CHAPTER 6. POTENTIAL SPATIO-VISUAL ADAPTATIONS
6.2 A Directive for Spatio-Visual Adaptation
Despite sparse publications on spatial-visualization-based adaptation, our experimental data and several reports in the literature point to a definite distinction: high-spatial-visualization users prefer and benefit from survey knowledge, while low-spatial-visualization users prefer and benefit from landmark knowledge. This is consistent with the landmark-route-survey acquisition process described in Siegel and White (1975). In this process, a user first recognizes landmarks, then links them together to form routes, and routes can give rise to a holistic, or survey,
understanding of an area. The connection between the LRS model and spatial visualization will be explained after presenting the next report.
Figure 6.2. Task scores from Nguyen (2012) after three differing training modes. Low-spatial-visualization and high-spatial-visualization participants exhibit differing performance profiles, flip-flopping in performance depending on treatment.
Bay & Ziefle (2008) trained 30 participants aged 9-14 on a menu navigation task for a smart phone, with the explicit goal of observing the interaction of landmark, route, and survey knowledge and spatial visualization ability. Spatial visualization was measured through the Tewes (1983) Mosaic Test. Training was conducted in three treatments: (a) a landmark mode, where participants were given the exact menu choices to complete the task; (b) survey mode, where participants were given the entire hierarchy of all possible selections; and (c) a combined mode of landmark, survey, and route knowledge, where participants were allowed to interact with the device for five minutes. High-spatial-visualization participants performed best after pure-survey training, while low-spatial-visualization participants performed best after pure- landmark training, outperforming high-spatial-visualization participants in time, number of steps taken, and number of undo actions.
Both Bay & Ziefle (2008) and Nguyen (2012) reported training modes that set low- spatial-visualization participants ahead of their high-ability counterparts. These findings were counterintuitive and striking in light of the existing literature: despite high-spatial-visualization users being the “favorites” on computer tasks, presentation modes exist that can overturn
expectations of performance. Of course, based on Williges, Elkerton, Vicente, & Hayes (1990), the impact of such designs depends heavily on task type:
However, just because the individual differences
have been assayed and isolated does not guarantee that the accommodation will be successful. This difficulty is acknowledged by Egan and Gomez (1985, p. 215): "The step of accommodating individual differences not only tests the analyses that precede it, but it also tests the theory of how an experimental manipulation ... will change the original task." (p. c-23)
Bay & Ziefle’s (2008) dichotomy of survey and landmark knowledge preferences mapping to high- and low-spatial-visualization participants were corroborated by our own data, which will be examined in the next section. But what is the nature of the relation between landmark preference, survey preference, and spatial visualization ability? The answers can be pieced together from the accounts in Rodes & Gugerty (2012) and Meneghetti, Gyselinck, Pazzaglia, & De Beni (2009).
Rodes & Gugerty (2012) investigated sixteen participants drawing a map from memory, after having used simulated aerial navigation software for an unmanned aerial vehicle. Spatial visualization ability was tested through Ekstrom et al.’s (1976) VZ-2 Paper Folding Test. After controlling for visual memory, spatial visualization ability was significantly associated with map draw error and therefore quality of recall! This outcome was surprising and counter-intuitive, as it suggested spatial visualization had a separate effect from visual memory on the construction and retention of survey knowledge. The marginal effect of spatial visualization ability could
explain why low-spatial-ability participants are less comfortable with survey knowledge and prefer landmark knowledge.
Meneghetti, Gyselinck, Pazzaglia, & De Beni (2009) ran a psychological study of 76 participants where recall of spatial and non-spatial text descriptions was measured while being interfered with through secondary tasks of spatial tapping and articulatory suppression. Spatial tapping consisted of tapping the four corners of a 30 x 24 cm rectangular board and interfered with visuospatial working memory, while the articulatory suppression task (repeating the
syllables “ba-be-bi-bo-bu”) interfered with verbal working memory. Spatial visualization ability was measured by the Vandenberg & Kuse (1978) Mental Rotations Test. High-spatial-
visualization participants were able to overcome the interference for the spatial text description (but not for the non-spatial description), while low-spatial-visualization participants suffered recall degradation for all treatments. These outcomes showed that spatial ability is used as an additional resource when processing spatial descriptions, and allowed high-spatial-visualization users to not require additional “executive resources”. As a result, high-VZ users appear to have extra capacity to manipulate and exploit survey knowledge that is subconscious, per the
outcomes from Brennan, Kelly, and Arguello (2014). In effect, high-VZ users appear to be have a preference for survey knowledge due to a modest comic-book-hero “superpower”: they can exploit survey knowledge as it arrives without engaging additional “executive resources”. On the other hand, low-spatial-visualization participants lack the spatial-visualization “superpower” and cannot exploit survey knowledge, instead preferring landmark knowledge.
Rhodes & Gugerty (2012) and Meneghetti, Gyselinck, Pazzaglia, & De Beni (2009) appear to have clarified how the landmark-route-survey process in Siegel and White (1975)
manifests itself in low- and high-spatial-visualization interface preferences. This view was corroborated by our data, as described in the next section.