The path to make sense of the puzzling nature of human cognitive behaviour requires
an intimate understanding of the general systems in the brain and how they work
in concert to manifest such behaviour. As a multi-disciplinary approach, it is at the
forefront of adapting, integrating and applying recent developments in data science,
and machine learning to further our understanding of the many facets of the brain.
These advancements led to a paradigm shift perceiving the brain as a biological system
composed of multi-scaled intertwined communities of functional networks. This
perspective gained immense popularity in recent years providing new insight into
the brain fundamental organisation and the neuronal mechanisms associated with
cognitive behaviour.
However, while highly promising, the application of these new techniques to
cognitive neurosciences is still a challenge. It requires neuroscientists to be able to
reformat the accumulated knowledge in the field, its big questions, theories and prior
knowledge within the confines imposed by these approaches. It also calls for adapting,
challenging and changing the interrogative capabilities of these methods from their
applied nature. In this thesis, I use an advanced suite of machine-learning methods and
data science approaches to investigate how multivariate patterns of brain activity and
connectivity dynamically vary with manipulation of behaviourally distinct aspects of
cognition. Specifically, I demonstrate that inspection of dynamic macro connectomic
provides additional insight to transcend our understanding of intelligence and working
memory. The multivariate inspection provided here was able to validate classic localist
findings in the field and tie them to the modern network perspective. Additionally,
this thesis develops a framework and tools to apply data science methodologies to new
and exciting research questions within cognitive neuroscience.