UC Riverside

Dense, individual-level datasets are an important resource for social neuroscience that enable models of brain responses to a wide range of naturalistic features, often investigated with movie fMRI. This data release provides high spatiotemporal resolution 3 Tesla BOLD fMRI data in three healthy participants while viewing four feature-length movies with sound (Grand Budapest Hotel, Forrest Gump, Planet Earth, Jiro Dreams of Sushi; ca. 520 minutes) and a movie trailer composite, together with clip and movie repeats. All movie functional data were acquired with partial brain coverage from an approximately axial slab covering ventral prefrontal and temporal lobes, including the amygdala, with a repetition time of 556 ms and complex-valued image reconstruction. The dataset is released in both unprocessed and minimally preprocessed forms with individual, high quality anatomic templates. Preprocessed fMRI data are provided in both individual template and FreeSurfer average surface spaces. Reference raw data are provided for a conventional face-object functional localizer task. Auxiliary physiological data includes 4-lead ECG waveforms, image-estimated respiratory waveforms and pupillometry. Quality metrics include temporal SNFR maps, high-resolution B₀ maps, EPI dropout and head motion parameters for all runs. Physiological noise estimation and cleaning used spatial independent component analysis and custom, per-subject component classification. Annotations for the movies provide automated low-level audiovisual and facial features, as well as emotion ratings from each participant. The presence of presumed BOLD neurovascular responses not associated with physiological or instrumentation noise was confirmed using temporally concatenated spatial independent component analysis (tcsICA).