The dynamic basis of cognition: an integrative core under the control of the ascending neuromodulatory system

Abstract

The human brain integrates diverse cognitive processes into a coherent whole, shifting fluidly as a function of changing environmental demands. Despite recent progress, the neurobiological mechanisms responsible for this dynamic system-level integration remain poorly understood. Here, we used multi-task fMRI data from the Human Connectome Project to examine the spatiotemporal architecture of cognition in the human brain. By investigating the spatial, dynamic and molecular signatures of system-wide neural activity across a range of cognitive tasks, we show that large-scale neuronal activity converges onto a low dimensional manifold that facilitates the dynamic execution of diverse task states. Flow within this attractor space is associated with dissociable cognitive functions, and with unique patterns of network-level topology and information processing complexity. The axes of the low-dimensional neurocognitive architecture align with regional differences in the density of neuromodulatory receptors, which in turn relate to distinct signatures of network controllability estimated from the structural connectome. These results advance our understanding of functional brain organization by emphasizing the interface between low dimensional neural activity, network topology, neuromodulatory systems and cognitive function.One Sentence SummaryA diverse set of neuromodulators facilitates the formation of a dynamic, low-dimensional integrative core in the brain that is recruited by diverse cognitive demands

Context-dependent architecture of brain state dynamics is explained by white matter connectivity and theories of network control

“…We acknowledge that a limitation of this study was a focus on discrete brain states with common spatial activity patterns across subjects rather than continu-ously fluctuating, overlapping functional modes of brain activity 9,80 . Nevertheless, this simplified approach also constituted a major strength of the study, because it allowed us to assess developmental trajectories and cognitive effects of previously unexplored brain dynamics across subjects in a large sample.…”

Mind-to-mind heteroclinic coordination: Model of sequential episodic memory initiation

“…• Analysis of high temporal resolution human fMRI data from a large sample of unrelated individuals in the study of Shine et al 8 suggests that the "integrative core of brain regions ... manipulates the lowdimensional architecture of the brain across an attractor landscape via highly conserved modulatory neurotransmitter systems"; reconstruction of statespace trajectories unambiguously confirms "existence of a low-dimensional, dynamic, integrated component that recurs across multiple unique tasks and demarcates a common cognitive architecture within the human brain". The authors of the study summarize: "Global brain states exist along a low dimensional manifold" 8 .…”

“…Since the coding occurs on the population level, dynamics of the modes is usually lowdimensional 6 . Low-dimensionality results from coherent activity of many elements that form modes, and can be extracted from the records by application of e.g., principal component analysis 7,8 . We assume that N different spatio-temporal patterns (brain modes) P i ( r, t), i = 1, 2, .…”

“…2a. Hence, it follows that the system (7,8) has an invariant set T 0 that is the direct product of Γ andΓ: T 0 = Γ ×Γ. Since both Γ andΓ are homeomorphic to the circle, T 0 is homeomorphic to the two-dimensional torus T 2 .…”

“…Since both Γ andΓ are homeomorphic to the circle, T 0 is homeomorphic to the two-dimensional torus T 2 . The equilibrium points belonging to T 0 are O ij = O i ×Õ j ; the eigenvalues of the linearized at these points system (7,8) are summarized in the Table I.…”