Our mind and the brain are fascinatingly rich and complex, constantly changing over time in response to the situations we are in. This thesis asks how large-scale neural dynamics give rise to higher-order cognitive processes that we experience in the real world, including event perception and comprehension, attentional engagement, and episodic memory. Moving beyond a focus on local brain areas, the work investigates how activity and interaction of the whole brain areas and networks reconfigure in relation to cognitive dynamics. Part 1 introduces three empirical chapters that use functional magnetic resonance imaging with naturalistic behavioral experiments in humans. Results demonstrate that attentional engagement is predicted by dynamic patterns of whole-brain functional connectivity, brain-wide state dynamics in a low-dimensional manifold reflect cognitive and attentional state dynamics, and the process of narrative comprehension guided by memory integration is characterized by changes in neural representation patterns. Part 2 introduces three theoretical chapters that review important literature and elaborate my opinion on how the field should study the brain to understand the mind and behavior. The work argues for the need to characterize neural dynamics in relation to cognitive dynamics, test for generalizability of brain-behavior relationship across contexts, use naturalistic and complex behavioral paradigms to complement targeted experiments, probe neural processes at multiple spatiotemporal scales and dimensionalities, and integrate findings across neuroscience subfields that use different model species and techniques. Together, this thesis demonstrates that the mechanisms of complex cognitive experiences in humans can be understood by looking at the biological and computational operations of the large-scale neural system.