Origins of Eye Movement Research

Human eyes make dozens of movements – saccades – every second. Analysis of eye movements therefore provides valuable insights into psychological and cognitive function in a number of real‐world tasks, including reading and visual exploration of computer displays (Goldberg et al., 2002). Eye movements are generally studied through acquisition and analysis of eye movement trajectories via eye tracking.

A saccadic eye movement involves the synchronised movement of both eyes at a rapid speed from one position to another. Up to 173,000 saccades are made each day (Abrams et al., 1989). The exact duration of a saccade will vary according to the visual input, although the speed of a saccade can be as high as 500 degrees per second (Rayner, 1998).

Any object or scene the eyes pass over during a saccade may not be properly visualised due to saccadic suppression (Bridgeman et al., 1975). This implies that information is derived from the visual scene during fixations. We study eye movements on the assumption that fixations are foveating points of interest.

56 There are a variety of techniques for measuring eye movements, which vary according to the extent to which they are invasive and in their accuracy. Unfortunately, these factors are inversely correlated. The most basic method of eye movement measurement is direct observation. We are reasonably good at being able to tell, broadly speaking, what another human is looking at. However, although this is the least invasive method, it is also the least accurate. Trajectory and velocity of saccades are largely impossible to track by observation, and this method would only be somewhat useful in recording the final fixation point of the eyes. Our capability to accurately record eye movement has improved considerably. Wade and Tatler (2005) provide a summary of eye tracking technologies and note that remote devices – not head mounted, but rather at a comfortable and not visually interfering distance from the participant – provide sufficient accuracy to record patterns of fixations while minimising interference with vision. This is the type of eye tracker used in the second experiment.

Çöltekin et al. (2009) studied map designs using eye movement analysis as well as usability metrics (e.g., speed and accuracy of comprehension). Çöltekin et al. (2009) state that ‘softcopy’ (digitally-based) maps are becoming the norm with the spread of internet access and mobile devices. The need for usable interface tools for digital cartographic mapping is therefore an important factor to consider for the future of maps. Basic map design principles, and to an extent the usability principles applied in the work of Çöltekin et al. (2009), are important to consider because maps are one of the oldest examples of a visual display of complex information. The evaluation carried out by Çöltekin et al. (2009) used two different interactive maps services, Natlas (2008) and Carto.net (2008), which have different layouts, in order to compare their usability. Çöltekin et al. (2009) state that it is a common assumption that more fixations “indicate a less efficient search strategy, longer fixations indicate difficulty with the display, and plotting scan paths and fixations will allow documenting” (Çöltekin et al., 2009). The authors provide an account of a user’s typical visual search pattern where two processes occur. The first is perceptual, during which the user locates the target area of interest, and the second is cognitive, where the user will processes the information at the target location. The same principle applies for any visual processing of a complex graphic. The visual processing of complex objects or a scene with many items of potential interest will not generate the same response from every person. People’s eyes will fixate on certain elements of the objects, and typically complex objects or scenes will contain many elements.

57 An early study by Yarbus (1967) on the 1884 art work by Ilya Repin called ‘An Unexpected Visitor’ shows that participants focused more on elements of the painting than they thought; as Yarbus put it, ‘useful and essential for perception’. Figure 2.2 shows the original work, while Figure 2.3 shows an example of participants recorded eye movement patterns over the course of a three minute period. Through a series of experiments such as this one Yarbus found participant eye movements were not necessarily determined by the number of details. This is interesting to note because it is linked to more modern research, where the number of elements presented to participants vary greatly. Yarbus also noted that light and dark details did not attract participant attention unless essential information lay within those details. In relation to the experiment conducted for this thesis, the lightest and darkest areas would be the ones containing the most information and therefore participants’ attention and foveation would be attracted to these areas of the visualisations.

Figure 2.2 Repin's ‘An Unexpected Visitor’, 1884

Yarbus (1967) concluded in his ground-breaking work that eye movement is at most only slightly dependent on the contents of a scene being observed by the eyes. However, in his experiment certain details were not highlighted as being particularly important. This experiment was exploratory in nature because there was no task requiring an answer, participants did not have to search for key elements in the scenes presented to them for three minutes at a time.

58 Figure 2.3 Illustration of participant eye movement, Yarbus 1967

A large component of this research is linked to the perception of a human being.