Complex dynamics of cognitive neural networks within normal and pathological brain states

Abstract

The classical view in systems neuroscience is that individual regions of the brain are specialised for the execution of specific tasks. This position is supported by a long history of lesion studies and more recently by a wealth of functional neuroimaging research, which has provided insights into how the anatomy of the brain relates to behaviour. However, in recent years this classical view has changed. Instead, new techniques have revealed intrinsic functional connectivity networks (ICNs) that reflect underlying patterns of structural connectivity, now considered to be the fundamental neural architecture supporting cognition and behaviour (Adams, 1982, Smith et al. 2009b, Honey et al. 2009). The activity of these ICNs is dependent on intrinsic fluctuations in their activity and the behavioural context, which dynamically reconfigure over time (Fox et al. 2005, Ances et al. 2009). Therefore, investigation of brain networks needs to consider not only the structural connections that constrain functional interactions, but also dynamic changes in functional interactions. An ongoing challenge, therefore, is to define a framework incorporating these neural dynamics that combines insights from structure, function and behaviour. Dynamical systems theory may provide a flexible framework that combines all of these levels. Current theory proposes that healthy neural dynamics operate in a multistable regime that promotes flexible information processing and behaviour (Kelso 2012, Tognoli and Kelso 2014, Shanahan 2012, 2010b, a, Friston 1997, Leech and Sharp 2014, Irner 2007). Large-scale multistable dynamics are likely constrained by underlying structural connections between brain regions. (Honey et al. 2009, Deco, Jirsa, McIntosh, Sporns, and Kötter 2009, Deco, Jirsa, and McIntosh 2011, Deco et al. 2008, Cabral et al. 2011, Cabral et al. 2012). However, it is unclear how such multistable dynamics may relate to behaviour, or how they are constrained by network structure. Moreover, whilst such dynamical accounts place considerable importance on the structural connectivity of the brain, it is unclear how the multistable dynamics are altered when pathological processes result in structural disconnections. In this thesis, I explore such dynamical accounts of the brain using a range of neuroimaging and computational approaches, and examine the implications of this level of description in one example of structural disconnection, namely the consequences of traumatic brain injury.