«Stefan Münzer*, Hubert D. Zimmer*, Maximilian Schwalm*, Jörg Baus+ & Ilhan Aslan+ *Department of Psychology, Brain and Cognition Unit + German ...»
Computer Assisted Navigation and the Aquisition of
Route and Survey Knowledge
Stefan Münzer*, Hubert D. Zimmer*, Maximilian Schwalm*, Jörg Baus+ & Ilhan
*Department of Psychology, Brain and Cognition Unit
German Research Center for Artificial Intelligence (DFKI)
Dr. Stefan Münzer
Postfach 15 11 50
Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 2 Abstract The incidental acquisition of spatial orientation knowledge when using a pedestrian navigation assistance system for wayfinding was compared to incidental learning during mapbased wayfinding. First-time visitors to a real environment (a zoo) took a guided tour. In the navigation assistance conditions, users were provided with direction information and viewbased pictures of the current intersection at each decision point, presented on a hand-held computer. In the map-based condition, participants derived route segments from a map (each segment comprising three or four intersections), and then walked the partial routes from memory. After walking, unexpected tests on route memory and survey knowledge were administered. Navigation assistance users showed good route knowledge and poor survey knowledge. In contrast, map users showed better survey knowledge and nearly perfect route knowledge. Variations of information presentation within navigation assistance conditions (auditory vs. visual direction command, additional presentation of allocentric spatial information) was not effective. Results are explained with an active encoding principle. Only information that is actually encoded, transformed, and/or memorized during the primary wayfinding activity, is incidentally learned. Since navigation assistance systems do not require users to encode, transform, and memorize spatial information, the spatial orientation knowledge of navigation assistance users is poor.
Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 3
1. Introduction When people navigate through a real environment, they might learn about the spatial configuration of the environment incidentally, i.e. as a side effect of wayfinding. This is particularly the case if maps are used. Maps support spatial learning in two ways. First, maps present spatial information as an allocentric survey representation of the environment, thereby showing the spatial relations between places within a stable reference frame. However, this presentation is orientation specific (e.g. Rossano & Warren, 1989; Rossano & Moak, 1998).
Second, using a map for wayfinding purposes requires elaborated cognitive processing of the spatial information to derive the route from the map. The cognitive processing presumably involves mental rotation in order to align the map with the present view of the environment, which may partly overcome the orientation specificity of the map. While until recently people have commonly used maps to find their way in a novel environment, nowadays people more and more rely on navigation assistance systems. Such systems are widely used in cars, and pedestrian navigation assistance systems are available on mobile phones and hand-held computers (Baus, Kray & Krüger, 2001; Baus, Cheverst, & Kray, 2005; Krüger et al., 2004;
Wasinger, Stahl & Krüger, 2003).
Navigation assistance systems are helpful and comfortable wayfinding aids because they can indicate the to-be-adopted direction from the current position and perspective of the user. As a consequence, the cognitive spatial processing requirement on the user's side is minimized. Particularly when used in cars, the assistance systems are efficient wayfinding aids that can contribute to safety. However, navigation assistance make wayfinding comfortable and easy also for pedestrians (e.g. for tourists or business people in a foreign city). Comfortable wayfinding assistance might furthermore be used by elderly or disabled persons who have, for instance, difficulties in orientation in a new environment or in reading Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 4 maps. However, the acquisition of spatial orientation knowledge is nevertheless considered desirable, for a number of reasons. First, people might simply wish to learn about the environment they are moving through, and after a few days in a new environment, people might wish to vary their routes, and might need less support by navigation assistance. Second, spatial orientation knowledge helps to plan routes that comprise a sequence of places. Third, assistance systems can fail or break down completely. Without spatial orientation knowledge, people would be lost.
While the acquisition of spatial orientation knowledge might thus be generally desirable, the situations in which assistance systems are used differ with respect to the cognitive resources that can be devoted to spatial learning. In the car navigation situation, the driver should focus on the driving, and therefore, efficiency and wayfinding comfort as well as minimal load for the interaction with the system are the primary goals. However, effiency and minimal load while wayfinding might not be primary goals for a tourist in a city. For pedestrian wayfinding, spatial orientation learning could well be supported by a navigation assistance system.
The acquisition of spatial knowledge when using a navigation assistance system might not longer occur incidentally as a consequence of the wayfinding effort. Unlike maps, assistance systems do not provide spatial information in a complete survey view with a stable reference frame. It might be the case that little is learned about the spatial configuration. In the present paper, the consequences of using navigation assistance for spatial learning of pedestrians are investigated. The present experiment was conducted in a real environment since it is possible that spatial learning is based on the stimuli in the environment rather than on information shown in a map or presented by an assistance system.
Humans acquire different forms of knowledge of large-scale space, depending on the learning experience. Route knowledge is the knowledge of places or landmarks and the routes that connect them. The according mental representation can be conceived of as a sequence of Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 5 view-based (egocentric) visual images of landmarks together with directions (Gillner & Mallot, 1998). When exploring an environment solely by navigation (i.e. without a map), the majority of people first acquire route knowledge. Since the mental representations of the landmarks are based on views from an individual’s perspective, Aginsky, Harris, Rensinck and Beusmans (1997) called memory for routes "visually dominated". In contrast, survey knowledge indicates an understanding of the spatial relationships between locations. Survey representations provide an overview over the spatial layout, based on an extrinsic frame of reference, i.e. it is a map-like representation from an allocentric perspective (Evans, 1980;
Hart & Moore, 1973; Kitchin, 1994; McNamara, Ratcliff, & McKoon, 1984; Siegel & White, 1975; Taylor & Tversky, 1992). This so-called "mental" or "cognitive" map allows flexible spatial orientation (e.g. drawing inferences about spatial relations between places, planning of routes not yet travelled). In general, the acquisition of survey knowledge appears desirable for successful and flexible orientation in an environment.
It has been proposed that with growing experience survey knowledge is developed from route knowledge quite automatically (Siegel & White 1975; Cousins, Siegel & Maxwell, 1983; Kirasic, Allen & Siegel, 1984). However it seems that the acquisition of survey knowledge is strategic and cognitively effortful rather than automatic. People who have acquired their orientation knowledge solely by navigation have difficulties with spatial judgments that require a bird’s eye view, even if they know the environment for months or years (e.g., Thorndyke & Hayes-Roth, 1982). People differ inter-individually with respect to preferences and strategies in building up survey knowledge from navigation experience (Kozlowsky & Bryant, 1977; Pazzaglia & De Beni, 2001), and only a minority deliberately choose to build up an allocentric, map-like mental representation when learning a route by navigation, i.e. to use a "spatially dominated strategy" (Aginsky et al., 1997).
The acquisition of survey knowledge is supported by an appropriate information format. Maps directly provide a structured model of the environment including all spatial Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 6 relations between the places in the environment. Investigations of map learning have shown that people built up a coherent spatial mental model when studying maps (e.g. Denis & Zimmer, 1992; Thorndyke & Hayes-Roth, 1982; Tlauka & Wilson, 1996; Rossano & Moak, 1998). In an everyday wayfinding situation, a map is used to derive a route. For this purpose, the information from the map as a model of the environment is extracted and transformed in order to make use of it for route derivation. For instance, the allocentric map perspective has to be transformed into the current egocentric perspective in the real environment, and vice versa. Thus, in the everyday wayfinding situation two presuppositions for spatial learning are fulfilled. First, the information in the map directly provides a survey model of the environment. Second, the information is actively processed, transformed, memorized etc.
during the wayfinding activity. This makes it probable that incidental spatial learning will happen if people use maps for wayfinding.
When using a navigation assistance system, however, people are presented with information that corresponds to their current perspective and view. Furthermore, depending on the design of the system, survey information is not presented at all, might be incomplete or might also correspond to the egocentric view. As a consequence, incidental spatial (survey) learning seems unprobable during wayfinding. According to the transfer appropriate processing account (Morris, Bransford, & Franks, 1977; McDaniel & Kearney, 1984) users will learn that information that has been processed to solve the goal of wayfinding. When using a navigation assistance system, route information (i.e. the combination of views and directions) is processed during travel. The incidental learning effect should therefore concern memory for the route, but not for spatial relations of places in the environment. In contrast, when using a map, survey information has been processed during wayfinding. In addition, using maps for wayfinding requires information transformation (as described above) while this is not necessary when using a navigation assistance system. The rather shallow processing of navigation assistance users should therefore result in less distinct (Jacoby, Computer Assisted Navigation and the Aquisition of Route and Survey Knowledge 7 Craik, & Begg, 1979) and less elaborated encoding (Craik & Tulving, 1975) compared to the deeper processing of spatial information by map users. As a consequence, navigation assistance users should have less accurate knowledge of routes and spatial relations than map users after wayfinding in a real environment. Consistent with this expectation, moderate route memory and only poor survey knowledge was found in navigation assistance users (Krüger, Aslan & Zimmer, 2004). However, in this prior study no map-based wayfinding condition was included to test directly the hypothesis of poor incidental spatial learning of navigation assistance users.