The most important operations of the brain, from vision and audition to complex movements and decision making, necessitate coordinated activity in populations of neurons. Within an area of neocortex, nearby neurons form complex networks of connections that help shape the activity patterns we observe. Since neurons are inextricably embedded within their local population, we must consider neural populations as a whole if we are to understand how the information in the outside world is interpreted and represented in the brain. In this thesis, I present two complementary approaches to study how groups of neurons cooperatively generate patterns of activity. Together, the results demonstrate how trial-to-trial variability of population activity can be explained by functional relationships within groups of neurons. The first approach identifies small ensembles of neurons that reliably spike in sequence to investigate spike propagation in emergent, spontaneous activity. By resolving spatiotemporal spike patterns into sequences, I describe single trial variability and substantiate evidence for neocortical assembly phase sequences. Secondly, I construct a functional graph from correlated activity between pairs of neurons to holistically represent shared variability in visually-evoked population dynamics. These graphs displayed a specific pattern of functional connections underlying accurate predictions of trial-to-trial variability, illustrating a signature of informative correlations in neocortical networks. These analyses delineate the ways in which ensembles of neurons coordinate their activity to shape population dynamics. Extending models of neural activity from single cells to the networks they form have been difficult. By strengthening the lexicon used to capture variability and describe population activity on single trials, we can better investigate the biological sources and behavioral consequences of variability.