Cognitive information transformation in functional brain networks

The human brain is a flexible information processing system. Across a range of simple and complex tasks, such as walking across the street to playing basketball, the brain transforms sensory information from the environment into corresponding motor actions. This sensory input to motor output transformation likely requires a sequence of complex neural computations implemented by brain networks. Though decades of cognitive neuroscience have made great progress in characterizing the functions of individual brain areas, less progress has been made in understanding exactly how these brain regions work in concert to implement the diverse cognitive computations underlying complex behaviors. In this thesis, I provide an account of how the brain's distributed functional networks implement neurocognitive functions and computations. First, I demonstrate how local cognitive task activations can be computed from the activity of other brain areas through distributed brain network connectivity patterns. This illustrates how intrinsic functional connectivity enables the transfer of task-relevant activations between brain regions. Second, I demonstrate how local cognitive information, such as sensory stimulus activations in sensory cortices, is transformed into motor activations in motor cortex through a sequence of computations governed by intrinsic functional connectivity during cognitive tasks in both humans and non-human primates. This demonstrates that the intrinsic brain network organization can provide insight into how the brain implements neurocognitive computations and transformations. Finally, I investigate the relationship between task activations and functional network connectivity from a dynamical systems perspective. Specifically, I demonstrate that task-state activity quenches ongoing functional correlations and variability, and that this quenching occurs due to a sigmoidal transfer function that describes local mean-field neural activations. This suggests that task-state functional network changes are meaningful, and reflect nonlinear relationships between brain regions. This provides a way forward to improve current models of neural computations and communication by leveraging nonlinear models of neural dynamics. Together, the results presented in this thesis provide a novel understanding of how functional brain network organization shapes cognitive computations.Ph.D.Includes bibliographical reference