THINGS-data

THINGS-data provides a multimodal dataset integrating functional magnetic resonance imaging (fMRI), magnetoencephalographic (MEG) recordings, and 4.70 million behavioral similarity judgments to study natural object representations in the human brain.

Key Features:

• Multimodal recordings: Includes functional magnetic resonance imaging (fMRI) and magnetoencephalographic (MEG) data collected to capture spatial and temporal aspects of neural responses.
• Large-scale behavioral data: Contains 4.70 million similarity judgments measuring perceived relationships among stimuli.
• Extensive stimulus set: Uses thousands of photographic images representing up to 1,854 distinct object concepts.
• Rich annotations: Provides accompanying annotations for each object concept to support varied hypothesis testing.
• Dense sampling of the visual world: Offers comprehensive coverage of object categories to enable in-depth exploration of object representation.
• Dataset integration: Multimodal structure supports integration across neuroimaging and behavioral measurements.
• Validation analyses: Includes analyses demonstrating the dataset's reliability and utility for hypothesis-driven and data-driven investigations.
• Illustrative examples: Presents five example analyses showcasing diverse potential uses, including neural and behavioral investigations.

Scientific Applications:

• Neural correlates of object recognition: Enables investigation of brain regions and patterns associated with object recognition using fMRI and MEG.
• Temporal dynamics of object processing: Supports analysis of temporal aspects of object representation using MEG recordings.
• Brain–behavior mapping: Facilitates linking neural responses to behavioral similarity judgments to study perceptual and cognitive representations.
• Large-scale hypothesis testing and reproducibility: Allows testing hypotheses across hundreds to thousands of object concepts and evaluating reproducibility of findings.
• Cross-dataset and interdisciplinary research: Permits integration of modalities to address questions spanning cognitive neuroscience and behavioral science.

Methodology:

Data comprise functional magnetic resonance imaging (fMRI) and magnetoencephalographic (MEG) recordings together with 4.70 million behavioral similarity judgments collected in response to thousands of photographic images representing up to 1,854 distinct object concepts.