrast compared to standard T1-weighted scans. This was followed by four 15.5-minute resting-state functional MRI (rs-fMRI) runs using a multiband EPI sequence, totaling approximately one hour of functional data per subject.
Study Design and Measurement
The dataset is characterized by its multimodal depth, covering various design and measurement types to capture a holistic view of the human phenotype.
The behavioral assessment spanned five primary domains:
- Personality & Habitual Behaviors: Including the "Big Five" personality traits, the "Dark Triad" (narcissism, Machiavellianism, psychopathy), and habits such as mobile phone and Facebook usage.
- Mind-Wandering & Mindfulness: This domain is a unique strength of the dataset, capturing "state" measures through probes administered during the resting-state scans to record participants' thoughts in real-time.
- Cognitive Control & Attention: Evaluated through tasks like the Conjunctive Continuous Performance Task (CCPT).
- Creativity: Assessing both divergent thinking and real-world creative achievement.
- Special Traits: Exploring phenomena like synesthesia and involuntary musical imagery (earworms).
Technical Validation and Data Quality
Technical validation confirmed the high quality of the dataset. Approximately 89.2% of fMRI runs exhibited minimal head motion, with a maximum framewise displacement of less than one voxel (2.3 mm). Temporal signal-to-noise (tSNR) maps showed robust signal across the majority of the cortex, despite standard susceptibility artifacts in orbitofrontal regions. Furthermore, the psychological measures demonstrated strong internal consistency, with an average Cronbach’s alpha of 0.78 for the translated German questionnaires, confirming the reliability of the behavioral data.
Open Science and Future Directions
The authors have demonstrated a strong commitment to open science by making the raw and preprocessed data available via platforms like OpenNeuro and the Harvard Dataverse. The data follows the BIDS (Brain Imaging Data Structure) standard, ensuring compatibility with modern analysis pipelines.
While the sample size is smaller than "big data" initiatives like the UK Biobank, the sheer depth of phenotyping provided here offers a unique goldmine for researchers interested in the nuanced links between brain architecture and the complexities of human experience. Future work may leverage factor-analytic approaches to distill stable behavioral scores from the extensive array of overlapping measures provided in this rich dataset.
Summary updated