Abstract
Large‐scale brain dynamics are believed to lie in a latent, low‐dimensional space. Typically, the embeddings of brain scans are derived independently from different cognitive tasks or resting‐state data, ignoring a potentially large—and shared—portion of this space. Here, we establish that a shared, robust, and interpretable low‐dimensional space of brain dynamics can be recovered from a rich repertoire of task‐based functional magnetic resonance imaging (fMRI) data. This occurs when relying on nonlinear approaches as opposed to traditional linear methods. The embedding maintains proper temporal progression of the tasks, revealing brain states and the dynamics of network integration. We demonstrate that resting‐state data embeds fully onto the same task embedding, indicating similar brain states are present in both task and resting‐state data. Our findings suggest analysis of fMRI data from multiple cognitive tasks in a low‐dimensional space is possible and desirable.
Highlights
Understanding large-scale brain dynamics is a major goal of modern neuroscience (Jorgenson et al, 2015)
We embedded resting-state data into the same task embedding to investigate differences in brain dynamics between resting-state and task performance. These results suggest that manifold learning can uncover an interpretable low-dimensional embedding for the study of brain dynamics in functional magnetic resonance imaging (fMRI) data
Using a recently validated manifold learning framework, named 2-step Diffusion Maps—2-step diffusion maps (2sDM) (Gao et al, 2019), we demonstrate that fMRI data from different tasks span the same low-dimensional embedding
Summary
Understanding large-scale brain dynamics is a major goal of modern neuroscience (Jorgenson et al, 2015). There is growing evidence to suggest that a low-dimensional space—hidden from direct observation, learned from the data, and derived from many brain regions—may be a suitable model for studying temporal brain dynamics (Gao & Ganguli, 2015). These low-dimensional spaces have been observed using a variety of neural recordings and animal models (Ahrens et al, 2012; Churchland et al, 2012; Kobak et al, 2016; Mishne et al, 2016; Santhanam et al, 2009). Data from richer tasks often project onto a larger portion of the manifold, violating linear approximations
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
