The human cognitive system is severely capacity limited. As such, only a select subset of visual input to the retina is fully processed by the cognitive system at any one point in time. To compensate, mechanisms of visual attention allow us to distribute these limited resources throughout the visual field according to properties of the visual input and our goals and intentions. Establishing the exact criteria by which visual input is selected to undergo further processing is central to understanding the processes that determine visual selection; processes that ultimately contribute to our conscious experience of the world. In this thesis, I focus on stimulus novelty and the role that task-expectancies play in determining visual selection. It is well documented that that which is new and unexpected often appears to standout in the environment and can attract visual attention (Horstmann, 2005; Meyer, Niepel, Rudolph & Schűtzwohl, 1991;Ranganath & Rainer, 2003;Yantis & Jonides 1990). However, the conditions under which this is the case and the mechanism by which we come to be sensitive to novel stimuli require some elucidation.Here I present 3 empirical chapters that explore the conditions under which novel and unexpected stimuli attract visual attention. In Chapter 2, I investigated whether attentional capture by unexpected stimuli reflects a form of stimulus driven capture and occurs contrary to the immediate goals on an observer. Across a series of experiments I found that an infrequent (~3-6% of trials) and irrelevant motion cue captured attention when observers had an active set for a specific target color.Critically, this effect was not observed when the same motion stimulus was presented frequently. Thus, task goals appear to modulate capture by stimuli that broadly conform to contextual expectations, while stimuli that violate these expectations appear to guide visual attention involuntarily.In chapter 3, I examined what it means for a stimulus to be 'unexpected'. The first presentation of an unexpected stimulus during visual search has been associated with large RT cost. In chapter 3, I independently varied prior exposure to an unexpected and irrelevant motion distractor and explicit knowledge of its occurrence to assess the contributions of each of these factors in determining the unexpectedness of the stimulus. Neither prior exposure, nor knowledge of occurrence served to attenuate the response to the unexpected motion distractor.ii These results suggest that the task expectations, which deviant stimuli are evaluated against, are highly context specific and derived from a process of implicit learning about stimulus relevance, and are a not shaped by explicit top-down knowledge about the likelihood of occurrence.In Chapter 4, I examined the time-course of attention shifts to unexpected stimuli. Previous authors (e.g., Horstmann, 2005;2006) have made the claim that shifts of attention to unexpected stimuli are delayed relative to expected stimuli and that this may suggest a dedicated orienting sy...