Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - CONSCBRAIN (How consciousness is shaped by neuronal network dynamics)

Teaser

This project aimed at developing computer models that could elucidate the physiological mechanisms of conscious experience in the brain. We all experience being awake and alert during most of the day, interrupted only by occasional episodes of fatigue, and fall asleep at night...

Summary

This project aimed at developing computer models that could elucidate the physiological mechanisms of conscious experience in the brain. We all experience being awake and alert during most of the day, interrupted only by occasional episodes of fatigue, and fall asleep at night with consciousness either fading entirely or reappearing in a different shape during dreaming. Yet, even when we are fully awake , our brains make us aware of only a fraction of all inputs from the environment, while most of our surroundings are processed unconsciously. Advanced brainimaging tools such as fMRI, EEG and MEG have opened a window on observing the neuronal signature of consciousness and have tried to untangle the mechanisms behind conscious and unconscious brain processes. This research also promises to find neurophysiological markers that can guide clinicians in correctly diagnosing and effectively treating neuropsychiatric disorders. Experimental findings inspired theoretical accounts of consciousness with one theory stating that awareness rises in any physical system capable of creating activity patterns which are both integrated and differentiated at the same time. Another framework states that stimuli compete in sensory areas for entering a global workspace situated in the front of the brain, where recurrent activity makes us becoming aware of the winning stimulus, while input processed outside the global workspace remains unconscious. The first objective of this project was to create whole-brain computer models that show dynamics similar to the different stages of consciousness seen during the sleep-wake cycle and find a computational marker for consciousness. In the second part of the project we tried to implement a computer model of the global workspace idea compatible with the rhythmic nature of brain activity to better understand the physiology behind access to consciousness.

Work performed

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.

Final results

The results obtained in this project advance our understanding of different levels of consciousness. They provide a mechanistic account of how the experimentally found changes in the interactions between anatomy and brain activity are generated by adjusting a single model parameter. For instance by fitting the model to individual patients, this parameter could potentially serve as a marker to determine the state of consciousness, as well as distinguish between vegetative state patients and those with minimal traces of consciousness. The theoretical framework about neuronal communication is expected to have a significant impact on our understanding of brain function and consciousness. It combines previously disjoint concepts comprising different modes of communication. It also gives a mechanistic explanation of the necessity for oscillations in different frequency bands, to establish or disrupt communication between brain areas. It is hypothesized that the creation of contrast between neuronal populations with firmly established communication, and those where communication was prevented through inhibition, is at the core of granting stimuli access to consciousness within a global workspace. We showed that this framework can indeed be applied to the global workspace architecture, and implemented in physiologically realistic computer models. Finally, the theory and model architecture might provide a fruitful avenue for understanding the mechanisms behind neuropsychiatric diseases that have been conceptualized as disorders of brain communication.