Dynamic reconfiguration of macaque brain networks during free-viewing of natural scenes

Natural vision involves the activation of a wide range of higher-level regions processing objects, motion, faces and actions. Here, we pursue a data-driven approach to explore how higher-level visual processes relate to the underlying structural and functional connectivity. Using a free-viewing paradigm in four awake rhesus macaque monkeys, we investigate how different visual scenes change functional connectivity. Additionally, we explore how such functional connectivity, as measured through fMRI, is related to the structural connectivity, as measured through diffusion weighted imaging. At first, we evaluate the consistency of the elicited free-viewing pattern using standard analytical techniques. We also evaluate the underlying structural connectivity via diffusion data by tracking white matter bundle projections from the visual cortex. We then reconstruct free-viewing and structural networks and quantify their properties. Centrality measures over the entire fMRI time-series revealed a consistent functional network engaged during free-viewing that included widespread hub regions across frontal (FEF, 46v), parietal (LIP, Tpt), and occipitotemporal cortex (MT, V4 and TE) among others. Interestingly, a small number of highly-weighted and long-length inter-hemispheric connections indicated the presence of long-range integrative properties during free-viewing. We hypothesized that during free-viewing, networks had the capacity to change their local and distal connections depending on the on-going changes in visual scenes. To capture these network dynamics, we depart from the static modular architecture of the structural networks and demonstrate that hubs in free-viewing networks reorganize according to the presence of objects, motion, and faces in the movie scenes indicating poly-functional properties. Lastly, we compare each NHP subject network and observe high consistency between individuals across the same network type with closer correspondence between structural networks (e.g., diffusion based and those partially assembled from tract-tracing). In summary, our network analyses revealed ongoing changes in large-scale functional organization present during free-viewing in the macaque monkey and highlight the advantages of multi-contrast imaging in awake monkeys for investigating dynamical processes in visual cognition. To further promote the use of naturalistic free-viewing paradigms and increase the development of macaque neuroimaging resources, we share our datasets in the PRIME-DE consortium.