Andrew T. Duchowski - Eye Tracking Methodology

At the second half of the 19th century, Von Helmholtz (1925) posited visual attention as an essential mechanism of visual perception. In his Treatise on Physiological Optics , he notes, We let our eyes roam continually over the visual field, because that is the only way we can see as distinctly as possible all the individual parts of the field in turn. Noting that attention is concerned with a small region of space, Von Helmholtz observed visual attentions natural tendency to wander to new things. He also remarked that attention can be controlled by a conscious and voluntary effort, allowing attention to peripheral objects without making eye movements to that object. Von Helmholtz was mainly concerned with eye movements to spatial locations, or the where of visual attention. In essence, although visual attention can be consciously directed to peripheral objects, eye movements reflect the will to inspect these objects in fine detail. In this sense, eye movements provide evidence of overt visual attention.

In contrast to Von Helmholtzs ideas, James (1981) believed attention to be a more internally covert mechanism akin to imagination, anticipation, or in general, thought. James defined attention mainly in terms of the what, or the identity, meaning, or expectation associated with the focus of attention. James favored the active and voluntary aspects of attention although he also recognized its passive, reflexive, non-voluntary and effortless qualities.

Both views of attention, which are not mutually exclusive, bear significantly on contemporary concepts of visual attention. The what and where of attention roughly correspond to foveal (James) and parafoveal (Von Helmholtz) aspects of visual attention, respectively. This dichotomous view of vision is particularly relevant to a bottom-up or feature-driven explanation of visual attention. That is, when considering an image stimulus, we may consider certain regions in the image that will attract ones attention. These regions may initially be perceived parafoveally, in a sense requesting further detailed inspection through foveal vision. In this sense, peripherally located image features may drive attention in terms of where to look next, so that we may identify what detail is present at those locations.

The dual what and where feature-driven view of vision is a useful preliminary metaphor for visual attention, and indeed it has formed the basis for creating computational models of visual attention, which typically simulate so-called low-level, or bottom-up visual characteristics. However, this view of attention is rather simplistic. It must be stressed that a complete model of visual attention involves high-level visual and cognitive functions. That is, visual attention cannot simply be explained through the sole consideration of visual features. There are higher-level intentional factors involved (e.g., related to possibly voluntary, preconceived cognitive factors that drive attention).

In the 1940s Gibson (1941) proposed a third factor of visual attention centered on intention. Gibsons proposition dealt with a viewers advance preparation as to whether to react and if so, how, and with what class of responses. This component of attention explained the ability to vary the intention to react while keeping the expectation of the stimulus object fixed, and conversely, the ability to vary the expectation of the stimulus object while keeping the intention to react fixed. Experiments involving ambiguous stimuli typically evoke these reactions. For example, if the viewer is made to expect words describing animals, then the misprint sael will be read as seal. Changing the expectation to words describing ships or boats invokes the perception of sail. The reactive nature of Gibsons variant of attention specifies the what to do, or how to react behavior based on the viewers preconceptions or attitude. This variant of visual attention is particularly relevant to the design of experiments. It is important to consider the viewers perceptual expectation of the stimulus, as (possibly) influenced by the experimenters instructions.