Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

Arrow of Time: an intensive fMRI dataset of brain responses to brief video presentations

Shufan Zhang¹; ¹#NAME?

Presenter: Tomas Knapen

Extensive sampling of neural activity during rich cognitive phenomena is critical for robust understanding of brain function. To this end, we introduce the Arrow of Time dataset, a novel resource that captures high-quality functional magnetic resonance imaging (fMRI) responses to thousands of richly annotated short (2.5s) naturalistic videos. Our goal is to provide a deep open-by-design dataset of brain responses to the field. Participants are exposed to up to 2000 video stimuli while ensuring strict fixation, across up to 15 scanning sessions. The extensive and varied stimuli are annotated using neural network-based object segmentation, action recognition, and semantic descriptions. This allows researchers to investigate how the brain processes and represents complex visual information, such as object recognition, scene understanding, and semantic processing. The dataset is generated using whole-brain 7 Tesla fMRI optimized for SNR, acquired at 1.6s and an isotropic voxel resolution of 1.7mm. After custom preprocessing geared towards optimal across-session coregistration, beta weights are estimated for each single video presentation using GLMsingle which projects out sources of noise and estimates the shape of the HRF at the single-voxel level. Initial quality assessments show noise ceiling values on a par with those in NSD. This type of high-quality fMRI data allows for rigorous testing and refinement of models related to brain function and cognition. As a reusable resource, the Arrow of Time dataset will help elucidate the neural basis of naturalistic visual processing in the hands of researchers in the fields of cognitive neuroscience, psychology, and artificial intelligence.

Topic Area: Visual Processing & Computational Vision

Extended Abstract: Full Text PDF