Nevertheless, ancient remnants of externalized thought continue to be found scattered all over the world. Frequent among them are images of people, objects, animals, events, tools, maps, and tallies. The images may be decipherable even if their meanings elude us. Take note of the content: people, objects, space, time, events, and number, the mainstays of our lives and undoubtedly the most common content to this day. Glance at any newspaper or website and you will see representations of creatures and things, space, time, events, and number, concepts core to human existence. So core that the brain, too, makes special note of each of them.
Definition of sorts
Importantly, these worldly expressions of thought are representations. Like symbols, they stand for something other than themselves, though unlike many symbols, they bear resemblance to what they represent. Remember Magritte’s pipe. They use place in space and marks in space to convey meaning. The marks can be depictions, icons, words, symbols, and simple geometric forms like dots and lines and boxes. The words often appear singly or in small groups; they aren’t organized into sentences and paragraphs like the prose you are reading now. That is, maps, charts, graphs, diagrams, and sketches—graphics of all kinds—can be multimodal. Like conversation.
COGNITIVE DESIGN PRINCIPLES
New representations of thought are being created by the second: recipes, explanations of political conflict, stock market ups and downs, scientific phenomena, instructions for assembling or operating or dancing. Designers should consider how users will understand the representations and how they will use them, but clearly, designers of visualizations, like designers of anything, cannot anticipate all the ways people will understand and use their designs. In fact, the new understandings and new uses can be creative and significant in themselves. These goals can be incorporated into two General Cognitive Design Principles.
Principle of Correspondence: The content and form of the representation should match the content and form of the targeted concepts.
Principle of Use: The representation should promote efficient accomplishment of the targeted tasks.
Take note: the principles can and will conflict, that is, they can suggest radically different designs. Hence the generations of refinement as well as the evolution of different solutions. Consider written language. It began all over the world by mapping words to pictograms, sketches of things that resembled the things. Resemblance aligns nicely with the correspondence principle, but ultimately not with the use principle. There are thousands of words, so thousands of characters to learn. And how to find resemblance for concepts like beauty, justice, and revolution? Negations, qualifications, and hypotheticals? The typical solution was to add characters that represented sound rather than appearance. Only once in history, a different kind of representation was invented: map sound to sight rather than meaning. Voilà, the alphabet! Sung by toddlers all over the country. But difficult for people who are deaf.
Despite these shortcomings, logographic writing has survived; over a billion people read it daily. Poetry in logographic languages has a feature lacking in poetry in alphabetic languages. Alphabetic languages allow alliteration, word play based on sound. Logographic languages have that too, but they also have another layer that allows visual word play. How delightful!
There is an important lesson to be drawn from the development of writing. Rarely is there a single best solution. Nearly everyone struggles with English spelling. Recall the First Law of Cognition: There are no benefits without costs; each solution has advantages and disadvantages. The same is true for design by nature, evolution: fish swim, birds fly, snakes slither. Different ways to move in the world, and they all work. But not all the time.
SPACE: MAPS
Ancient maps reveal so much about thinking. Many early maps are more user-friendly than many designed-by-experts contemporary ones. Figure 8.2 shows what is currently thought to be the world’s oldest map, a pocketable one incised on a stone block weighing about two pounds and left in a cave in northern Spain over 15,000 years ago.
FIGURE 8.2. Paleolithic map, world’s oldest, from Abauntz Cave, Spain, ca. 13,600 BCE.
Perhaps you can see what thrilled the archaeologists who found it, that the map depicts paths as well as features of the landscape around the cave, including mountains and rivers. Rather large landmarks. Harder to see are groups of animals on both sides of the river gorge. The archaeologists speculated that it might have been used to record a hunt or perhaps to plan one. Mountains, rivers, and animals are visible aspects of the scene, placed in a spatial array, so the representation is part a depiction of a scene and part a map of one. The map shows two perspectives at once, an overview of the terrain and head-on views of features of the terrain. This double-perspective and partly depictive representation is certainly neither as realistic as a scene nor as abstract as a map, but, just as certainly, it is a helpful one, both for orientation and for planning.
Our ancestors mapped not only the ground but also the sky. One can only wonder why. The cycles of the moon were the source for many early calendars. Mapping the stars is more puzzling; perhaps their awesome beauty. Perhaps a belief that celestial bodies, so high in the sky, observe us, guard over us, control us, beliefs shared by contemporary astrology. Ancient maps of constellations have been discovered in many sites. The famous Lascaux cave in southern France has a map of the Pleiades going back at least twenty thousand years, as does the nearby Tête du Lion cave.
FIGURE 8.3. Stick and shell map used by Marshall Islands navigators.
Maps come in many forms. North Coast Native Americans made maps by annotating their slightly cuffed left hand, placing cities and towns along their index finger and thumb, Quebec at the tip of the index finger, Montreal at the joint, New York at the juncture of the finger and thumb to the hand. Michiganders hold up their left hand, fingers straight, mimicking the shape of the state, with Lake Huron occupying the gap between the index finger and the thumb. They then point to the location of their home town on their hand. Inuits who navigated the coasts of Greenland carved pocket-sized maps from wood, with curves matching the undulations of the coast. The maps fit inside a mitten and could be followed tactilely—no need to remove the glove and freeze the fingers. They floated in case they fell into the water. So clever!
Equally clever were the floating bamboo and shell “stick” maps designed by Marshall Islanders in the South Pacific and used for navigating long distances on the open ocean, like the one in Figure 8.3. The bamboo sticks represented the ocean swells and currents and wind, essentially the highways of the ocean, and the cowrie shells stood for the islands, too far from each other to be seen.
FIGURE 8.4. T-O map, Leipzig, eleventh century.
In the Middle Ages in Europe, cartography was schematic and the vast majority of the population traveled only in imagination, so maps seemed to serve spiritual, religious, and political ends. Maps like the one in Figure 8.4 were frequent, called T-O because of their form. The Holy Land, the birthplace of Jesus, is in the center. Note that Oriens, east, is at the top. Oriens means “rising,” as in the rising sun, and as in Orient and orient. The horizontal bar of the T is the Indian Ocean separating Asia from Europe and Africa, and the vertical bar is the Mediterranean Sea separating Europe and Africa. A single ocean surrounds the world as was then known in Europe. The spatial locations and sizes are vague and boundaries are missing. Don’t worry if you can’t read the small print. What is important to see is that spatial layout is highly schematic: the Holy Land is in the center of the world, and the three continents then known are separated by bodies of water.
It is not clear why European maps were later turned 90 degrees so that north was and remains up, but that change in perspective happened around the time magnetic north was discovered. The even older maps attributed to Ptolemy were also north-up; some speculate that that was pragmatic because most of the land known at the time extended east-west. Maps don’t just map space, they map so much more.
Diverse and imagina
tive as they are, these maps share core features. They aren’t always to scale. They mix perspectives. They depict as well as map. They omit masses of information. This isn’t just from ignorance or lack of technical sophistication, it’s by design. What sparse information they show differs for the different examples. The precise ins and outs of the coastline for the Inuit canoeists. The ocean swells and islands for the Marshall Islands navigators. The terrain and animals for the Iberians. The spiritual for Europeans in the Middle Ages. What’s included is exactly the information that the users of the maps needed, uncluttered by information they didn’t need.
With those features in mind, let’s skip a few millennia to a map used today by millions of people. The London Tube map. I can’t reprint it here, but Harry Beck’s map of the London Underground entered the world in 1931 and has since been imitated by transport systems all around the globe. In an odd way, it resembles the stick maps used by the Marshall Islanders only, instead of ocean currents and islands, we have underground lines and stops. It shows—but distorts—a simplified skeleton of the train lines, depicted as lines running vertically, horizontally, or diagonally, by no means an accurate reflection of their pathways. The insight that inspired the map came from electronic circuit diagrams, a fascinating instance of anachronistic analogical reasoning. Geography doesn’t matter for electricity. What matters are paths and connections, gateways to other paths. The same for commuters, Beck reasoned. What commuters needed were the paths from station to station and the connections to other Tube lines, not geographic accuracy. His design met resistance (unintended pun) from the powers that be but was an instant hit with commuters. It’s so legible. The Tube lines are color-coded. The horizontal, vertical, or diagonal lines are easy for the eye to follow. The stops are indicated by name and perpendicular blips, and the connections to other Tube lines are clearly marked by circles. There’s even some geographic information, enough so I once used the Tube map to successfully get to a place that was off the street map I was using for navigation in the ancient times before ubiquitous smartphone maps.
Like the historic maps, the Tube map includes only a small fraction of the possible information. And again, like the earlier maps, what it includes is exactly what users typically need: paths and points where actions can be taken, specifically, switching Tube lines, entering, or exiting. The Tube map goes further: it distorts distances and directions. The alert reader will remember that memory for environments does the same. And that schematic maps, even distorted ones, can be helpful, even more helpful than veridical ones because they are easy to read, to understand, and to use. As such, well-designed schematic maps follow both General Cognitive Design Principles.
You might wonder whether the missing information and distortions confuse people. Like all diagrams and, for that matter, all communications, maps are designed to be used in a context. The context can be what’s perceptible in the surroundings and it can also be shared knowledge in the mind. The context typically (but not always) provides the missing information and resolves ambiguity and distortion. That’s in the implicit contract between communication partners. Users understand that maps and other diagrams work that way, that they’re meant for a particular set of users in a specific context, known to both. The Tube map is meant to be used with the Tube, assembly instructions with the object to be assembled. We have the same understanding for language. If someone says, “Isn’t it cold in here,” it’s understood as an indirect request to close a window or turn down the AC. If your passenger-navigator says, “Turn right,” you know to turn at the next intersection, not right now.
The London Tube map also adds words and symbols. Most good maps and diagrams are multimodal, like natural conversation, which uses far more than words, such as intonation, gesture, and stuff in the world.
Maps can be designed for multiple purposes or different maps for different purposes. Maps can allow way finding and exploring an environment and planning excursions and rerouting traffic and locating bike lanes and so much more. Maps can form a foundation for explaining history, as the Aztecs did in their codices, colorfully showing the migrations of their ancestors over space and time.
FIGURE 8.5. Snow’s 1854 map of central London, with cholera cases represented by dots.
Maps can explain wars, as the newspapers did during World War II, showing the size and movements and alliances of troops in Europe day by day. They can track spread of disease, the first step in finding causality, as the dogged physician John Snow famously did in the cholera epidemic in London in 1854. No one knew then what caused cholera. Snow asked that each case be recorded on a map of central London, as shown in Figure 8.5. He observed that many cholera cases clustered around the Broad Street pump and ordered the pump handle removed.
That virtually ended the epidemic and, at the same time, initiated the science of epidemiology, still strongly based on maps. Maps can foster sleuthing and inferencing and discovery and prediction, whether it’s spread of disease or tracking terrorists or paths of hurricanes. They can allow making sense of voting patterns, famine and flood, and demographic data like population shifts and economic disparities. They can allow explanations of social, religious, political, linguistic, genetic, and technological change and their consequences.
Try thinking of activities central to our lives that do not involve movement in space. Not easy. Maps are a natural for showing space, and because the eye quickly sees locations and clusters and direction—remember gestalt grouping and common fate—maps foster inferences about phenomena in space and movement in space. Space and movement in space are the ground on which individual, social, political, biological, chemical, and physical processes take place. We can also create conceptual maps, and maps for theories, that show relations among ideas and changes in ideas and relations.
RULES OF THUMB FOR DESIGN OF MAPS AND OTHER THINGS
Here are four Rules of Thumb for map design that we’ve gleaned from maps, ancient and modern. Most have been tested in one way or another. Together, they conform to the two cognitive design principles: they help to ensure that the representation corresponds to the targeted concepts and that the representation is straightforward to use for the targeted tasks.
Map elements and relations in real space to elements and relations in representational space. That’s the page, virtual or actual.
Include only the information that’s useful for the task, uncluttered by irrelevant information that can distract or confuse.
Exaggerate, even distort, the useful information to make it easy to find and follow.
Add words and symbols where useful to clarify the critical information.
These rules of thumb apply to maps and to many other diagrams as well. These are best practices, gathered from observation, analysis, and experience. But there are more direct methods to determine cognitive design guidelines for specific cases, methods grounded in empirical research.
Cognitive design guidelines for route maps: Show paths and turns
Route maps are a special case of maps, but a common one. Route maps take you from A to B, from one location to another. Now a smaller jump in time, to the late nineties, for a bit of geographic scientific serendipity. Long before smartphones, a pair of grad students in computer graphics just down the street from my office came up with a prescient idea: develop an algorithm to generate route maps that would allow people to easily get from A to B. At the time, the custom routes that could be downloaded from websites were superimposed on highway maps and were almost useless. They were on a single scale, so the tricky parts, getting to and from the freeways, were too small to be seen and the route itself was buried in irrelevant clutter. The students had found our work on effective sketch maps. Together with the work of others, we had shown that people produced, preferred, and performed better with schematic route maps, ones that showed the relevant paths and places to turn, even if directions and distances weren’t accurately depicted.
What characterized those maps came to be used as cognitive des
ign guidelines specifically for maps. Cognitive design guidelines can be developed empirically for any thinking tool. For maps: clearly show the paths and the points where actions are taken, typically landmarks. Exact distance and direction are less important. Agrawala and Stolte applied those principles to create a terrific algorithm, one that produced an enormous number of A’s to B’s quickly and that was loved by users in beta testing. Applying those principles did double duty: it earned one of the students a PhD and it was sold and used by millions. Perhaps even more, it was the beginning of rapid-paced developments in map technology. And it served as a paradigm case for using empirical methods to reveal cognitive design guidelines and applying them to other design cases.
The Three Ps for finding cognitive design guidelines: Production, preference, performance
We developed an empirical program that co-opted experienced users as designers to uncover specific design principles. Here’s the program, applied to route maps. Production: one group of experts creates maps, in this case, route maps. Nowadays just about every literate adult is an expert in using maps. Preference: another group rates them for quality. Performance: a third group uses the highly rated maps for navigation. If the features gleaned from the productions are also preferred and go on to help performance, then, bingo! we have the cognitive design principles for route maps and the program for designing other graphics and more. The Three Ps Program to reveal cognitive design principles has since been applied more generally, with a string of successes. Another paradigm case was developing cognitive design principles for instructions for putting something together, and we’ll get to that soon.
Mind in Motion Page 21