The flexibility and robustness of social insect colonies, when they cope with challenges as integrated units, raise many questions, such as how hundreds and thousands of individual local responses are coordinated without a central controlling process. Answering such questions requires: 1. Quantifiable collective responses of colonies under specific scenarios; 2. Decomposability of the collective colony-level response into individual responses; and 3. Mechanisms to integrate the colony- and individual-level responses. In the first part of my dissertation, I explore coordinated collective responses of colonies in during the alarm response to an alarmed nestmate (chapter 2&3). I develop a machine-learning approach to quantitatively estimate the collective and individual alarm response (chapter 2). Using this methodology, I demonstrate that colony alarm responses to the introduction of alarmed nestmates can be decomposed into immediately cascading, followed by variable dampening processes. Each of those processes are found to be modulated by variation in individual alarm responsiveness, as measured by alarm response threshold and persistence of alarm behavior. This variation is modulated in turn by environmental context, in particular with task-related social context (chapter 3). In the second part of my dissertation, I examine the mechanisms responsible for colonial changes in metabolic rate during ontogeny. Prior studies have found that larger ant colonies (as for larger organisms) have lower mass-specific metabolic rates, but the mechanisms remain unclear. In a 3.5-year study on 25 colonies, metabolic rates of colonies and colony components were measured during ontogeny (chapter 4). The scaling of metabolic rate during ontogeny was fit better by segmented regression or quadratic regression models than simple linear regression models, showing that colonies do not follow a universal power-law of metabolism during the ontogenetic development. Furthermore, I showed that the scaling of colonial metabolic rates can be primarily explained by changes in the ratio of brood to adult workers, which nonlinearly affects colonial metabolic rates. At high ratios of brood to workers, colony metabolic rates are low because the metabolic rate of larvae and pupae are much lower than adult workers. However, the high colony metabolic rates were observed in colonies with moderate brood: adult ratios, because higher ratios cause adult workers to be more active and have higher metabolic rates, presumably due to the extra work required to feed more brood.