«Abstract. Since ancient times, people have devised cognitive artifacts to extend memory and ease information processing. Among them are graphics, ...»
Some Ways that Maps and Diagrams Communicate
Department of Psychology, Stanford University
Stanford, CA 94305-2130
Abstract. Since ancient times, people have devised cognitive artifacts to extend
memory and ease information processing. Among them are graphics, which use
elements and the spatial relations among them to represent worlds that are
actually or metaphorically spatial. Maps schematize the real world in that they
are two-dimensional, they omit information, they regularize, they use inconsistent scale and perspective, and they exaggerate, fantasize, and carry messages. With little proding, children and adults use space and spatial relations to represent
relations, temporal, quantitative, and preference, in stereotyped ways, suggesting that these mappings are cognitively natural.
Graphics reflect conceptions of reality, not reality.
1 Introduction One candidate for an intellectual achievement separating humankind from other kinds is the creation of cognitive artifacts, of external devices that extend the human mind.
They range from using fingers for counting or fashioning bends in trees to mark trails to powerful computers or global positioning systems. Cognitive tools augment the mind in two major ways: they reduce memory load by externalizing memory, and they reduce processing load by allowing calculations to be done on external rather than internal objects and by externalizing intermediate products. Of course, external representations have other benefits as well. They take advantage of people’s facility with spatial representations and reasoning, they are more permanent than thoughts or speech, they are visible to a community (cf. Tversky, in press, b). Written language is prominent among useful cognitive tools, but graphics of various types preceded written language and serve many of the same functions. What renders graphics privileged is the possibility of using space and elements in space to express relations and meanings directly, relations and meanings that are spatial literally as well as metaphorically.
Early graphics, found in cultures all over the world, include not only lines on paper or bark, but also inscriptions on trees, paintings in caves, incisions in bones, carvings in wood, scarifications on bodies, and more. They portrayed things that took up space, animals, objects, and events, actual or imagined, or spaces, the prime example being maps. It was only in the late 18th century that graphics began to be used, in the West, to portray non-spatial, abstract information, most notably, economic data, such as Ch. Freksa et al. (Eds.): Spatial Cognition II, LNAI 1849, pp. 72-79, 2000.
Springer-Verlag Berlin Heidelberg 2000
balance of trade over time (Beniger and Robin, 1958; Tufte, 1990). I will first characterize ways that graphics convey the essentially visual through a discussion of maps, the prime example of these graphics. Then I will characterize ways that diagrams visualize the non-visual.
2 Characterizing Maps
One of the most typical and ubiquitous of graphics is a map. Nowadays, when people think of maps, they think of the world map in school classrooms or the road map in the glove compartment of the car or the city map in the tourist guide. Yet the maps that have been invented across cultures throughout human existence are both far more and far less than these. They seem to share two features, though by no means strictly.
Maps typically portray an overview, and they reduce and roughly preserve scale. As we shall see, in practice, maps are not limited to overviews, and scale is not always consistent. Here, we analyze those maps, characterize what they do and do not do, and bring those insights to the study of diagrams in general.
2.1. Maps are Two-Dimensional
Although the worlds that maps typically represent are three-dimensional, maps are typically two-dimensional. This is for a number of converging reasons. Obviously, it is easier to portray a two-dimensional space on a piece of paper than a three-dimensional space. But, in addition to the constraints of the medium, there are cognitive reasons for portraying maps as two-dimensional overviews. First, it seems that people readily conceive of three-dimensional environments as two-dimensional overviews, in itself a remarkable cognitive achievement. This is attested by the invention of twodimensional overview maps by diverse and dispersed cultures as well as their spontaneous invention by children. Next, three-dimensional diagrams are difficult to construct and difficult to comprehend (e. g., Cooper, 1984; Gobert, 1999). Architects and other designers, for example, prefer to first construct two-dimensional plans or overviews and two-dimensional elevations before integrating them into threedimensional sketches or models (Arnheim, 1977). As Arnheim points out, the considerations important for plans differ from those important for elevations. Like maps, plans display the spatial relations among large structures, providing information useful for navigating among them. Plans, then, are a presentation useful for evaluating function. Elevations provide information about what the structures look like, important for recognition of them. Elevations are a presentation useful for evaluating aesthetics.
In addition to these reasons, for many purposes, three-dimensional information about environments is simply not needed, and may even interfere; the spatial relations among the large features is sufficient information.
74 Barbara Tversky
2.2. Maps Omit Information
The next thing to notice about maps is that they omit information. One of the reasons for this has to do with the very nature of mapping. Borges’ invented fable of an Empire where the Art of Cartography was perfected so that a map of the Empire the size of the Empire could be created was just that, a fable, an absurdity (Borges, 1998).
The very usefulness of a map comes from its reduction of space. Reductions in size require reductions in information to be useful. An aerial photograph does not make a good map. Maps omit information because much of the information in space is not only not relevant, but also gets in the way of finding the essential information. Maps are typically designed for a communicative purpose; that purpose determines what information should be kept and what information can be eliminated. Consider, for examples, two kinds of maps created by seafaring cultures (Southworth & Southworth, 1982). Coastal Eskimos carried carved wood outlines of the coastline with them in their canoes to guide them in their travels. The Marshall Islanders in the Pacific, who navigate among islands too distant to be seen for much of their voyages, constructed maps out of bamboo sticks and shells. The shells indicated islands and the sticks ocean currents, identifiable from the flotsam and jetsam that accumulates along them. For more familiar examples, consider the information useful for a map to guide drivers in the city or, alternatively, a map to guide hikers in the mountains. Details of types of roads and intersections are important to the former, whereas topographical details are important to the latter.
2.3. Maps Regularize
Yet another characteristic of maps in practice is that they simplify and regularize information. A classic example is the London subway map, which has served as a model for subway maps all over the world. The London subway system, like many subway systems, is quite complex, with many different lines and intersections. The information important to travelers includes the general direction of the lines, the stops, and the intersections with other lines. The specific directions, path curvatures, and distances are not usually critical. So the lines on the London subway map are presented as straight lines, oriented vertically, horizontally, or diagonally, ignoring the finer distinctions of local curves and orientations. This simplification, however, facilitates computing the desired information, the general directions and the connections, and the distortions produced by the regularization do not cause sufficient errors to create problems.
2.4. Maps Use Inconsistent Scale and Perspective
Road maps illustrate another common feature of maps. They use inconsistent scale.
Indeed, roads, rivers, railroads, and other important environmental information portrayed in maps would simply not be visible if scale were consistently adopted. In addition, many maps violate consistent perspective. Consider, for example, a popular Some Ways that Maps and Diagrams Communicate 75 kind of tourist map. These present an overview of the city streets superimposed with frontal views of frequently visited landmarks, such as churches and public buildings.
Presenting both perspectives in a single map is a boon to tourists. It allows them to navigate the streets in order to find the landmarks, and then to recognize the landmarks when they see them. Maps with mixed perspectives are by no means a modern invention. For example, maps portraying overviews of paths and roads and frontal views of structures and living things are clearly visible in petroglyphs dating back more than 3000 years in northern Italy (Thrower, 1996).
2.5. Maps Exaggerate, Fantasize, and Carry Messages, Aesthetic, Political, Spiritual, and Humorous In 1916, King Njoya presented the British with a map of his kingdom, Banum, in northwestern Cameroon. To impress the Europeans with his modernity, he put 60 surveyors to work for two months to construct the map. While fairly accurate, the map exaggerates the size of the capital, and locates it, incorrectly, smack in the center of the kingdom (Bassett, 1998). Maps influenced by the political, mythical, historical, or fantastic are common all over the world. In medieval Europe, T-O maps were popular.
They were called that because they looked like T’s embedded in O’s, the circular background for the world. East, the direction of the (presumed) beginning of the world, Adam and Eve, the rising sun, was at the top (hence the word “oriented” from oriens or east). The top bar of the T was formed by the Nile on the south or the left and the Dan on the north or the right. The Mediterranean formed the vertical bar. Such maps portrayed religious beliefs and reflected elegant geometry and symmetry more than actual geography. They also added decorative depictions, of Adam and Eve in the Holy Land, of the four winds, and more. Maps mixing geography, beliefs, and history are not unique to Europeans. T-O maps appeared in Asia (e. g., Tibbetts, 1992), Similar maps appeared in preColumbian Mesoamerica, for example, a map showing the imagined or real migrations of the ancestors superimposed on a geographic map (Mundy, 1998). Maps of the heavenly spheres appeared in both Europe and Asia (Karamustafan, 1992).
Humorous maps enliven newspapers, books, and journals. Perhaps best known are the “New Yorker’s View of the World” maps of Steinberg that graced the covers of the New Yorker as well as many dormitory rooms. Such maps take a local perspective so that close distances loom larger than far distances. They also include landmarks likely to be of interest to the New Yorker and omit those of less interest. A more recent example from the New Yorker was a map of New York City as the palm of a hand, with Broadway as the lifeline and the boroughs as fingers. These are but a few of many, many examples of maps that are designed to convey far more than geography, and that sacrifice geographic accuracy for other messages.
Put briefly, maps, those produced by professionals as well as amateurs, are schematic (for a related view, see Freksa, Moratz, and Barkowsky, this volume).
Schematic maps are created for a specific goal or goals, usually communicative, and they distill and highlight the information relevant to those purposes. They eliminate 76 Barbara Tversky extraneous information to remove clutter, making the essential information easier to extract. They simplify and even exaggerate this information. People’s minds also schematize spatial and other information. In fact, many of the ways that minds schematize correspond to the way that maps schematize. Internal representations of environments omit and regularize information, they mix perspectives, reduce dimensionality, and exaggerate. This can lead to internal “representations” that are impossible to realize even as a three-dimensional world (e. g., Tversky, 1981; in press, a). Matching external schematizations to internal ones may also facilitate processing information from maps. Of course, the match between internal schematizations of environments and external schematizations of maps is no accident; both are products of human minds. Moreover, these same processes, omission, use of inconsistent information, regularization, exaggeration, politicization, beautification, and more, appear in other depictions and external representations.
3 Characterizing Graphics
When people talk or think about abstract concepts, they often do so in terms of spatial concepts (e. g., Clark, 1973; Lakoff and Johnson, 1980). There is good reason for this.
Spatial competence appears early in life and is essential for survival (e. g. Tversky, in press, a). Bootstrapping abstract thought onto spatial thought should allow transfer of spatial agility to abstract agility. Graphic visualizations of abstract concepts and processes such as those in economics, biology, chemistry, physics, mathematics, and more should pay similar benefits. An examination of graphics produced by children and adults throughout history and across cultures reveals some general characteristics of the way they use space and the elements in it to convey meaning (Tversky, 1995; in press, b).
3.1. Spatial Relations