Mark, D. M., 1993. Human spatial cognition. In Medyckyj-Scott, D., and Hearnshaw, H. M., editors, Human Factors in Geographical Information Systems, Belhaven Press, 51-60.

 

Human Spatial Cognition

 

David M. Mark

National Center for Geographic Information and Analysis

Department of Geography

State University of New York at Buffalo

Buffalo, New York 14261 U.S.A.

 

 

Introduction

Human spatial cognition is fundamental to human life itself, in both obvious and subtle ways. The obvious way is that almost all of us move around each day, and must deal in real time with a wide range of geographic concepts, features, and hazards. One less obvious way is the spatial basis for a great many metaphors that help us shape and understand more abstract conceptual domains. In this paper, after presenting some definitions, I will concentrate on reviewing a variety of ways of categorizing and conceptualizing spatial concepts, especially geographical ones. At the end, I will make some brief remarks regarding implications for human-computer interaction involving geographical software, themes that no doubt will be picked up elsewhere in this book.

Before beginning the presentation of models of geographic knowledge and spatial cognition, it is appropriate to provide definitions of some fundamental terms and associated concepts.

Cognition. Cognition is difficult to define, but is commonly used to refer to a wide range of mental processes—thought, reasoning, memory, perception, etc. There are some disagreements over several details of the definition, however. Liben (1988) state that 'there is a further issue of whether or not one insists on cognizance (awareness) in the individual' (Liben, 1988, p. 172). She distinguishes spatial thought (conscious) from spatial storage (not directly available to consciousness). In their definition of spatial cognition, Hart and Moore appear to include non-conscious spatial knowledge: '[S]patial cognition is the knowledge and internal or cognitive representation of the structure, entities, and relations of space; in other words, the internalized reflection and reconstruction of space in thought' (Hart and Moore, 1973, p. 248). In this paper, the inclusive definition of spatial cognition will be adopted, thus including memory (storage)

Perception. In this paper, the word 'perception' will be used to refer to sensation in the brain in the immediate presence of sensory stimuli; it explicitly excludes memory, reflection, conscious reasoning, etc. If this definition, which corresponds more or less with definitions in cognitive science and psychology, is adopted, then the use of the term by geographers under 'environmental perception' is incorrect.

Cognitive science. .'Cognitive science is a new field that brings together what is known about the mind from many academic disciplines: psychology, linguistics, anthropology, philosophy, and computer science' (Lakoff, 1987, p. xi). The interdisciplinary nature of cognitive science is part of its essence—it is not simply psychology re-titled.

Metaphors are a fundamental cognitive process (Lakoff and Johnson, 1980; Lakoff, 1987). Broadly defined, a metaphor exists when an unfamiliar conceptual domain is understood in terms of a familiar one. Lakoff (1987, p. 276) stated his idea of the essential aspects of metaphor:

'Each metaphor has a source domain, a target domain, and a source-to-target mapping. To show that the metaphor is natural in that it is motivated by the structure of our experience, we need to answer three questions:

1. What determines the choice of a possible well-structured source domain?

2. What determines the pairing of the source domain with the target domain?

3. What determines the details of the source-to-target mapping?'

Metaphors are a fundamental part of human cognition, and many metaphors have spatial source domains. Metaphors also are very important in user interface design for all computer systems, including GISs (Gould and McGranaghan, 1990; Mark, 1992; Kuhn and Frank, 1991).

'Spatial' vs. 'Geographic(al)'. There is considerable overlap in the meanings of the terms 'spatial' and 'geographical'. The Oxford English Dictionary defines spatial in several ways, of which the most appropriate seems to be 'of, pertaining, or relating to space'; space is then defined at its most general level as 'denoting area or location.' Geographical is defined by the O.E.D. as 'of or pertaining to geography', with geography defined as 'the science which has for its object the description of the earth's surface.' 'Spatial' is thus seen as a very general term, and 'geographical' as more specific. The differences in the O.E.D. definitions are consistent with the author's preference to see 'geographical' as that subclass of 'spatial' which deals with relations at scales ranging from a few metres to the size of the planet. Examples that would be included under 'spatial' but not 'geographical' include shapes of molecules; typical CAD (computer-assisted design) applications; and relative positions of stars and other bodies in astronomy. A final point regards the uses of 'geographic' and 'geographical'; whereas both are acceptable spelling alternatives in both British and American usage, 'geographic' is the preferred form in North America, and 'geographical' in Europe.

 

Kinds of Spatial Information

Classification is a powerful tool in science and in everyday life. Information that forms a continuum in the real world may be difficult to comprehend; thus, a division of such continua into sets of mutually-exclusive and collectively exhaustive categories ('kinds') may aid in comprehension and discussion. Classifying spatial knowledge and cognitive abilities is perhaps even more problematic that applying classification in the natural sciences, but seems almost essential nevertheless.

Thus, despite potential difficulties, this paper will be organised around classifications of spatial knowledge and cognition. Three different classification dimensions will be used to draw out different aspects of the topic. First, I will use something I call the 'nature' of spatial knowledge, dividing spatial knowledge into three broad categories: declarative, procedural, and configurational. Next, a new classification based on 'sources of spatial information for cognition' will be outlined. This distinguishes haptic spaces, pictorial spaces, and transperceptual spaces, which are dominated by information from touch and bodily motion, from vision and hearing, and from wayfinding experience and inference, respectively. And thirdly, the paper will outline a classification based on human interaction with the world and associated linguistic use, which introduces size and scale as explicit factors in the classification. After presenting a synthesis across these classifications, the paper will then outline some critical differences across users that user interfaces will have to accommodate, and lastly present some relations of these models to human-computer interaction.

Classification by the Nature of the Spatial Knowledge

There have been many efforts to review and categorize spatial knowledge, especially geographic knowledge. Some of these works have concentrated on the development of spatial concepts and abilities in young children. The work of Jean Piaget (especially Piaget and Inhelder, 1956) was and is very influential, although much debate has centred around the issue of the exact ages at which Piaget proposed that these abilities developed, the over-all schema presented by Piaget has stood the test of time quite well. Some workers have suggested that adults in unfamiliar environments may recapitulate the stages of spatial knowledge as developed by young children. This extension of Piagetian work is more speculative and uncertain than its application in developmental psychology. This paper will not deal with development or acquisition of spatial knowledge or concepts in any detail, but will concentrate on relatively stable forms and aspects of spatial knowledge in adults in familiar environments. For a recent review of developmental issues in spatial cognition, see Blades (1991).

Many authors have divided knowledge of geographic space into 2, 3, or 4 'kinds'. It is clear, however, that three types of knowledge of geographic space seem to be recognizable in all of these works.

Declarative Geographic Knowledge. Declarative geographic knowledge could also be called 'geographic facts' (Freundschuh, 1991). This term is used here to denote any knowledge of geographic space, and in one sense is a superset of the other two kinds of spatial knowledge listed below, especially configurational. Typical of declarative knowledge are facts about the locations, sizes, populations, etc., of geographic objects:

'Canberra is the capital of Australia'

'Mount Saint Helen's in in the State of Washington, U.S.A.'

'National Parks make up about 10% of Costa Rica by area'

'Indonesia is the only Asian country that is cut by the equator'

Such facts might be known by a person who cannot even identify Australia, Washington, Costa Rica, or Indonesia on a world map, never mind locate the named features within those regions. Declarative geographic knowledge may be acquired from real-world experience (as part of procedural knowledge) or from maps (configurational), and later abstracted as isolated facts. Declarative knowledge also can be acquired more directly from books, newspapers, films, television, personal communication, and other sources.

Procedural Geographic Knowledge. Procedural knowledge of geographic space is evidenced by the ability of people to find their ways from place to place. It is 'procedural' in the sense that it allows a person to perform a task, in this case navigation. Procedural knowledge may contain information that is not accessible to the conscious mind.(see discussion below). Kuipers (1975) modelled procedural knowledge of space, which he called sensorimotor knowledge, as a collection of 'view-action pairs' (V,A): when the environment provides a certain set of visual and other sensory cues (the 'view'), the traveller should perform some 'action' (such as a certain turn). Complete procedural knowledge also contains information on the view that will follow the action ( (V,A)->V), and would allow a person to describe the route. Kuipers pointed out that partial knowledge, in which some or all of these links to subsequent views are missing, still allows travel, since after the action, the environment supplies the next view. This is consistent with Donald Norman's ideas that knowledge is often divided between 'the world' and 'the head', that is, that some of the knowledge needed to perform a task is learned by the performer, but other essential information is in the external world (Norman, 1989, pp. 54-80). Liben (1988, p. 173) made a similar point:

'A fundamental distinction may be made between "doing" and "knowing." The former concerns activities in space that may be accounted for on the basis of perceptual cues, the latter those that must be attributed to some kind of internal spatial knowledge (spatial storage), or aware and able to manipulate (spatial thought)'

What is being called 'procedural knowledge' in this paper in its pure form is a combination of spatial storage with the perceptual clues of 'doing'.

Configurational Geographic Knowledge. Configurational knowledge of geographic space is 'map-like', and often has or approximates a Euclidean geometry. Knowledge of geographic space does not have to be perfect in order to be configurational; various states of partial configurational knowledge may exists. The 'lowest' form of configurational knowledge might just show connections between objects ('topology'), and indeed Kuipers (1975) and some other authors have identified 'topological knowledge' as a kind of stage intermediate between 'procedural' and 'configurational'. Topology might then be enhanced by general distance and direction information among nodes or 'anchor points' (see Couclelis et al., 1987; MacEachern, 1992), by correct angles at road junctions, and perhaps eventually by orientation, scale, and location of the area relative to other locations or latitude-longitude coordinates. In its fully-developed form, configurational knowledge would allow a person to estimate absolute distances and directions between known points as accurately as they could do that while looking at a geometrically-correct map.

Transformations Among These Kinds of Geographic Knowledge. A critical question is the relation among these three kinds of geographic knowledge, and perhaps even more critical is the ability of human cognitive processes to transform geographic knowledge from one form to another. Declarative knowledge can be thought of as a superordinate to configurational knowledge, and thus no transformation procedure is needed. However, non-configurational declarative knowledge can also develop from configurational through 'forgetting' of the configuration itself. For example, one might at one point learn that Paris is the capital of France in the context of a map of Europe showing capital cities, and might be able to reproduce that map as an accurate configuration from memory; later, however, one may still know the fact 'Paris is the capital of France' but have forgotten the relative position of Paris within France, or even of France within Europe.

Declarative knowledge can be obtained in straightforward manner from procedural knowledge of routes and paths. However, it is important to note that not all facts contained within procedural knowledge are accessible to the mind outside of the context of those procedures. For example, while driving in an urban setting, one might recognize a turning point by a particular restaurant; however, one might not be able to give the location of that restaurant, or even recognize the name of the restaurant. Transfer of knowledge from procedures to declarative facts probably happens spontaneously to some extent, but is not automatic or universal.

Procedural knowledge can be derived from configurational knowledge. This happens whenever we read a map to plan a route, since the map's configuration must be in the mind, at least on a short-term basis, in order for the planning to take place. People often plan novel routes without access to printed maps, and this requires mental access to configurational knowledge (at least partial knowledge at that level), followed by inference to produce the transformation.

Derivation of configurational knowledge of geographic space, given only knowledge of routes at a procedural level, is perhaps the most interesting and most controversial transformation. Kuipers (1978) presented a computer model (TOUR) that allowed configurational knowledge of such transperceptual spaces to be inferred from procedural knowledge of routes. However, work by Lloyd (1989), Freundschuh (1991), Golledge et al. (1992), and other researchers has cast doubt on whether humans actually do infer configurational information from procedural knowledge under normal circumstances. As suggested by Mark and McGranaghan (1986), it is probably that for most people, configurational knowledge of transperceptual (geographic) spaces is derived mostly from maps, which then provide a reference frame for other information acquired experientially; without access to maps, it seems that knowledge of transperceptual spaces often remains at a procedural level.

Classification by Sources of Spatial Information for Cognition

Recently, Mark (1992) presented a model of three fundamental and distinct concepts of space used in human spatial cognition, differentiated according to the perceptual of cognitive source of that information. In this model, haptic spaces are defined in the first instance by touching and bodily interaction, pictorial spaces are understood in through the visual experiences, and transperceptual spaces are learned primarily through inference during wayfinding. However, Mark also argued that these spaces are arranged hierarchically in the order listed above, with each being built in part on concepts from the previous ones. Thus, pictorial spaces are understood in part through metaphorical extensions of haptic concepts; and similarly, transperceptual spaces are understood in part by metaphorical extensions of concepts and terms from pictorial and to a lesser extent haptic spaces. The connection between this model of spaces and Lakoff's metaphor definition is that the spaces to the right are to some extent talked about, and perhaps 'understood', in terms of the spaces to the left. The confirmatory conjecture is that metaphors and mappings from right to left are much less common. If true, then the left-hand spaces are more fundamental or basic or (perhaps) primitive, and the right-hand spaces more conceptual and derived.

 

Figure 1: 'Metaphors' or 'mappings' between kinds of spaces. The spaces to the right are structured in part by concepts transfered by metaphor from spaces to the left (from Mark, 1992).

Haptic spaces. Sensorimotor and haptic perception are the most important early form of spatial information that reaches the mind, and in many ways is the most basic. Solid body motion is central to haptic spaces, and since solid-body motion and Newtonian physics lead to a Euclidean geometry, such a geometry applies to haptic spaces.

Pictorial spaces. Pictorial spaces are based primarily on visual perception, although the auditory and olfactory senses also contribute to a pictorial concept of space. The visual system collects fields of relative brightness in different electromagnetic channels at the retina, and converts these sensations into higher level perceptual sensations in the visual cortex. Haber and Wilkinson (1982, p. 25) claimed that the visual system goes farther than a two-dimensional picture-like image: 'the visual system attempts to interpret all stimulation reaching the eyes as if it were reflected from a scene in three dimensions'. The geometry of a two-dimensional image of objects is connected to the three-dimensional geometry of those objects through projective geometry. People talk about the visual scene in part using the language of touch and manipulation, which is one form of evidence that pictorial space is metaphorically grounded in haptic space.

Transperceptual spaces. 'Transperceptual space is composed or assembled in the mind from a number of independent haptic or pictorial spaces or objects experienced over time' (Mark, 1992). Downs and Stea (1977) coined this term to refer to the fact that such spaces are not perceived all at once. Later in the book, Downs and Stea (1977, p. 199) contrasted the terms 'small-scale perceptual space' and 'large-scale geographic space', thus implying 'geographic space' as at least an approximate synonym for both 'large-scale space' and 'transperceptual space'. Kuipers (1978, p. 129) used the term large-scale space, to refer to a 'space whose structure cannot be observed from a single viewpoint.'

Classification by Experiential Interaction and Linguistic Uses

Experiential realism is a philosophical stance expounded by George Lakoff in his 1987 book Women, Fire, and Dangerous Things. Experiential realism claims that each individual builds a personal mental model of the world, but that the cognitive models of different individuals are similar in their basic elements because most human bodies and senses are very similar, and are interacting with the same 'real world.' The cornerstone for Lakoff's philosophical position is the work of psychologist Eleanor Rosch. Rosch (1973, 1978) developed a theory of basic-level concepts that are commonly used by humans in reasoning and discourse. Other concepts that are either superordinate or subordinate to these in a hierarchy of conceptual categories are not used as much, and have less definite properties. Another closely-related model is that of radial category structures. Lakoff (1987) argues that most cognitive categories are not well-modelled by mathematical sets, or even fuzzy sets. More often, a category has prototypical members, and other things are added to the category by resemblances of various sorts to things already in the category, which leads to an internal category structure that may have several distinct branches leading out from a core; things at the ends of different branches may have few if any objective properties in common.

Central to the theories of Rosch and Lakoff is the idea that cognitive categories are often based on interactions between the human body and senses and the external world. At a workshop on 'Languages of Spatial Relations' held in Santa Barbara in January 1989, linguist David A. Zubin presented a model of spatial objects and spatial concepts that is grounded in this philosophical approach. Zubin has not subsequently published a more detailed description of his model, but Mark et al. (1989a) reported some details of the model.

Zubin defined four kinds or types of spatial objects or spaces, which he denoted by letters 'A' through 'D'. Type A includes a variety of small, manipulable objects. They are typically less than the size of the human body, and are known in a 'holistic' sense—for example, if we see one side of a cup or a book, we do not wonder what the other side looks like. Additional examples of Type A objects are small plants and animals and various hand-sized artifacts. In contrast, Type B objects are typically larger than human bodies, and typically are neither manipulable nor moveable. Zubin's examples included "the outside of a house, an elephant, trees, large machines, a fence, a mountain, a pond in the woods" (Mark et al., 1989). Zubin's Type C include scenes that are perceived by scanning. Examples include fields, caves, or the insides of large rooms within buildings. And lastly, Zubin's Type D included spaces that cannot be perceived as units, but must be built up from components. Typical examples are geographic spaces, but the entire inside of a mult-roomed building is also a member of Type D in Zubin's model.

This model mixes 'objects' (A,B) and 'spaces' (C,D), and the details will have to be developed and formalized if the model is to be used in cognitive science or GIS. However, it introduces directly the importance of size (scale) of things relative to the human body, as well as modes of human-world interaction and sensing. For example, Zubin's type A objects have their essence in the world of haptic and sensorimotor experience, whereas type D spaces are equivalent, perhaps completely, with the transperceptual spaces of Downs and Stea (1977) discussed in the last section. Another point to note is that paper maps are clearly type A objects, that represent type D spaces.

At that workshop, Leonard Talmy (another linguist) noted that the way in which individuals interact with an object may be more important than size alone; they may 'chunk' pieces of the large scale space (Mark et al., 1989a). For example, an automobile is almost certainly a type A object, even though it is larger than a human being. Zubin also noted that we can (mentally) 'convert' objects from one type to another. Talmy also noted that there is no need to talk about this as an absolute scale effect (Mark et al., 1989a).

The connection between 'Zubin spaces' and the metaphor-grounding model based on sensory sources of spatial concepts outline above is obvious, and in fact Zubin's concepts were one of the sources of evidence that led to that model (Figure 2). One point of apparent disagreement is that Zubin's 'type B' and 'type C' are both listed under 'pictorial spaces', because in Zubin's original definitions they both are based on visual scanning. Perhaps the same sorts of things can be viewed as type B when they are objects ('figures') and as type C when they are spaces ('grounds'); clearly, additional work is needed to clarify these distinctions and establish definitions.

Figure 2: Relation of Zubin's model of spatial objects and spaces to the 'metaphors' or 'mappings' between kinds of spaces presented in Figure 1.

 

Synthesis Across the Classifications

Interestingly, decades ago, Piaget presented similar concepts in 'The Child's Conception of Space' (Piaget and Inhelder, 1956). That book is divided into two broad sections: Topological Space (Part One) and Projective Space (Part Two). And the first two chapters under 'Topological space' are entitled 'Perceptual space, representational space, and the haptic perception of shape'; and 'The treatment of elementary spatial relationships in drawing—"Pictorial Space"'. The similarity is not limited to the names, but goes into a developmental sequence for conceptualizations of space, proceeding from 'perceptual or sensori-motor space' to 'the recognition of shapes ("haptic perception")' to space in drawings. However, Piaget does not extend the work to wayfinding, and how models of maze-like transperceptual spaces are built, but rather discusses how children typical 'progress' to more formal geometrical concepts, including perspective ('projective space').

These spaces can also be related back to the 'kinds' of spatial knowledge presented in the first main section of this paper. Configurational knowledge is especially interesting, since in its most developed form it has a Euclidean geometry and is 'map-like'. It may be that configurational knowledge belongs to the realm of haptic spaces and Zubin's 'Type A' objects, just as artifactual maps do. And this might in turn account for the fact that building such spaces from just wayfinding experience within a transperceptual space is often difficult, and may not happen at all in the course of normal activities. A stronger claim is that Euclidean configurations mis-represent geographic spaces as such places are experienced. If one needs a configurational model of geographic space, The power of the map lies in the way it allows spatial reasoning methods from the familiar, haptic, table-top world (Zubin's 'Type A') to be applied to transperceptual spaces (Zubin's 'Type D'); but the cost is that the map does not represent the world as it is experienced.

Human Differences

Before going on to discuss, briefly, the importance of spatial cognition to human computer interaction and user interfaces for GIS, one additional set of issues must be raised: differences within the human species. While most of the models discussed above a generalized enough that they may approach universality, their relative importance varies by culture, language, and training, as well as at an individual level.

Cultural and linguistic differences. The relation of concepts to language and culture is somewhat controversial. A scientist would say that the world exists in some objective sense, and is the same for everyone. The extreme statement of the opposite view is the approach of linguist Benjamin Lee Whorf:

We cut nature up, organize it into concepts, and ascribe significance as we do, largely because we are parties to an agreement to organize it in this way—an agreement that holds throughout our speech community and is codified in the patterns of our language. ... We are thus introduced to a new principle of relativity, which holds that all observers are not led by the same physical evidence to the same picture of the universe , unless their linguistic backgrounds are similar, or can in some way be calibrated. (Whorf, 1940, pp. 213-214.)

While many linguists today reject Whorf's extreme form of 'linguistic relativism', milder forms, such as those presented by Leonard Talmy (Talmy, 1983) in his paper 'How Language Structures Space'. Mark et al. (1989b) and Gould et al. (1991) have argued that natural language is an important factor in the design of user interfaces for GIS, but did not present empirical evidence. Thus the importance of language and culture, relative both the differences discussed in the next two sections, remains an open question.

Disciplinary and experiential differences. Professionals and other 'educated' people are often trained to see and categorize the world in particular ways. Those trained in a western scientific culture may believe that science identifies truth, and take naive realism into their everyday lives as well as in their professional practice. Foresters may see a forest as stands, with definite boundaries, while an ecologist might see the same forest as plant communities and ecotones with no sharp boundaries. At the same time, an environmentalist may note the aesthetic and perhaps spiritual value, and a logger may see a source of income for his family and of personal danger. It even seems that within a single discipline or professional, there are schools of thought with their own conceptual models. Such differences almost certainly apply to concepts of and in geographic space, although the topic has hardly been studied.

Individual differences. Last but certainly not least, there are individual differences in many aspects of cognition. Some of these may be correlated with handedness or gender, but individual variability in spatial tasks is high (see Gould, 1989). One approach to this in user interface design is the adaptive interface, which has software to detect individual differences, and adapt accordingly. Another way is to adjust the user's spatial concepts through training or experience with a GIS. Again, both of these solutions are open areas for further research.

 

Relations to Human-Computer Interaction

Human spatial cognition relates to human-computer interaction for geographical information systems in a number of ways. One is in the area of geographic concepts. A recent agenda for HCI research in GIS identified geographic concepts as one of four major themes (Mark et al., 1992). Ideally, the user interface of a computer program should present the user with concepts that are consistent with the user's mental model of that phenomenon in the real world, which in the case of GIS means geographic concepts. However, this raises a major problem for software developers: how can a system accommodate the great range of individual mental models of the world that are due to culture, language, experience, and individual differences? Adaptive user interfaces may be a possibility in the more distant future, but at present the best we can hope for is the inclusion of tools that will allow the user, the vendor, or some third party to customize the user interface.

A second way in which human spatial cognition relates to user interface design is that different kinds of spatial knowledge lead to different human-computer interaction paradigms for any software. This became obvious in a workshop conducted at the 1990 ACSM Computer Human Interaction Special Interest Group (CHI'90) by Werner Kuhn, Max Egenhofer, and Andrew Frank (Kuhn and Egenhofer, 1991), and recently was discussed by Mark (1992),

A third area in which spatial cognition is important is in thinking about the space of the computer display device or the hard-copy map. Zhang and Mark (1992) noted that managing space for objects on the screen in a cartographic design program also involves spatial knowledge. In this case, it is knowledge of how objects can behave in the Euclidean world of Zubin's 'Type A' objects, which primarily belong to sensorimotor space. The fact that this knowledge of the graphical world of the screen or paper must be combined somehow with knowledge of the geographical world that the GIS or other software represents is perhaps what makes user interfaces for GIS an especially interesting and challenging topic for theoretical and applied research.

Each of the above areas in which spatial cognition is connected to GIS provides a wealth of non-trivial research topics. However, in addition, the three areas sketched above interact. Designers of user interfaces will have to be careful to avoid cofusion between spatial concepts of the interface and the spatial concepts of the geographic world that the GIS represents.

Acknowledgements

This paper is a part of Research Initiative #13, 'User Interfaces for GIS', of the National Center for Geographic Information and Analysis, supported by a grant from the National Science Foundation (SES-88-10917); support by NSF is gratefully acknowledged. The author also wishes to thank Scott Freundschuh, Werner Kuhn, Mike Gould, and Andrew Frank for many valuable discussions that helped shape the ideas presented in this paper.

 

References Cited

Blades, M., 1991. The development of the abilities required to understand spatial representations. In Mark, D. M., and Frank, A. U., editors, Cognitive and Linguistic Aspects of Geographic Space. Dordrecht, Netherlands: Kluwer, pp. 81-116.

Couclelis, H., et al., 1987. Exploring the anchor-point hypothesis of spatial cognition. Journal of Environmental Psychology, 7: 99-122.

Downs, R. M., and Stea, D., 1977. Maps in Minds: Reflections on Cognitive Mapping. New York: Harper and Row.

Freundschuh, S. M., 1991. Spatial Knowledge Acquisition of Urban Environments from Maps and Navigation Experience. Unpublished doctoral dissertation, Department of Geography, State University of New York at Buffalo.

Golledge, R. G., Gale, N., Pellegrino, J. W., and Doherty, S., 1992. Spatial knowledge acquisition by children: Route learning and relational distances. Annals, Association of American Geographers, 82(2): 223-244.

Gould, M. D., 1989. Considering individual cognitive ability in the provision of usable navigation assistance. Proceedings, VNIS'89, IEEE Conference on Vehicle Navigation and Information Systems, Toronto, September 12-14, 1989: 443-447.

Gould, M. D., Mark, D. M., and Gavidia Gadea, C., 1991. Resolucio de problemas geograficos en ingles y español: Implicaciones para el diseño del los sistemas de informacion geografica. Proceedings, III Conferencia Latinoamericana sobre Sistemas de Informacion Geografico, Viña del Mar, Chile, October 21-25 1991, v. 1, pp. 32-41.

Gould, M. D. and McGranaghan, M. (1990) Metaphor in geographic information systems. Proceedings, Fourth International Symposium on Spatial Data Handling, Zurich, Switzerland 1: 433-442.

Haber, Ralph Norman, and Wilkinson, Leland, 1982. Perceptual components of computer displays. IEEE Computer Graphics & Applications, May 1982, 23-35.

Hart, R. A., and Moore, G. T., 1973. The development of spatial cognition: A Review. In Downs, R. M., and Stea, D., editors, Image and Environment: Cognitive Mapping and Spatial Behavior. Chicago, Aldine Publishing Company, pp. 246-295.

Kuhn, W., and Egenhofer, Max J., editors, 1991. Visual interfaces to Geometry. Santa Barbara, CA: National Center for Geographic Information and Analysis, Report 91-18.

Kuhn, W., and Frank, A. U., 1991. A formalization of metaphors and image-schemas in user interfaces. In Mark, D. M., and Frank, A. U., editors, Cognitive and Linguistic Aspects of Geographic Space. Dordrecht, Netherlands: Kluwer, 419-434.

Kuipers, B.,1978. Modeling spatial knowledge. Cognitive Science 2, 129-153.

Lakoff, G., 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind, Chicago: University of Chicago Press.

Lakoff, George, and Johnson, Mark, 1980. Metaphors We Live By. Chicago: University of Chicago Press.

Liben, L. S., 1988. Conceptual issues in the spatial development of spatial cognition. In Stiles-Davis, J., Kritchevsky, M., and Bellugi, U., editors, Spatial Cognition: Brain Bases and Development. Hillsdale, New Jersey: Lawrence Erlbaum Associates, pp. 167-194.

Lloyd, R., 1989. Cognitive maps: Encoding and decoding information. Annals, Association of American Geographers, 79: 101-124.

MacEachren, A. M., 1992. Application of environmental learning theory to spatial knowledge acquisition from maps. Annals, Association of American Geographers, 82(2): 245-274

Mark, D. M., 1992. Spatial Metaphors for Human-Computer Interaction. Proceedings, Fifth International Symposium on Spatial Data Handling, Charleston, South Carolina, August 1992, v. 1, 104-112.

Mark, D. M., Frank, A. U., Kuhn, W., McGranaghan, M. Willauer, L., and Gould, M. D. , 1992. User Interfaces for Geographic Information Systems: A Research Agenda. Proceedings, ASPRS/ACSM Annual Meeting, Albuquerque, New Mexico.

Mark, D. M., Frank, A. U., Egenhofer, M. J., Freundschuh, S. M., McGranaghan, M., and White, R. M., 1989a. Languages of Spatial Relations: Initiative Two Specialist Meeting Report. Santa Barbara, CA: National Center for Geographic Information and Analysis, Report 89-2.

Mark, D. M., Gould, M. D., and Nunes, J., 1989b. Spatial language and geographic information systems: cross-linguistic issues. Proceedings, 2nd Latin American Conference on Applications of Geographic Information Systems, Merida, Venezuela, 105-130.

Mark, D. M., and McGranaghan, M., 1986. Effective provision of navigation assistance for drivers: A cognitive science approach. Proceedings, Auto-Carto London, London, England, September 14-19, 2: 399-408.

Norman, Donald A., 1998. The Design of Everyday Things. New York: Doubleday Currency

Piaget, J., and Inhelder, B., 1956. The Child's Conception of Space. London: Routledge & Kegan Paul.

Rosch, E., 1973. On the internal structure of perceptual and semantic categories. In T. E. Moore (editor), Cognitive Development and the Acquisition of Language, New York, Academic Press, pp. 111-144.

Rosch, E., 1978. Principles of categorization. In E. Rosch and B.B. Lloyd (editors) Cognition and Categorization. Hillsdale, NJ: Erlbaum, pp. 27-48.

Talmy, L., 1983. How language structures space. In H.Pick and L. Acredolo (editors) Spatial Orientation: Theory, Research and Application. Plenum Press.

Whorf, B. L., 1940. Science and linguistics. Technology Review (M.I.T.), 42(6), 39-231 and 247-286. Reprinted in Carroll, John B., , editor, 1956. Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf. Cambridge, Massachusetts: The M.I.T. Press, pp. 207-219.

Zhang, F.-b., and Mark, D. M., 1992. Object-Oriented Spatial Knowledge Representation and Processing: Formalization of Core Classes and Relationships. Proceedings, Fifth International Symposium on Spatial Data Handling, Charleston, South Carolina, in press.