When viewing a complex event, it is necessary to identify and calculate the relationships between different entities in the event. For example, when viewing a caused motion event (e.g. a man raking leaves into a basket.), people need to identify the Agent (man), the affected object or Patient (leaves), the Instrument (rake) and the Goal (basket). In this paper we explore how this process of event apprehension proceeds using eye-tracking methodology. Our study indicates that viewers extract event components rapidly, but some components can be extracted faster than others. Moreover, there is a structure to saccade patterns when participants are asked to identify specific event components. In caused motion events, attention is allocated to the Patient during the early stages of processing even when the Patient is not the target. We discuss implications of this work for how people perceive complex events.