The first part of the project was dedicated to build a whole-brain computer model that can reproduce different dynamics seen at different levels of consciousness ranging from awake to sleep and anesthetized states. We tried to explain experimental findings showing the brain's ability to produce different dynamical states characterized by changing levels of synchronized brain activity and how the anatomical wiring of the brain affects this synchronization. Importantly, different levels of consciousness were delineated by the specific composition of such dynamical states with showing a large repertoire of states during waking and a reduced repertoire during loss of consciousness. To explain these results we constructed a whole-brain model in which the activity of each brain area was described by mathematical equations following the Hopf formalism and regions interacted through a pattern of connections that was taken from empirical data. Similar to the experimental findings, this model was able to change its patterns of global synchronization and expression of anatomical information as a function of the bifurcation parameter, the principal control switch in the model that characterizes a transition between noisy and oscillatory dynamics in each brain areas. When this parameter was adjusted to match the empirically found repertoire of dynamical states (see Figure 1), the model was found to reproduce unconscious brain activity during different stages of sleep (N1-N3), when set to a dynamical working point with dominating noisy local activity. In contrast, awake states were best matched with a value of the bifurcation parameter, where local dynamics randomly switches between noisy and oscillatory dynamics. The result of these random switches was a complex pattern of global synchronization and desynchronziation with varying degrees of dynamically expressed anatomical connectivity that was similar to activity in the awake state. These changes in the model parameter can be biologically interpreted as altering the level of global excitability in the brain, possible mediated by different neuromodulatory systems, which tune its ability to create complex patterns and awareness. Parts of this project have been published in a scientific journal, presented at an international conference and another manuscript is in preparation.
In the second part we implemented the global workspace model of consciousness by incorporating oscillatory signals to account for the oscillatory properties of neuronal information processing. We first developed a novel theoretical framework that is based on the theory of nonlinear dynamical systems, and formalizes how two brain areas communicate with each other. This theory assigns a vital role to resonance mediated by neuronal oscillations for routing signals between two areas and identifies varying roles for different frequency bands in the communication process. The framework distinguishes between fast, non-oscillatory, routing from slower communication based on fast oscillations and describes how attention and plasticity can modify the speed of efficacy of communication. These ideas were then implemented in an anatomical architecture mimicking the global workspace model of consciousness and using the Wilson-Cowan model of neuronal activity. In this model, fast oscillatory signals were routed along two separate pathways towards global workspace areas, where they were maintained through recurrent neuronal activity. The signal first entering the global workspace enhanced slower frequencies in the competing pathways which had a blocking effect on routing of fast oscillatory signals and prevented its entrance into the global workspace. Moreover, the winning stimulus sent back rhythmic signals that enhanced and accelerated its routing to the global workspace reflecting the impact of attention. Manuscripts are in preparation and work on the theoretical framework is ready to be submitted to a high impact journal.
