The aim of this dissertation is to explore the link between long-term attention and memory by looking at two types of data, Hollywood film and reaction time data. Data from a variety of sources are analyzed using a sequence of techniques designed to detect patterns in the data related to dimensionality. These patterns have been linked in to long-scale fluctuations in human attention, amongst other natural phenomena, but there is much research to be done before concluding whether long-term temporal structure plays a meaningful role in human attention. This question is investigated through four main chapters: (1) An introduction to the concept of timeseries analysis, fractal structure, and how these phenomena may be linked to neural systems; (2) An analysis of the structure of shot lengths in Hollywood film, with several techniques simulated and compared; (3) A review and analysis of historical reaction time data collected before the advent of film; (4) A series of novel laboratory experiments which investigate the link between attention and memory for visual stimuli presented in different temporal structure.