Monitoring cognitive vigilance during attentive tasks, such as the psychomotor vigilance test (PVT), helps evaluate performance. EEG can be used to track the type of attentive control being used during task progression: top-down (TD; goal-driven) or bottom-up (BU; salient driven). This is contingent on the dichotomous view of attention (Gaspelin & Luck, 2018), which excludes a third proposed mechanism: selection bias (Awh et al., 2012). In this literature review, it is proposed that low frequency beta (β) waves are a strong index of TD control. BU control, however, lacks a clear EEG metric such as proposed gamma (𝛾) waves . Additionally, mission constraints prevent the ideal use of both spatial fMRI and temporal EEG. For such a metric, a follow up investigation measuring changes in β wave activity from baseline may provide an indirect metric of BU control. Furthermore, a general lack of understanding of BU control and its possible task subsets warrants exploration. Keywords: psychomotor vigilance task (PVT), electroencephalogram (EEG), top-down control, bottom-up control, beta (β) waves, gamma (𝛾) waves
Literature Review
The Mechanisms of AttentionThe conventional view of attention has been viewed as a dichotomous mechanism: top-down control or bottom-up control. Top-down control (TD) is simply defined as the goal-driven type of attention including both volition and stimuli context (i.e. trying to stay focused while reading in a noisy library). In contrast, bottom-up control (BU) can be simply defined as salient driven or distraction driven attention (i.e. shifting attention to a loud noise instead of reading the book) [1]. TD and BU control is also commonly associated with feedforward (BU) and feedback (TD) descriptions of neural activity [35]. While the terminology of BU control is generally agreed upon, the scope of TD control is debated and not universally agreed upon. Proposals of a third category of attention involve "selection bias": the reward history and selection history of a specific stimuli [2,3]. It was argued that context of the stimuli, while internal by nature, should not be classified as TD control since volition and context are vastly different [2]. The mechanisms of these two components (volition vs context) are inherently different: selection context "is fast, automatic and occurs without much, if any, effort" ([3], p. 29). It was thus proposed that most selection is conducted via selection bias, therefore reducing the frequency of use of TD control [3]. In contrast, the conventional and current AFRL view of TD control can be summarized as "'top-down' processing to refer to any perceptual phenomenon that is influenced by context, learning, or expectation, which would include selection history" ([1], p. 25). This is not to say that current AFRL cognitive models, such as ACT-R, cannot comport with "selection bias" as the mechanics of cognition are present. It is important to note these are terminology disagreements (no experimentation has taken place).